article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
spin peierls compounds such as cugeo have solitons in their incommensurate phase @xcite the magnetic profile of which can be measured by nmr techniques @xcite and which must be accounted for by any theoretical model . recently a systematic study of solitons in the heisenberg spin peierls model was done by quantum monte carlo methods@xcite , with results that differ essentially from those obtained in the xy model@xcite . however
, these authors did not determine the distribution of spins and the corresponding nmr signal in the incommensurate phase , which are the subject of the present paper .
here we calculate the magnetic structure of solitons in the heisenberg spin peierls model via the dmrg method of white @xcite and also within the hartree fock approximation @xcite .
our ( essentially exact ) dmrg results in the dimerised phase agree with those of ref@xcite and they are also surprisingly close to the results of a hartree fock calculation . the magnetic profiles in the incommensurate phase that we compute by hartree fock disagree with those reconstructed from nmr data @xcite .
however , if we eliminate the oscillations in these profiles by averaging over even and odd sites , we find spin distributions that agree qualitatively with those obtained by nmr in ref[@xcite ] . the spin gap peierls in cugeo and similar materials @xcite may be due to ( i ) magnetoelastic couplings @xcite and/or ( ii ) interactions among the chains @xcite .
we consider the simplest realisation of case ( i ) : @xmath0 @xmath1 are spin @xmath2 operators , all energies are measured in units of @xmath3 and @xmath4 is the only parameter of this model .
the @xmath5 are due to elastic deformations @xmath6 the fluctuations of which are thought to be stabilised by inter chain interactions and which are treated classically .
this approximation ignores the competing energy scales of phonons @xcite and magnons for cugeo ( an attempt to go beyond the static approximation was undertaken in , for example , @xcite ) .
we also ignored second nearest neighbour exchange couplings that might be needed in a realistic model of cugeo @xcite .
the stationary point of @xmath7 with respect to the deformations @xmath8 is given by @xmath9 and this inhomogenous classical background @xmath5 is the key difficulty in tackling the spin peierls hamiltonian of eq([spinpeierlsmodel ] ) .
we use the dmrg method of white in the original spin variables , while for hartree fock we use the jordan - wigner transform @xcite of eq([spinpeierlsmodel ] ) : @xmath10 as usual , the hartree fock ( hf ) approximation @xcite is given by an energy functional @xmath11 + \mu b\sum s_{n}^{z}\mbox { } \nonumber
\\ \mbox { } s_{n}^{z}\mbox { } & = & \rho _ { n , n}-\frac{1}{2 } \nonumber\end{aligned}\ ] ] plus a consistency condition @xmath12
to test our dmrg @xcite and hf approaches , we consider the ground state energy of eq([spinpeierlsmodel ] ) for prescribed alternating dimerisation @xmath13 at @xmath14 . in the hf calculation
there is an oscillating effective coupling @xmath15 that reflects oscillations in @xmath16 and that includes oscillations of the hartree fock variables .
we find ( details are given in an appendix ) @xmath17 where @xmath18 , @xmath19 are standard elliptic functions with modulus @xmath20 .
figure(1 ) displays , descending in energy , ( i ) the ground state energy of the renormalised xy model @xmath21 \(ii ) the hartree - fock energy and ( iii ) the dmrg result ( dmrg reproduces the bethe ansatz result @xmath22 at @xmath23 ) .
the dmrg and hartree fock results are close and at @xmath24 where the system decouples into independent dimers , the hartree fock approximation is exact .
less satisfactory is the incorrect gap of the hartree - fock result at @xmath25 that was avoided by bulaevski@xcite by assuming the hartree fock parameter to be uniform along the chain .
we conclude that the hartree fock approximation is distinct from the renormalised xy model .
both in dmrg and hf we iterate eq([elastic ] ) to converge towards the fixed point .
we test this procedure on the @xmath26 solitons of the xy spin peierls model of eq([xy ] ) or , equivalently the charge=@xmath2 solitons of the peierls model of polyacetylene @xcite .
we recall that in the xy spin peierls model @xmath27 for half the points @xmath28 $ ] on a chain of odd length @xmath29 with periodic boundary conditions @xcite .
this is due to the fact that the hamiltonian couples only even with odd points and is of the form @xmath30 with a spectrum that is symmetric under @xmath31@xmath32 by the symmetry @xmath33 there are @xmath34 eigenvalues of @xmath35 plus the eigenvalue @xmath36 of the form @xmath37 which alone contributes to @xmath38 and which is responsible for the vanishing of @xmath38 on half the lattice .
all this is borne out by figure 2 : + the magnetic profile of a soliton in the full heisenberg spin peierls model as obtained by dmrg and hf are given in fig.3 . clearly , this profile is quite different from that obtained in the xy spin peierls model .
+ the dmrg result gives the symmetric profile in fig.3 and it is hard to differentiate between the two profiles .
since dmrg and hf give comparable results for the spin density in the dimerised phase and since dmrg is difficult for long inhomogenous chains , we shall use hf to estimate the distribution of spins in the incommensurate phase@xcite .
+ we crudely fix the parameter @xmath39 by requiring that the period of the soliton structure be comparable to that observed in the nmr experiments .
consider the configuration of @xmath13 and @xmath40 , multiplied by an oscillating factor , in the incommensurate phase for @xmath41 , @xmath42 , @xmath43 in fig.4 .
+ the nmr data reflects the distribution of spins along the chain or the number of points that carry a prescribed value of spin .
the configuration of spins as displayed in fig.4 ( modulo an oscillating factor ) , gives rise to the distribution of fig.5 .
we see immediately that this distribution is incompatible with the nmr data of @xcite : it contains positive and negative spins and the value of @xmath44 disagrees by a factor of @xmath45 with the value quoted by @xcite .
so we are forced to conclude that the static heisenberg spin peierls model for cugeo is incompatible with the nmr data .
the most plausible resolution of this + discrepancy is that the motion of a nuclear spin in the nmr experiment averages over the dynamics of the spin plus lattice system . to mimic such an average that is , strictly speaking , outside of the domain of validity of our calculation
, we take averages @xmath46 between even and odd spins .
the distribution of these average spins that is given in fig.6 is qualitatively similar to the distribution of spins in the xy model , in that the averages have one sign only .
the value of @xmath44 for the distribution of averaged spins is of the same order as in the nmr data of @xcite , but the agreement with the data is only on a crude and qualitative level .
we have calculated the magnetic structure of solitons in a static heisenberg spin peierls model via dmrg and the hartree fock approximation .
thanks to the fact that the renormalised xy model is not the hf approximation of the heisenbergmmodel we found reasonable agreement between hf and dmrg results in the dimerised phase . within our static model
, we obtained a spin distribution in the incommensurate phase that is incompatible with the nmr data on cugeo .
we then assumed that the nmr data reflect averages in time over fluctuating spins and lattice deformations and mimicked such internal dynamics ( that is beyond the scope of our static approach ) by taking averages over even and odd spins .
the resulting spin distribution agrees qualitatively with the nmr data .
we conclude that the dynamics of the spin plus lattice system must be taken into account in a proper theoretical description of the spin peierls transition of the cugeo system .
* acknowledgements * : we are indebted to alexander buzdin for introducing us to the spin peierls problem and for showing us the nmr results .
we thank mladen horvatic and claude berthier for comments on their data and sergei brasovskii , lev bulaevski and kazumi maki for comments on the theory .
we also thank goetz uhrig for calling our attention to the qmc calculatios of ref@xcite that we had been unaware of .
to correct an error in the litterature , we give the details of the hartree fock calculation in the dimerised phase i.e. for prescribed and uniform @xmath13 with @xmath47 , see eq([hartreefock ] ) : @xmath48 we are interested in alternating couplings : @xmath49 fourier transforming the effective hamiltonian and diagonalising it @xmath50
\nonumber \\ \lambda & = & \sqrt{k^{2}\cos ^{2}p+\delta k^{2}\sin ^{2}p } \nonumber \\ \psi ( p ) & = & u(p)\alpha ( p)+v^{*}(p)\beta ( p ) \nonumber \\ \psi ( p+\pi ) & = & v(p)\alpha ( p)-u^{*}(p)\beta ( p ) \nonumber\end{aligned}\ ] ] we find two consistency conditions : @xmath51 = -\frac{% 1}{\pi } \left ( { \bf k}-{\bf d}\right ) \nonumber \\
\delta t & = & < \sum_{|p|<\pi /2}\left [ u^{*}(p)v(p)-v^{*}(p)u(p)\right ] e^{ip}=-\frac{\delta k}{\pi k}{\bf d } \nonumber \\ \kappa & = & \sqrt{1-\left ( \frac{\delta k}{k}\right ) ^{2 } } \nonumber\end{aligned}\ ] ] at @xmath52 both conditions can be combined into one fixed point equation for @xmath53 : | we compute the magnetic profile of spin peierls solitons in a simple heisenberg model with magneto elastic couplings , using independently the dmrg method of white and the hartree fock approximation .
the results obtained in such a static model are incompatible with existing nmr data on cugeo , but the distribution of averaged spins agrees qualitatively with the data .
we conclude that the dynamics in the spin plus lattice system must be included in a more detailed theory of the spin peierls transition in cugeo .
pacs numbers : 75.10.jm , 75.50.ee , 76.60.-k |
systems of cold atoms in optical lattices have provided a highly controllable testing ground for quantum many - body physics @xcite .
particularly , a transition from superfluid ( sf ) to mott - insulator ( mi ) has attracted much attention . the sf to mi transition
can be induced by increasing the depth of the optical lattice potential , and has been experimentally realized in 1d @xcite , 2d @xcite , and 3d @xcite .
it has been established that a system of cold bosonic atoms in an optical lattice can be quantitatively described by the bose - hubbard ( bh ) model @xcite , @xmath0 when the lattice is sufficiently deep , i.e. in the tight binding regime . here
@xmath1 annihilates a boson at the lowest level localized on the @xmath2th site and @xmath3 is the number operator .
the hopping energy @xmath4 corresponds to the overlap integral of two nearest - neighboring wannier orbitals , and decreases exponentially when the lattice depth increases .
the onsite interaction energy @xmath5 increases algebraically with the lattice depth .
there is another important parameter , the number of particles per site @xmath6 ( also called as filling factor ) that is implicit in eq . ( [ eq : bhh ] ) .
the ratio @xmath7 controls the ground state phase of the bh model .
when @xmath8 , the sf phase is favored at any filling factors .
when @xmath7 increases at an integer filling factor such that @xmath9 , quantum fluctuations drive the system to the mi phase .
since the sf to mi transition is one of the most remarkable phenomena emerging in the bh model , many previous studies have made efforts to determine the transition point both numerically @xcite and experimentally @xcite . especially , theorists have accurately calculated the ratio @xmath10 at the transition point for different filling factors and dimensionalities . in 2d and 3d ,
quantum monte carlo simulations have provided @xmath10 at @xmath11 @xcite while the transition points at arbitrary filling factors have been calculated with use of the strong - coupling expansion ( sce ) techniques @xcite . in 1d , @xmath10 has been determined at low filling factors , namely at @xmath11 @xcite and @xmath12 @xcite , using the quasi - exact numerical methods of density - matrix renormalization group ( dmrg ) @xcite and time - evolving block decimation ( tebd ) @xcite . however , the transition points in the high filling region ( @xmath13 ) are yet to be obtained because of higher computational cost .
difficulty in treating the high filling region stems also from the fact that sce fails to give an accurate transition point in 1d because the transition is of the berezinski - kosterlitz - thouless ( bkt ) type . in this paper
, we calculate the critical points of the sf to mi transition in 1d for arbitrary integer filling factors , including the region of very high filling factors in which the bh model is equivalent to the @xmath14 quantum rotor model @xcite .
we emphasize that the determination of the transition points in the quantum rotor regime is important in the sense that the quantum rotor model effectively describes a regular array of josephson junctions @xcite and liquid @xmath15 absorbed in nanopores @xcite . in order to locate the transition point ,
one usually calculates static quantities such as the single - particle density matrix @xmath16 and the density - density correlation function @xmath17 @xcite .
in contrast , here we use a dynamic quantity , that is , energy splitting @xmath18 caused by tunneling between two states with macroscopically distinct currents ( or winding numbers ) @xcite . in our previous work @xcite , we have shown that the region of high filling factors can be efficiently treated with tebd by imposing a lower bound for the occupation number at each site in addition to an upper bound .
it has been also shown that the tunneling splitting can be accurately computed by means of tebd . using these prescriptions , we calculate @xmath18 to determine the transition points .
we find that the transition point as a function of @xmath6 is well approximated by @xmath19 where @xmath20 denotes the dimensionality of the system ( e.g. @xmath21 for 1d ) , and the constants @xmath22 , @xmath23 , and @xmath24 are numerically determined .
we show that one can express @xmath25 also in 2d and 3d , which has been obtained in ref .
, in the form of eq .
( [ eq : critu ] ) with different values of the constants . given these results
, one can immediately know the transition points for any filling factors and any realistic dimensions .
this paper is organized as follows . in sec .
[ sec : model ] , we introduce the system and the model considered here and explain how to determine the transition point from the tunneling splitting .
we qualitatively discuss the relation between the tunneling splitting and the sf to mi transition from a perspective of the instanton method . in sec .
[ sec : xxz ] , in order to examine the performance of the suggested procedure , we apply it to the 1d hardcore bh model with nearest - neighbor interactions that is exactly solvable by means of the bethe ansatz . in sec .
[ sec : ucvsnu ] , we calculate the transition point in the bh model of eq .
( [ eq : bhh ] ) as a function of the filling factor . in sec .
[ sec : con ] , we summarize the results .
we consider a system of 1d lattice bosons in a ring - shaped geometry , i.e. with a periodic boundary .
we assume for the moment that the system is in the sf phase . in the classical limit , which corresponds to the limit of @xmath26 in the bh model , a state with a finite homogeneous current
can be metastable , and its quasimomentum per particle is discretized as @xmath27 , where the integer @xmath28 is the winding number , @xmath29 the number of lattice sites , and @xmath30 the lattice spacing .
we suppose a situation that two states with different winding numbers , say @xmath31 and @xmath32 , are degenerate .
we define the winding number difference between the two states @xmath33 as @xmath34 .
when @xmath35 is an integer and @xmath7 is finite , umklapp scattering processes can induce phase slips via quantum tunneling to couple these two states . as a result ,
the degeneracy is broken and there emerges the energy splitting @xmath18 that quantifies the tunneling rate . applying the instanton techniques @xcite to the phenomenological tomonaga - luttinger ( tl ) liquid model , kashurnikov _ et al .
_ have derived the following scaling formula of the energy splitting at the smallest possible winding number difference @xmath33 with respect to @xmath29 @xcite , @xmath36 where the tl parameter @xmath37 is defined such that the effective action for the phase of the bosonic field @xmath38 is described as @xmath39.\end{aligned}\ ] ] here @xmath40 is the sound speed .
it is well - known that the sf phase is favored when @xmath41 and that the sf to mi transition of the bkt type occurs at @xmath42 @xcite . at the transition point
, the renormalization of the tl parameter leads to a logarithmic correction on the scaling of @xmath18 as @xcite @xmath43 in the next section , we will see that the inclusion of the logarithmic correction is important to accurately calculate the transition point .
given the scaling formulas of eqs .
( [ eq : esp ] ) and ( [ eq : esplog ] ) , one can determine the transition point from the energy splitting as follows .
first , using tebd , we numerically compute @xmath18 as a function of @xmath29 in the way described in ref . .
we next fit a function @xmath44 to the numerical data , where @xmath45 and @xmath46 are free parameters , and extract the exponent @xmath46 .
we determine the transition point from the condition that @xmath47 .
let us explain a reason why the exponent @xmath46 is equal to unity at the transition point from a different viewpoint , which is the relation between the sf to mi transition and the validity of the instanton formula of eq .
( [ eq : esp ] ) .
an important point is that the instanton techniques used to derive eq .
( [ eq : esp ] ) are based on the so - called dilute gas approximation ( dga ) , in which the instantons are assumed to be well separated from each other in the path - integral trajectories that contribute to the partition function @xcite . in our system of 1d lattice bosons , since an instanton is regarded as a vortex in the space - time plane , the breakdown of dga means that many vortices are created in the space - time coordinate so that they destroy the long - range coherence of bosonic phases , leading to the quantum phase transition to the mott insulator . in short
, the breakdown of dga signals the mott transition @xcite . in general
, the condition under which the dilute gas approximation is valid is that the size of an instanton along the imaginary - time axis @xmath48 is much smaller than the tunneling time @xmath49 . in the present case ,
since the winding number difference @xmath50 is of the order of unity , the instanton size is of the order of the system size , i.e. @xmath51 .
meanwhile , @xmath18 is given by eq .
( [ eq : esp ] ) . obviously , when @xmath52 , the condition that @xmath53 is held in the thermodynamics limit such that dga is valid , i.e. the system is in the sf phase .
thus , the condition for the mott transition is given by @xmath47 .
in this section , we analyze the 1d hardcore bh model with the nearest - neighbor interactions , @xmath54 in order to illustrate that the critical point of the 1d superfluid to insulator transition can be accurately calculated along the procedure described in the previous section . here , @xmath55 is the nearest - neighbor interaction , and @xmath56 and @xmath57 are the creation and number operators of a hardcore boson at site @xmath2 .
we also include the phase twist @xmath58 in the hopping term in order to control the winding number of states .
the hardcore constraint means that the maximum occupation number at each site is unity .
this constraint leads to the identities between the operators of hardcore bosons and those of @xmath59-spins , namely @xmath60 and @xmath61 , which tell us that the model of eq .
( [ eq : hcbhh ] ) is equivalent to the spin-@xmath59 xxz model and is exactly solvable by means of the bethe ansatz @xcite .
according to the exact solution , there is a quantum phase transition between the sf and the density - wave insulator at half filling due to the competition between @xmath55 and @xmath4 .
when @xmath62 , the particles favor to be delocalized and the system is in the sf phase . when @xmath63 increases , the umklapp scattering tends to localize the particles , and the transition to the insulating state occurs at @xmath64 . in the following , we show that our procedure provides a numerical value of the transition point that is very close to the exact one .
( color online ) the time evolution of the current velocity @xmath65 in the dynamics of the model of eq .
( [ eq : hcbhh ] ) , where @xmath66 and @xmath67 . ]
( color online ) the time evolution of the overlaps @xmath68 ( blue solid line ) and @xmath69 ( black dashed line ) in the dynamics of the model of eq .
( [ eq : hcbhh ] ) , where @xmath66 and @xmath67 . ] since the sf to mi transition occurs at @xmath70 in the model of eq .
( [ eq : hcbhh ] ) , the minimum winding number difference in possible phase slip processes is @xmath71 . in order to obtain the energy splitting for @xmath71
, we first prepare the ground state of eq .
( [ eq : hcbhh ] ) with @xmath72 , which is a state with winding number @xmath73 . for dealing with our system with a ring - shape geometry ,
we use tebd for a periodic boundary condition @xcite . taking the state with @xmath73
as the initial state and setting @xmath74 , we next carry out the real - time evolution . since two states with @xmath73 and @xmath75 are degenerate when @xmath76 , it is expected that the supercurrent dynamics exhibit a coherent oscillation between these two states induced by quantum tunneling . to demonstrate this
, we calculate the time evolution of the current velocity given by @xmath77 where @xmath78 is the total number of particles . in fig .
[ fig : vel ] , @xmath65 for @xmath66 and @xmath79 is shown .
we clearly see that the velocity oscillates between @xmath80 and @xmath81 , i.e. between the states with @xmath73 and @xmath75 .
we also calculate the overlap @xmath82 of the wave function with the ground state @xmath83 of the hamiltonian ( [ eq : hcbhh ] ) with @xmath84 . in fig .
[ fig : ovlp ] , we show overlaps @xmath68 and @xmath69 , which reconfirm that the wave function @xmath85 coherently oscillates between @xmath86 and @xmath87 . we note that similar quantum - tunneling dynamics have been found also for quantum vortices in anisotropic traps @xcite and supercurrents in two - color optical lattices @xcite .
once the tunneling dynamics in real time are obtained , we can extract the energy splitting @xmath18 by fitting @xmath68 using the function @xmath88 where @xmath18 , @xmath89 , and @xmath90 are free parameters .
( color online ) the energy splitting versus the number of lattice sites @xmath29 for the 1d hardcore bose - hubbard model of eq .
( [ eq : hcbhh ] ) at half filling .
we take @xmath79 ( black diamonds ) , @xmath91 ( blue squares ) , and @xmath92 ( red diamonds ) .
the solid lines represent the best fits to the respective numerical date with the function of eq .
( [ eq : fitf ] ) , where @xmath93 , @xmath94 , and @xmath95 for @xmath79 , @xmath91 , and @xmath92 . ]
( color online ) the exponent @xmath46 versus @xmath63 for the 1d hardcore bose - hubbard model of eq .
( [ eq : hcbhh ] ) at half filling .
@xmath46 is extracted by fitting the numerical data of @xmath18 versus @xmath29 upto @xmath96 .
we use the fitting function of eq .
( [ eq : fitf ] ) with the logarithmic correction for red squares , while we use eq .
( [ eq : fitfd ] ) without the correction for the black circles .
the solid lines are guides to the eyes .
the dashed line represents @xmath47 , which is the condition to determine the transition point . ] in fig .
[ fig : delvsl ] , we plot the energy splitting between @xmath86 and @xmath87 as a function of the number of lattice sites @xmath29 up to @xmath96 .
the numerical data are very well fitted by the function of eq .
( [ eq : fitf ] ) . from this fitting , we extract the exponent @xmath46 and calculate it varying @xmath63 , as shown by the red squares in fig . [
fig : beki ] .
as expected , @xmath46 monotonically decreases with @xmath63 .
the condition that @xmath47 gives the transition point as @xmath97 .
this value is so close to the exact result , @xmath98 , that the relative error , @xmath99 , is smaller than @xmath100 .
we emphasize that our procedure provides an accurate value of the transition point even when the system size is relatively small .
for instance , when we take the number of lattice sites up to @xmath101 for the fitting of the energy splitting , we find that @xmath102 and that the relative error is still within @xmath103 .
this is a clear advantage of the use of the energy splitting over the correlation functions that require a substantially larger system size .
we also note a disadvantage of our procedure that one needs a system with periodic boundaries in which the bipartite entanglement entropy is twice as large as that in a system with open boundaries . in order to corroborate the necessity of the logarithmic correction in the fitting function of eq .
( [ eq : fitf ] ) , we instead use another function that does not include the correction , @xmath104 for the fitting .
the black circles in fig .
[ fig : beki ] represent @xmath46 extracted with @xmath105 , and the value of the transition point in this case is @xmath106 .
the relative error is @xmath107 , which is much larger than the case with the logarithmic correction .
.[tab:1duc ] the critical values @xmath108 of the 1d bh model for several values of the filling factor @xmath6 . [ cols="^,^",options="header " , ] the function of eq .
( [ eq : critu ] ) is a good approximation also for the transition points in 2d and 3d . to show this , we depict in fig .
[ fig:2d3d ] the numerical data of @xmath109 versus @xmath6 for 2d ( blue squares ) and 3d ( black diamonds ) , which are obtained using sce in ref . , together with the best fit with the function of eq .
( [ eq : critu ] ) represented by the solid lines .
the transition points calculated from eq .
( [ eq : teich ] ) are also plotted with the dashed lines .
there we see that the function of eq .
( [ eq : critu ] ) approximates the transition points as accurately as eq .
( [ eq : teich ] ) .
the values of the numerical constants @xmath110 in the fitting function are summarized in table [ tab : arbd ] .
in summary , we have studied the superfluid to mott insulator transition in a system of one - dimensional lattice bosons .
we have shown that the transition points in 1d can be accurately determined from the energy splitting between two degenerate states with distinct winding numbers .
we obtained the transition points for the bose - hubbard model with arbitrary integer filling factors , including the high filling limit corresponding to the quantum rotor regime .
we have found a simple analytical formula of eq .
( [ eq : critu ] ) that well approximates the transition point versus the filling factor for 1d , 2d , and 3d . with this formula ,
the transition points can be obtained easily and immediately for any fillings and any realistic dimensionalities .
the authors thank e. altman , n. prokofev , b. svistunov , and s. tsuchiya for stimulating discussions .
i. d. thanks boston university visitors program for hospitality .
a. p. was supported by nsf dmr-0907039 , afosr fa9550 - 10 - 1 - 0110 , and the sloan foundation .
the computation in this work was partially done using the riken cluster of clusters facility .
the same argument is valid also for the superflow decay via quantum tunneling from a state with a certain winding number . for more details , see i. danshita and a. polkovnikov , _ quantum phase slips in one - dimensional superfluids in a periodic potential _ , arxiv:1110.xxxx ( 2011 ) . | we study the quantum phase transition between the superfluid and the mott insulator in the one - dimensional ( 1d ) bose - hubbard model . using the time - evolving block decimation method
, we numerically calculate the tunneling splitting of two macroscopically distinct states with different winding numbers . from the scaling of the tunneling splitting with respect to the system size
, we determine the critical point of the superfluid to mott insulator transition for arbitrary integer filling factors .
we find that the critical values versus the filling factor in 1d , 2d , and 3d are well approximated by a simple analytical function .
we also discuss the condition for determining the transition point from a perspective of the instanton method . |
studies of @xmath2 and @xmath4 mesons and their interactions with nucleons provide an interesting window on axial u(1 ) dynamics .
the flavor - singlet @xmath5 channel is characterized by a large okubo - zweig - iizuka ( ozi ) violation : the masses of the @xmath2 and @xmath6 mesons are 300400 mev larger than the values they would have if these mesons were pure would - be goldstone bosons associated with spontaneously broken chiral symmetry @xcite .
one needs extra mass in the singlet channel associated with nonperturbative topological gluon configurations and the qcd axial anomaly @xcite .
the strange quark mass induces considerable @xmath2-@xmath6 mixing . for the physical @xmath2-@xmath6 mixing angle @xmath7 degrees
, the singlet component in the @xmath2 has the potential to induce a factor of 2 increase in the @xmath2-nucleon scattering length @xmath8 relative to the value one would expect if the @xmath2 were a pure octet state @xcite .
the flavor - singlet goldberger - treiman relation @xcite relates @xmath4-nucleon coupling to the small flavor - singlet axial charge measured in the proton spin puzzle @xcite .
the properties of @xmath2 and @xmath4 mesons should manifest themselves in production and decay processes and in their interaction with nuclear matter .
these processes are being studied in experiments from threshold @xcite through high - energy collisions where anomalously high branching ratios have been observed for @xmath9 and @xmath10-meson decays to an @xmath6 plus additional hadrons @xcite . at the hadronic level
, it is expected that the @xmath4 meson can be produced through the exchange and rescattering of mesons , through the excitation of an intermediate baryonic resonance or via the fusion of virtual mesons @xcite .
so far the @xmath11 process has been studied quantitatively only within ( i ) an effective lagrangian approach , assuming that the s@xmath12(1535 ) resonance is dominant @xcite and ( ii ) within a covariant effective meson - nucleon theory including meson and nucleon currents with the nucleon resonances s@xmath12(1650 ) , p@xmath12(1710 ) and p@xmath13(1720 ) @xcite . however , for @xmath6 production there is an additional potential contribution where glue is excited in the `` short distance '' ( @xmath14-fm ) interaction region of the proton - nucleon collision and then evolves to become an @xmath6 in the final state . this gluonic contribution to the cross section for @xmath15 , proposed by bass @xcite ,
is beyond the contributions associated with meson exchange models .
there is no reason , a priori , to expect this ozi violating process to be small .
because glue is flavor - blind it contributes with the same strength in both the @xmath15 and the @xmath16 reactions .
it is not possible to establish the relative contributions of possible reaction mechanisms responsible for @xmath4-meson production based only on the presently available cross sections for the @xmath17 reaction @xcite .
therefore , one has to investigate @xmath4 production as a function of spin and isospin degrees of freedom .
such studies were conducted in the case of the @xmath2 meson @xcite .
the ratio @xmath18 has been measured for quasifree @xmath2 production from a deuteron target up to 109 mev above threshold @xcite .
one finds that @xmath19 is approximately energy independent with a value of @xmath20 in the energy range of @xmath21 mev signifying a strong isovector exchange contribution to the @xmath2 production mechanism .
the ratio @xmath22 has not been measured to date , and the existing predictions differ drastically depending on the model .
cao and lee @xcite assumed , by analogy to the production of the @xmath2 meson , that the production of the @xmath4 meson proceeds dominantly via the s@xmath12(1535 ) resonance .
as a consequence they predicted within an effective lagrangian approach an @xmath23 value equal to the experimentally established @xmath19 value . in contrast , kaptari and kmpfer @xcite predicted a value of @xmath23 close to @xmath241.5 in the kinematic range of the cosy-11 experiment with the dominant contribution coming from the meson conversion currents . in the extreme scenario of glue - induced production saturating the @xmath4 production cross section , the ratio @xmath25 would approach unity after correction for the final - state interaction between the two outgoing nucleons .
the su(3 ) wave functions of the @xmath2 and @xmath4 mesons differ due to iso - singlet gluonic and strangeness degrees of freedom .
hence , it is reasonable that singlet currents will play a greater role in @xmath4 than in @xmath2 production .
proton - proton data are available from cosy and saturne @xcite .
to investigate the production mechanism it is important to provide an empirical base with spin and isospin dependence of the cross sections . in this article
we present first results on the measurement of the @xmath11 reaction .
the cosy-11 detector system @xcite is schematically depicted in fig.[cosy11 ] .
a quasifree proton - neutron reaction was induced by a proton beam @xcite with a momentum of 3.35 gev / c impinging on a deuteron cluster target @xcite .
a detailed description of the functioning of the detectors and of the experimental technique is given in a recent cosy-11 article devoted to measurement of the @xmath26 reaction @xcite and in ref .
therefore , we describe it only briefly here .
the experiment is based on the registration of all outgoing nucleons from the @xmath27 reaction .
the proton moving forward is measured in two drift chambers and scintillator detectors and the neutron is registered in the neutral particle detector .
the proton moving backward is measured with a dedicated silicon - pad detector @xcite . in the data analysis
the backward proton from the deuteron is considered as a spectator that does not interact with the bombarding proton , but escapes untouched and hits the detector carrying the fermi momentum possessed at the time of the reaction .
the total energy available for the quasifree proton - neutron reaction is calculated for each event based on the momentum vectors of the spectator and beam protons under the assumption that the spectator proton is on its mass shell at the moment of the collision .
the absolute momentum of the neutron is determined from the time of flight between the target and the neutron detector and its direction is derived from the hit position defined as the center of the detection module whose signal was produced as the first one . reaction .
protons are registered in two drift chambers d1 and d2 and in the scintillator hodoscopes s1 and s3 .
the s1 detector is built of 16 vertically arranged modules , whereas the s3 hodoscope is a nonsegmented scintillator wall viewed by a matrix of 217 photomultipliers .
an array of silicon pad detectors ( spectator detector ) is used for registration of spectator protons .
neutrons are registered in the neutral particle detector consisting of 24 independent detection units . to distinguish neutrons from charged particles
a veto detector is used .
elastically scattered protons are measured in the scintillator detector s4 and the position - sensitive silicon detector @xmath28 .
the sizes of the detectors and their relative distances are not to scale .
, scaledwidth=35.0% ] the veto detector installed in front of the neutral particle detector discriminates signals originating from charged particles , whereas the @xmath29 quanta are discernd by means of the time of flight measured between the target and the neutral particle detector where the time of the reaction in the target is reconstructed based on the trajectory and velocity of the forward scattered protons detected in drift chambers and s1 and s3 scintillators .
figure [ banan_all_pads ] presents the momentum distribution of protons considered as spectators ( crosses ) compared with simulations taking into account a fermi motion of nucleons inside the deuteron ( solid line ) . in the analysis of the @xmath11 reaction spectator protons with momenta ranging from about 35 to about 115 mev / c were taken into account .
the lower limit is caused by the noise level of the silicon pad detectors .
the _ spectator noise cut _ was determined for each pad of the detector separately , using the spectra of energy loss triggered by a pulser with a frequency of 1 hz in addition to other experimental triggers .
protons with a momentum larger than about 115 mev / c were not taken into account because we used the second layer of the double - layer spectator detector as a veto to reduce the background from charged pions .
the shapes of the simulated and experimental spectra shown in fig .
[ banan_all_pads ] agree quite well , and the small difference will be taken for the estimation of the systematical uncertainties of the final results caused by the assumption of a fermi momentum distribution of nucleons inside the deuteron . the accuracy of the excess energy determination was estimated based on monte carlo simulations . for this purpose we simulated @xmath30 events of the @xmath11 reaction taking into account the fermi motion of the nucleons inside the deuteron , the size of the target @xcite , the spread of the beam momentum and the horizontal and vertical beam size @xcite .
simulated data were analyzed in the same way as the experimental data , taking into consideration the energy and time resolution of the detectors , resulting in an excess energy resolution of @xmath31 mev .
it is worth stressing that using this method for the @xmath32 reaction the standard deviation of the excess energy was derived to be @xmath33 mev @xcite , which is comparable to the @xmath34 mev obtained under similar conditions at the promice / wasa setup @xcite . for further analysis
the collected data were grouped according to the excess energy .
the available range of excess energy @xmath35 above the @xmath4-meson production threshold has been divided into three 8 mev - wide intervals .
the choice of the width of the interval is a compromise between statistics and resolution and reflects the accuracy ( fwhm ) of the determination of the excess energy . in figs [ mm_da_q0_8](a ) ,
[ mm_da_q0_8](c ) , [ mm_da_q0_8](e ) the experimental mass distributions determined for the excess energy ranges @xmath36 $ ] , @xmath37 $ ] and @xmath38 $ ] mev are shown by solid lines .
for all distributions we observe an increase in counting rate toward the kinematical limit . the shape is mostly caused by the detector acceptance . a signal from the @xmath4-meson production
is expected on top of the continuous multipion mass distribution at the position corresponding to the @xmath4 mass ( 0.95778 gev/@xmath39 @xcite , which is denoted by the arrow in fig .
however , no enhancement around this region is seen . to extract a signal
the histograms corresponding to positive @xmath35 intervals were compared to the background established from the data below the threshold . for each distribution of the missing mass spectra at @xmath40 ,
a corresponding background spectrum was constructed from events with @xmath35 belonging to the range @xmath41 $ ] mev by applying the method described in detail in a dedicated article @xcite .
this method was already applied successfully to the analysis of the @xmath32 reaction @xcite . as a cross - check , to gain more confidence in the applied procedure we constructed alternative background distributions taking events from the @xmath35 range from -20 to 0 mev .
the result of this investigation has shown a consistency of both background shapes of the order of 1% .
next , for each distribution for @xmath40 the corresponding background spectrum was normalised for mass values smaller than 0.25 gev/@xmath42 where no events from the @xmath4 meson are expected . in this
region events correspond to one - pion production , for which the total cross section remains nearly constant in the range of the excess energy shown , because for a single - pion production the excess energy is high above the threshold @xcite .
the dashed lines in fig .
[ mm_da_q0_8 ] depict the background spectrum , shifted to the kinematical limit and normalized accordingly .
the normalization in the missing mass range significantly below the mass of the @xmath4 meson ( in the range from 0.75 to 0.85 gev/@xmath42 ) leads to the same result .
figures [ mm_da_q0_8 ] ( b ) , [ mm_da_q0_8](d ) , [ mm_da_q0_8](f ) show distributions of the missing mass for the @xmath0 reaction as determined after subtraction of the background . because of the low statistics and very low signal - to - background ratio the signal from @xmath4-meson created in the proton - neutron collision is statistically insignificant .
therefore , we can only estimate the upper limit for the @xmath4 meson production in the @xmath0 reaction .
reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] reaction ( solid lines ) .
distributions were obtained for ranges of excess energy @xmath35 from 0 to 8 mev ( a ) , from 8 to 16 mev ( c ) and from 16 to 24 mev ( e ) .
corresponding background spectra are shown as dashed lines . *
( b , d , f ) * experimental missing mass distributions of the @xmath0 reaction for excess energy ranges from 0 to 8 mev ( b ) , from 8 to 16 mev ( d ) and from 16 to 24 mev ( f ) determined after background subtraction .
vertical bars indicate statistical errors .
the arrow depicts the nominal value of the @xmath4 mass.,title="fig:",scaledwidth=27.0% ] to estimate the systematic error of the number of @xmath43 events , the change in the missing mass spectra was studied by varying different parameters describing the experimental conditions , discussed in the following analysis and simulation @xcite .
the change in the global time offset of the neutral particle detector by 1 standard deviation of its time resolution @xmath44 ns ) resulted in a systematic error of the cross sections of 3% . a change in simulations of the time resolution of the neutron detector ( 0.4 ns ) by @xmath450.2 ns resulted in @xmath455% changes in the cross section .
the variation in the cut on the noises in the spectator detector within the limits of energy resolution results in a change of cross sections by @xmath457% . in the simulation we also checked that varying the beam momentum resolution arbitrarily by @xmath451 mev / c around its nominal value of 3.35 mev / c resulted in changes in the cross section values by @xmath453% . as discussed previously , and demonstrated in ref .
@xcite , the systematic error owing to the method used for the background subtraction is equal to 1% .
the uncertainty owing to the model dependence calculation of the fermi momentum distribution is equal to 2% @xcite and was estimated as the difference in results determined using the paris @xcite versus the cd - bonn @xcite potentials .
it is worth stressing that also a difference between the experimental fermi momentum distribution and the predictions based on the paris potential leads to a variation in the final result of about 2% .
a displacement of the spectator detector by 1 mm ( the estimated accuracy of its position determination ) changes the number of events by 5% .
the error of the detection efficiency , predominantly caused by the uncertainty in the efficiency of the neutral particle detector @xcite , is determined to be no larger than 5% @xcite .
we also took into account the systematic error ( @xmath46 @xcite ) originating from the luminosity determination .
the nuclear `` shadow effect '' ( estimated to about 4.5% @xcite ) was not taken into account because the reduction of the beam flux seen by the neutron , owing to the shielding by a spectator proton , is expected to be the same for the @xmath11 reaction and quasielastic scattering that was used for determination of the luminosity .
similarly we have neglected the effect owing to the reabsorption of the produced meson by the spectator proton , because even for the case of @xmath2-meson production this effect reduces the cross section by a factor of about 3% @xcite and the proton-@xmath2 interaction is much stronger than the proton-@xmath4 interaction @xcite .
finally , the total systematic error of @xmath47 was estimated as the quadratic sum of the nine independent systematic errors discussed and is equal to @xmath48 .
the luminosity of @xmath49 was established from the number of quasifree proton - proton scattering events applying the method described in the dedicated article @xcite .
the acceptance of the detector setup and efficiency was determined based on monte carlo studies .
the signal from the @xmath4 meson in the determined missing mass spectra is statistically insignificant .
therefore only the upper limit of the total cross section for the quasifree @xmath50 reaction was extracted .
the number of @xmath4 mesons can be calculated as the difference between the number of events @xmath51 in the peak for the signal ( solid line in fig .
[ mm_da_q0_8 ] ) and the number of events @xmath52 in the background of the peak ( dashed line in fig .
[ mm_da_q0_8 ] ) : @xmath53 the range for the integration was chosen based on the simulation of the missing mass distributions @xcite . assuming that no @xmath4 mesons are observed @xmath54 in the experiment , the value of 1.2815 @xmath55 @xmath56 gives the upper limit of the total cross section for the @xmath0 reaction at a 90% confidence level .
the result is shown in fig .
[ cross_pn ] and in table [ table_pn ] .
.excess energy range , upper limit of the total cross section for the @xmath0 reaction , mean value of the @xmath17 total cross section according to the parametrization given in ref .
@xcite , upper limit of the ratio @xmath23 as a function of the excess energy .
energy intervals correspond to the binning applied . [ cols="^,^,^,^ " , ] the horizontal bars in fig .
[ cross_pn ] represent the intervals of the excess energy for which the upper limit of the total cross section was calculated .
reaction as a function of the excess energy ( open squares ) .
upper limit for the total cross section for the @xmath0 reaction as a function of the excess energy ( filled circles).,scaledwidth=50.0% ] the total cross section for the @xmath17 reaction was measured in previous experiments @xcite .
it reveals a strong excess energy dependence , especially very close to threshold .
this dependence must be taken into account when comparing the results for the @xmath0 reaction that were established for 8-mev excess energy intervals .
therefore , for a given interval of excess energy , we have determined the mean value of the total cross section for @xmath17 reaction using the parametrization of fldt and wilkin @xcite fitted to the experimental data @xcite . figure [ ratio_etap ] presents the upper limit of the ratio @xmath1 of the total cross section for the @xmath0 and @xmath17 reactions as a function of the excess energy ( arrows ) .
the corresponding values are listed in table [ table_pn ] .
the corresponding ratios for the @xmath2 meson are also shown as open squares in fig [ ratio_etap ] .
for the @xmath2 meson , the value of @xmath19 is @xmath576.5 at excess energies larger than @xmath58 mev @xcite , which suggests the dominance of isovector - meson exchange in the production mechanism .
the decrease in @xmath19 close to the threshold @xcite may be explained by the different energy dependence of the proton - proton and proton - neutron final - state interactions @xcite . of the total cross sections for the @xmath0 and @xmath17 reactions ( arrows ) in comparison with the ratio @xmath59 determined for the @xmath2 meson @xcite ( open squares ) .
the superimposed solid line indicates the result of a fit to the @xmath19 data taking into account the final - state interaction of nucleons @xcite .
the dotted line presents the result of calculations performed under assumption of dominance of the s@xmath12(1535 ) resonance in the production process @xcite , and the dashed line denotes the result obtained within a covariant effective meson - nucleon theory including meson and nucleon currents with nucleon resonances s@xmath12(1650 ) , p@xmath12(1710 ) and p@xmath13(1720 ) @xcite , scaledwidth=50.0% ] for the @xmath4 meson the upper limit of the ratio for the excess energy range @xmath36 $ ] mev is nearly equal to values of the ratio obtained for the @xmath2 meson , whereas for larger excess energy ranges , @xmath37 $ ] and @xmath38 $ ] mev , the upper limits of the ratio are lower by about 1 standard deviation each .
the dotted curve in fig . [ ratio_etap ] corresponds to the prediction for @xmath23 by cao and lee @xcite and exceeds the observed limit , showing the importance of other exchange currents in the @xmath4 production .
the prediction of kaptari and kmpfer @xcite is well within the upper bound .
a smaller @xmath23 than @xmath60 is also consistent with a possible greater role for singlet currents in @xmath4 production than @xmath2 production .
if there are important new dynamics in the @xmath4 production process relative to @xmath2 production , a key issue is the relative phase @xcite of possible additional exchanges compared to the isovector currents that dominate @xmath2 production .
the observed limit thus constrains modeling of the production processes . to confirm these interesting observations and to go farther ,
new experimental investigations with improved statistics are required .
to determine the excitation function of the total cross section for the quasifree @xmath0 reaction near the kinematical threshold , we performed an experiment at the cooler synchrotron cosy at the research centre jlich using the cosy-11 detector system .
the quasifree @xmath0 reaction was induced by a proton beam with a momentum of 3.35 gev / c on a deuteron target .
all outgoing nucleons have been registered , whereas for @xmath4-meson identification the missing mass technique was applied .
the upper limit of the total cross section for the @xmath0 reaction in the excess energy range between 0 and 24 mev has been determined .
the total cross section for the @xmath17 reaction was measured in previous experiments at the same energy range .
combining these data , we have determined the upper limit of the ratio @xmath61 .
the comparison of the @xmath19 ratio with the upper limits of @xmath23 established here suggests nonidentical production mechanisms for the @xmath4 and @xmath2 mesons , which are dominated by a strong isovector exchange contribution . to pin down the detailed reaction mechanism further theoretical as well as experimental investigations with better statistics
are required .
we are grateful for useful comments of burkhard kmpfer , piotr salabura and joanna stepaniak .
the work was partially supported by the european community - research infrastructure activity under the fp6 and fp7 programs ( hadron physics , rii3-ct-2004 - 506078 , primenet no .
227431 ) , by the polish ministry of science and higher education under grant nos 3240/h03/2006/31 , 1202/dfg/2007/03 , and 0082/b / h03/2008/34 , by the german research foundation ( dfg ) , by the ffe grants from the research center jlich , by the austrian science fund , fwf , through grant p20436 and by the virtual institute `` spin and strong qcd '' ( vh - vp-231 ) .
9 s. d. bass , phys .
* t 99 * , 96 ( 2002 ) . s. d. bass , acta phys .
* b * proc .
* 2 * , 11 ( 2009 ) .
shore , hep - ph/9812354 ; lect .
notes phys .
* 737 * , 235 ( 2008 ) [ hep - ph/0701171 ] s. d. bass and a. w. thomas , phys .
b634 * , 368 ( 2006 ) .
shore and g. veneziano , phys .
b244 * , 75 ( 1990 ) ; t. hatsuda , nucl . phys . * b329 * , 376 ( 1990 ) . s. d. bass , rev .
phys . * 77 * , 1257 ( 2005 ) .
p. moskal , m. wolke , a. khoukaz , w. oelert , prog .
49 * , 1 ( 2002 ) ; p. moskal , habilitation dissertation no . 364 ( jagiellonian university press , cracow , 2004 ) , arxiv : hep - ph/0408162 .
p. ball , j. m. frere and m. tytgat , phys .
* b 365 * , 367 ( 1996 ) . h. fritzsch , phys .
* b 415 * , 83 ( 1997 ) . k. nakayama , h. f. arellano , j. w. durso , and j.speth , phys .
rev . * c 61 * , 024001 ( 2000 ) . k. nakayama and h. haberzettl , phys . rev .
* c 69 * , 065212 ( 2004 ) . l. p. kaptari , b. kmpfer , eur . phys .
j. * a 37 * , 69 ( 2008 ) ; b. kmpfer ( private communication ) .
x. cao and x .- g .
lee , phys . rev .
* c 78 * , 035207 ( 2008 ) ; x. cao ( private communication ) .
s. d. bass , phys .
* b 463 * , 286 ( 1999 ) .
p. klaja _ et al .
* b 684 * , 11 ( 2010 ) ; a. khoukaz _ et al .
j. * a 20 * , 345 ( 2004 ) ; p. moskal _ et al .
_ , phys . lett .
* b 474 * , 416 ( 2000 ) ; f. balestra _ et al .
* b 491 * , 29 ( 2000 ) ; p. moskal _ et al .
lett . * 80 * , 3202 ( 1998 ) ; f. hibou
_ et al . , _ phys . lett . * b 438 * , 41 ( 1998 ) .
h. caln _ et al .
* c 58 * , 2667 ( 1998 ) .
p. moskal _ et al .
* c 79 * , 015208 ( 2009 ) .
r. czyykiewicz _ et al .
* 98 * 122003 ( 2007 ) .
s. brauksiepe _ et al .
instr . & meth .
* a376 * , 397 ( 1996 ) .
d. prasuhn _ et al .
& meth . * a 441 * , 167 ( 2000 )
. h. dombrowski _ et al .
& meth . * a 386 * , 228 ( 1997 )
. j. klaja , ph.d .
thesis , jagiellonian university , 2009 , + e - print : arxiv:0909.4399 .
r. bilger _
et al . , _ nucl .
instr . & meth .
* a 457 * , 64 ( 2001 ) .
m. lacombe _ et al .
* b 101 * , 139 ( 1981 ) .
p. moskal _ et al .
. instr . & meth .
* a 466 * , 488 ( 2001 ) . c. amsler
_ et al . , _ phys .
* b 667 * , 1 ( 2008 ) .
p. moskal _ et al . ,
_ j. phys . *
g 32 * , 629 ( 2006 ) .
r. czyykiewicz , diploma thesis , * jl-4017 * , jagiellonian university , 2002 .
r. machleidt _ et al .
* c 63 * , 024001 ( 2001 ) .
j. klaja et al . , acta phys . polon . * b * proc . suppl .
* 2 * , 31 ( 2009 ) .
e. chiavassa et al .
* b 322 * , 270 ( 1994 ) .
e. chiavassa et al .
* b 337 * , 192 ( 1994 ) .
p. moskal et al . , phys . lett .
* b 482 * ( 2000 ) 356 .
p. moskal and r. czyykiewicz , aip conf
. proc . * 950 * , 118 ( 2007 ) .
p. moskal _ et al .
phys . * a 22 * , 305 ( 2007 ) .
g. fldt and c. wilkin , phys . lett .
* b 382 * , 209 ( 1996 ) . g. fldt and c. wilkin , phys . rev .
* c 56 * , 2067 ( 1997 ) . c. wilkin ( private communication ) .
g. fldt and c. wilkin , phys .
scripta * 64 * , 427 ( 2001 ) . | the upper limit of the total cross section for the @xmath0 reaction has been determined near the kinematical threshold in the excess energy range from 0 to 24 mev .
the measurement was performed using the cosy-11 detector setup , a deuteron cluster target , and the proton beam of cosy with a momentum of 3.35 gev / c .
the energy dependence of the upper limit of the cross section was extracted exploiting the fermi momenta of nucleons inside the deuteron .
comparison of the determined upper limit of the ratio @xmath1 with the corresponding ratio for @xmath2-meson production does not favor the dominance of the @xmath3 resonance in the production process of the @xmath4 meson and suggests nonidentical production mechanisms for @xmath2 and @xmath4 mesons . |
since several decades polymer brushes find abiding interest for various applications , and have been studied very intensively by experiment , analytical theory , and computer simulations @xcite .
the structure of these soft polymeric layers and their response to various external perturbations depends in a delicate way on various control parameters such as molecular weight , grafting density , quality of the solvent , and character of the interactions between monomeric units and the substrate to which the macromolecules are grafted .
however , inspired by the early works @xcite , mostly brushes formed from completely flexible chains were considered , where the elastic response is entirely entropic , due to the configurational degrees of freedom of the tethered chain molecules . in such brushes
, chains may be nevertheless strongly stretched in the direction perpendicular to the substrate , to avoid unfavorable monomer - monomer overlap in the case of sufficiently dense grafting .
effects of intrinsic chain stiffness , as are expected for semi - flexible polymers @xcite , did find rather little attention @xcite , apart from the case where one considers grafted semi - flexible chains in nematic solvents @xcite and their liquid crystalline order @xcite .
we also draw attention to some related problems where rod - like molecules densely packed and anchoring at substrates play a role , such as in langmuir monolayers of surfactants and related systems @xcite .
experimentally , the interactions between dna - grafted colloids @xcite , or poly(acrylic acid ) brushes with variable rigidity @xcite , have been comprehensively investigated by f. kremer and collaborators . in recent years , semi - flexible polymers find increasing interest particularly in the context of biophysical systems , such as cytoskeleton actin bundles @xcite biofilaments @xcite , and also biobased polymer brushes were recently created @xcite .
thus , a more comprehensive simulation study of brushes formed from semi - flexible polymers would be desirable to theoretically clarify the effect of chain stiffness on structure and dynamics of such systems , including also their response to external perturbations , such as shear and compressive forces . as is well known , one important possible application of polymer brushes formed from flexible polymers is their use as a lubricant @xcite , although the theoretical aspects of this response to shear and compression are still under discussion @xcite . in our preliminary publication @xcite , indeed , an anomalous response of brushes formed from rather stiff chains was found , caused by an onset of orientational order parallel to the substrate . in the present paper
we shall give a more complete account of this problem , focusing mainly on the structure and static behavior of such stiff brushes subject to compression .
in the next section [ sec_model ] , the model that is studied will be introduced and the simulation method will be described .
section [ sec_structure ] examines the parallel and perpendicular chain linear dimensions in rigid brushes as a function of bending rigidity and grafting density over a wide range , as well as the profiles of monomer density in the direction perpendicular to the grafting substrate , and the end monomer distribution .
then in section [ sec_compress ] we elaborate the topic that was already briefly addressed in our preliminary study , compression of such brushes by a piston , presenting new results , including also the variation with grafting density and chain length .
the hysteresis associated with the transition towards orientational order will also be investigated .
section [ sec_theory ] is devoted to an attempt for theoretical consideration within the mean - field approximation of the brush response to deformation . a brief account on the recovery dynamics of compressed stiff brushes
is then given in section [ sec_dynamics ] .
our conclusions are summarized in section [ sec_summary ] .
we start from the standard model for polymer brushes formed from completely flexible polymers @xcite and complement it by adding a bond angle potential .
this standard model is a bead - spring model in the continuum , where bonded effective monomers interact with the well - known finitely extensible non - linear elastic ( fene ) potential @xcite @xmath3 , \ ; 0 < r < r_o , \ ] ] where @xmath4 is the distance between the beads , and the spring constant @xmath5 as well as the maximum distance @xmath6 between neighboring monomers will be specified below .
in addition , for any pairs of monomers ( both bonded and non - bonded ones ) a truncated and shifted lennard - jones potential acts , the so - called weeks - chandler - andersen ( wca ) potential @xcite @xmath7 , \ ; r < r_c = 2^{1/6 } \sigma , \ ] ] while @xmath8 .
note that this potential is cut off in its minimum and both potential and forces are continuous for @xmath9 .
we henceforth take @xmath10 as our unit of length , and @xmath11 ( also boltzmann s constant @xmath12 is taken as unity , as usual ) .
no explicit solvent molecules are included , eqs .
[ eq1 ] , [ eq2 ] describe interactions between effective monomers and solvent molecules implicitly only , corresponding to the very good solvent regime .
the constants of the fene potential are chosen as @xmath13 so that the total potential between two subsequent monomers along the chain ( @xmath14 has a minimum at about @xmath15 .
the flexibility of the chains now is varied by introducing a bond angle potential @xmath16 that depends on the angle @xmath17 between the bonds between monomer pairs @xmath18 and @xmath19 respectively @xmath20\ ] ] for rather stiff chains we have @xmath21 and hence eq .
[ eq4 ] can be approximated as @xmath22 which shows that our polymer model can be viewed as a discretized version of the well - known kratky - porod model @xcite of semiflexible chains ( but , unlike the kratky - porod model , the present model fully accounts for excluded volume effects and is hence suitable for the study of rather dense systems as well ) .
note that the effective persistence length of an isolated chain would be @xmath23 ; of course .
due to packing effects in dense systems , local nematic order may arise as a result of interchain interactions that lead to nematic short or long range order , and then the persistence length can be appreciably larger ( see e.g. @xcite ) . for small @xmath24 , the persistence length is , in general , not a well defined quantity @xcite .
the chains are grafted at a planar @xmath25 impenetrable surface , choosing a square lattice arrangement of grafting sites , the lattice spacing chosen according to the desired grafting density @xmath26 . alternatively , also
a random arrangement of grafting sites has been considered , as discussed below .
the substrate surface exerts a repulsive interaction of a form analogous to eq .
[ eq3 ] on all monomers , @xmath27 , \quad 0 < z < z_c = \sigma _ w 2^{1/6}.\ ] ] here @xmath1 is the distance of a monomer from the wall , and the parameters are chosen as @xmath28 . in @xmath29-directions ,
periodic boundary conditions are used , and in @xmath1-direction the simulation box is closed with another repulsive wall ( at distance @xmath30 from the grafting surface ) where a potential of the same type as eq .
[ eq6 ] acts ( simply @xmath1 needs to be replaced by @xmath31 on the right hand side of eq .
[ eq6 ] ) .
this upper wall has no physical effect , as long as @xmath30 exceeds significantly the height , @xmath32 , of the unperturbed brush in equilibrium .
however , we shall consider also the effect of compressing the brush by reducing @xmath30 to values @xmath33 . as an initial condition ,
all chains are put into a straight rod configuration , choosing all angles @xmath34 .
note that the first bond is constrained to the @xmath1-direction , perpendicular to the grafting surface , so in the initial state all chains are oriented along the @xmath1-axis .
then the chain configurations are relaxed by standard molecular dynamics ( md ) simulations , applying the velocity verlet algorithm @xcite and a langevin thermostat .
thus the coordinates @xmath35 of the effective monomers evolve according to the newton equations of motion @xmath36 where the mass of effective monomers @xmath37 is chosen to be unity as well , @xmath38 is the total potential ( containing the terms from the interactions between monomers , eqs .
[ eq1 ] , [ eq2 ] , [ eq4 ] and the wall - monomer interaction , eq.[eq6 ] ) , @xmath39 is a friction coefficient and @xmath40 a random stochastic force .
the latter satisfies the fluctuation - dissipation relation @xcite ( @xmath41 denote the cartesian coordinates ) @xmath42 the friction coefficient @xmath39 is chosen as @xmath43 .
this choice ensures efficient equilibration .
we also note that with our choice of units , the md time unit is unity as well , @xmath44 . in our simulations , @xmath24 is varied from @xmath45 ( flexible chains ) up to @xmath46 , and chain lengths from @xmath47 to @xmath48 , so cases are included where the contour length and the persistence length are of the same order of magnitude .
grafting densities are varied from @xmath49 ( the `` mushroom regime '' @xcite ) to @xmath50 ( in the latter case monomer density inside of the brush corresponds to a concentrated solution ) . the number of chains was typically chosen to be @xmath51 , but in a few cases the lateral box linear dimensions were doubled ( so that @xmath52 ) so as to verify that finite size effects are still negligibly small .
snapshot pictures of polymer brushes for the case @xmath53 , @xmath54 , @xmath55 , and two choices of @xmath56 : @xmath45 ( fully flexible chains ) ( a ) , @xmath57 ( semi - rigid chains ) ( b ) .
the distance between observer and objects is the same in both cases.,title="fig : " ] snapshot pictures of polymer brushes for the case @xmath53 , @xmath54 , @xmath55 , and two choices of @xmath56 : @xmath45 ( fully flexible chains ) ( a ) , @xmath57 ( semi - rigid chains ) ( b ) .
the distance between observer and objects is the same in both cases.,title="fig : " ] as an example , fig .
[ fig1 ] shows typical snapshot pictures for a well - equilibrated brush at a grafting density @xmath54 and @xmath55 for two choices of @xmath56 , @xmath58 .
one can clearly see that for the flexible case the chain conformation is locally rather random , the overall stretching of the chains in @xmath1-direction becomes visible on larger scales only .
in contrast , the stiff chains show at most long wavelength transverse undulations , like a bunch of pliant rods .
this behavior is exactly what one expects to occur for worm - like chains , which hence are well modeled by our simulation . in the next section
we shall analyze the properties of these brushes in more detail .
we start with a discussion of the monomer density profile , @xmath59 , and the distribution of free ends , @xmath60 , for a typical choice of @xmath61 and @xmath62 , @xmath63 and @xmath64 , varying @xmath56 ( fig . [ fig2]a ) .
one notes that near the wall there is always a pronounced layering effect , which strongly increases with @xmath56 , although the density of the stiffer chains is smaller in the center of the brush ( near @xmath65 , as a rough measure of brush height @xmath66 , we may take the inflection point of the profile @xmath59 near its final decay for large @xmath1 ) .
this density decrease with increasing @xmath56 must occur since the stiffer chains are more stretched and the brush height @xmath32 increases with increasing @xmath56 up to its maximum value while the number of monomers stays constant .
this effect that stiffer chains exhibit more pronounced layering than flexible chains has already been pointed out for a different model @xcite .
note that the actual value of the density is @xmath67 and hence about @xmath68 for @xmath69 for @xmath70 , and @xmath71 for @xmath72 .
these densities are much smaller than melt densities , indicating a rather unexpected high degree of order for the semi - flexible brushes near the wall .
[ fig2]b shows that the distribution of chain ends becomes sharper with increasing chain stiffness .
the half - width @xmath73 scales as @xmath74 for large @xmath56 ( i.e. , @xmath75 scales inversely with the persistent length ) .
( a ) density distribution @xmath59 of the effective monomers in the brush , and distribution of chain ends , @xmath60 , plotted vs. the distance @xmath1 from the grafting surface , for @xmath76 , and three choices of @xmath56 , as indicated . the density profile for @xmath77
is indicated by shaded area .
note the normalizations @xmath78 , @xmath79 .
( b ) half - width of @xmath80 vs @xmath81 for @xmath82 and @xmath83 , shown as a @xmath84-@xmath84 plot .
the inset shows the relative increase of the brush height with @xmath56 . the asymptotic value @xmath85 is indicated by the dashed horizontal line.,title="fig : " ] ( a ) density distribution @xmath59 of the effective monomers in the brush , and distribution of chain ends , @xmath60 , plotted vs. the distance @xmath1 from the grafting surface , for @xmath76 , and three choices of @xmath56 , as indicated . the density profile for @xmath77 is indicated by shaded area .
note the normalizations @xmath78 , @xmath79 .
( b ) half - width of @xmath80 vs @xmath81 for @xmath82 and @xmath83 , shown as a @xmath84-@xmath84 plot .
the inset shows the relative increase of the brush height with @xmath56 .
the asymptotic value @xmath85 is indicated by the dashed horizontal line.,title="fig : " ] the height of the brush @xmath86 saturates for @xmath87 at its maximum value of fully stretched chains , @xmath88 , see the inset of fig .
[ fig2]b , so the distribution of chain ends becomes a delta - function ! fig .
[ fig3 ] presents the chain linear dimensions as functions of @xmath56 , for @xmath89 and @xmath90 .
as expected , the chains get more stretched as the polymers get stiffer .
interestingly , also the linear dimensions in @xmath29-directions parallel to the wall initially increase even stronger than the linear dimensions in @xmath1-direction . due to
the increased stretching , the density in the brush has decreased , and hence there is more space for the monomers to occupy . in particular , the large magnitude of this effect is surprising .
it is also evident from the snapshot - fig .
we also note that the linear dimensions @xmath91 reach their maximum in the range @xmath92 , i.e. , a range where the contour length is just a few times larger than the persistence length .
note that for a single semi - flexible chain one predicts that transverse fluctuations of monomers in the rod - like limit scale as @xcite @xmath93 normalized mean square gyration radii components and end - to - end distance components @xmath94 , @xmath95 in the directions perpendicular @xmath96 and parallel @xmath2 to the grafting surface , plotted vs. @xmath56 for @xmath82 and @xmath83 .
dashed horizontal lines indicate the asymptotic ratio values for @xmath97 whereby @xmath98 and @xmath99 ^ 2 $ ] . ] thus , fig .
[ fig3 ] suggests that a similar decrease of the linear dimensions in the @xmath29-direction ( transverse to the average orientation of the rod - like grafted chains in the brush , which is the @xmath1-direction ) occurs also in fairly dense brushes .
however , much more data ( for still larger values of @xmath56 ) would be required to test whether eq . [ eq9 ] holds here quantitatively .
note that for @xmath72 the @xmath29-components of the mean square gyration radius are still larger than their counterparts for flexible chains , although the brush height is close to its maximum value . for @xmath100
, these @xmath29-components must tend to zero .
[ fig4 ] studies the variations of chain linear dimensions with @xmath62 . while for flexible chains one knows that in the semi - dilute limit @xmath101 , and hence @xmath102 , that is , our data are compatible with a variation @xmath103 , but the variation with @xmath26 is distinctly weaker .
the values reached for these large values of @xmath62 indicate that the chains are stretched out like straight rods , since in this rod limit we simply have @xmath104 and @xmath105 , respectively .
note that for @xmath106 thus a saturation is reached .
( a ) plot of @xmath107 and @xmath108 vs. @xmath62 for @xmath109 and two choices of n , n = 40 and n = 80 . note the logarithmic ordinate scale ; the straight lines indicate an empirical exponential relation , @xmath110 .
( b ) log - log plot of @xmath111 and @xmath112 vs. @xmath62 .
straight lines indicate variations proportional to @xmath113 .
all data refer to @xmath114 .
both data for @xmath82 and @xmath115 are shown .
horizontal straight lines in ( b ) indicate the asymptotic results when the chains are fully stretched out like a straight rod along the @xmath1-axis , i.e. , @xmath116 and @xmath117 , respectively.,title="fig : " ] ( a ) plot of @xmath107 and @xmath108 vs. @xmath62 for @xmath109 and two choices of n , n = 40 and n = 80 .
note the logarithmic ordinate scale ; the straight lines indicate an empirical exponential relation , @xmath110 .
( b ) log - log plot of @xmath111 and @xmath112 vs. @xmath62 .
straight lines indicate variations proportional to @xmath113 .
all data refer to @xmath114 .
both data for @xmath82 and @xmath115 are shown .
horizontal straight lines in ( b ) indicate the asymptotic results when the chains are fully stretched out like a straight rod along the @xmath1-axis , i.e. , @xmath116 and @xmath117 , respectively.,title="fig : " ] similarly , for flexible chains one would have @xmath118 ; however , the present data suggest a much more rapid decay of @xmath119 with @xmath62 than this weak power law .
empirically we find that the decrease is compatible with an exponential decay : unfortunately , we do not have any explanation for this puzzling behavior .
however , it should be remarked that for semi - flexible brushes the regime where the crossover from semi - dilute brushes to mushroom - like behavior occurs is found for much smaller grafting densities than for brushes formed from flexible chains of the same length .
this fact simply results from the observation that the size of a chain in dilute solution scales as @xmath120 for flexible chains , while for semi - flexible chains the size is much larger @xcite , @xmath121 for very long chains @xmath122 ) and @xmath123 for chains with @xmath124 ; the bond length is denoted as @xmath125 in this scaling description , and can be identified with the distance @xmath126 of the minimum of our effective potential between neighboring monomers in our model . since we use here rather stiff chains of medium length , we are in the regime @xmath127 with our simulations always .
hence the crossover from mushrooms to semi - dilute brushes occurs for @xmath128 given by @xmath129 the scaling behavior typical for semi - flexible semi - dilute brushes hence is expected for @xmath130 .
this regime clearly is not explored in our simulations , which rather address the regime from concentrated solutions to melt densities .
in this section we consider the change of the state of the brush when it is compressed by a flat structureless piston parallel to the grafting surface that is brought to a height @xmath30 (= distance from the grafting surface ) less than the height @xmath32 of the free brush .
[ fig5 ] shows a few representative snapshot pictures and fig .
[ fig6 ] corresponding density profiles .
one can see that for the snapshot pictures of polymer brushes for the case of l = 32 , n = 40 , @xmath83 , and two choices of @xmath56 , @xmath131 ( a , b ) and @xmath132 ( c , d ) , applying a compression characterized by @xmath133 ( a , c ) and @xmath134 ( b , d ) .
, title="fig : " ] snapshot pictures of polymer brushes for the case of l = 32 , n = 40 , @xmath83 , and two choices of @xmath56 , @xmath131 ( a , b ) and @xmath132 ( c , d ) , applying a compression characterized by @xmath133 ( a , c ) and @xmath134 ( b , d ) .
, title="fig : " ] snapshot pictures of polymer brushes for the case of l = 32 , n = 40 , @xmath83 , and two choices of @xmath56 , @xmath131 ( a , b ) and @xmath132 ( c , d ) , applying a compression characterized by @xmath133 ( a , c ) and @xmath134 ( b , d ) .
, title="fig : " ] snapshot pictures of polymer brushes for the case of l = 32 , n = 40 , @xmath83 , and two choices of @xmath56 , @xmath131 ( a , b ) and @xmath132 ( c , d ) , applying a compression characterized by @xmath133 ( a , c ) and @xmath134 ( b , d ) . , title="fig : " ] flexible chain the density increase caused by the compression leads to a more pronounced layering of the monomers near the grafting surface ; also at the compressing upper surface the parabolic decay of the density profile found in the free brush is replaced by the density oscillations in the compressed brush .
for very strong compression the layering at the compressing wall is even more pronounced than near the grafting surface , where the brush conformation is more constrained due to the condition that the first bond of each grafted chain must density distribution of the effective monomers in the brush , @xmath135 , plotted vs distance @xmath1 from the grafting surface , for @xmath63 , @xmath136 , for two choices of @xmath137 , namely , @xmath138 and @xmath139 .
two choices of @xmath56 are included namely fully flexible chains @xmath140 , case ( a ) , and semiflexible ones @xmath141 , case ( b).,title="fig : " ] density distribution of the effective monomers in the brush , @xmath135 , plotted vs distance @xmath1 from the grafting surface , for @xmath63 , @xmath136 , for two choices of @xmath137 , namely , @xmath138 and @xmath139 .
two choices of @xmath56 are included namely fully flexible chains @xmath140 , case ( a ) , and semiflexible ones @xmath141 , case ( b).,title="fig : " ] be oriented in the @xmath1-direction perpendicular to the grafting surface . for the semi - flexible grafted chains the compression has the effect to clearly reduce the extent in @xmath1-direction over which layering can be seen , in comparison with the corresponding uncompressed brush .
the snapshot pictures ( fig . [ fig5 ] ) give a qualitative interpretation for this surprising behavior : while for the flexible brush the chains get uniformly compressed and the density increases gradually , the rather stiff chains exhibit a collective bending , and the interplay of the bending of the stiff chains and the dense packing of the effective monomers destroys the periodicity of the local density that we see in the uncompressed case .
it is found that the onset of collective chain bending does not start immediately when the distance @xmath30 between the compressing piston surface and the grafting substrate become equal to the uncompressed brush height @xmath32 , but only when the degree of compression @xmath142 exceeds a threshold value .
this fact is clearly recognized ( fig . [ fig7 ] ) when we study the normalized pressure @xmath143 ( note that the normal pressure @xmath144 , i.e. the diagonal component of the pressure tensor @xmath145 , is easily sampled using the virial theorem @xcite ) . for a free , uncompressed brush in thermal equilibrium @xmath146 , of course .
for very small compression , @xmath147 , the pressure is almost immeasurably small .
note , however , that there is some arbitrariness in defining @xmath32 exactly : here we have defined @xmath32 from the condition that @xmath59 \{fig .
[ fig2]a } has decreased to about @xmath148 of its value in the flat region of @xmath59 where the oscillations of @xmath59 have just decayed .
note that the definition in terms of the first moment @xmath149 of the density profile , @xmath150 , that is often used for flexible brushes and holds for the parabolic profile of the self - consistent field theory @xcite , is not useful for rather dense semi - flexible brushes , and hence not used here .
( a ) normal pressure @xmath143 for compressed brushes with chain length @xmath151 and grafting density @xmath152 plotted vs. the degree of compression , @xmath142 .
the brush contains @xmath153 chains .
both a flexible brush @xmath140 ) and a brush formed from rather stiff chains ( @xmath154 ) are included ; for the latter also the order parameter @xmath155 \{eq.[eq10 ] } and the orientational order parameter @xmath156 \{eq .
[ eq11 ] } are shown , as indicated .
@xmath156 plotted vs. @xmath137 for the case @xmath63 .
two values of the grafting density are shown @xmath157 and @xmath158 , respectively ) for both flexible @xmath140 ) and semi - flexible @xmath159 ) chains . ] while for flexible brushes , @xmath160 , the increase of the pressure is completely gradual for the full range of compressions , and can be described by a power law ( roughly @xmath161 for @xmath162 , as will be discussed below ) , the variation of the pressure for stiff chains is very different : there occurs a much faster rise up to a maximum , then the pressure decreases again to a shallow minimum , before a slow increase of the pressure occurs again . at large compressions
the pressure in semi - flexible brushes clearly is distinctly _ smaller _ than in their flexible counterparts !
the pressure maximum is due to the onset of collective orientational ordering of the chains in a particular direction in the @xmath29-plane .
a convenient way to measure this ordering in terms of the unit vectors @xmath163 of the projections of the last bond vector of each chain into the @xmath29-plane is @xmath164 as illustrated by the snapshots in fig .
[ fig_vec]a , b . if these unit vectors @xmath163 are randomly oriented in the @xmath29-plane , we obviously have @xmath165 , and this is what is observed for zero ( or very small ) compression .
this nonzero - plateau of @xmath155 at zero compression is hence just a trivial finite size effect . on the other hand ,
if spontaneous symmetry - breaking occurs , and all unit vectors are parallel to each other , we would have @xmath166 .
such definitions of root mean square order parameters are well known from simulation studies of phase transitions in isotropic ferromagnets @xcite .
this gradual onset of ordering also shows up in the average angle that bonds make with the @xmath1-direction . thus fig .
[ fig7]b includes also data on @xmath167 defined as usual by @xmath168 while in the initial period of compression , up to the pressure maximum , @xmath167 stays essentially constant , then a gradual decline of @xmath156 sets in .
recall that @xmath169 for rods oriented along the @xmath1-axis while @xmath170 for rods oriented parallel to the @xmath29-plane .
however , this smooth behavior of the order parameters @xmath171 and @xmath172 , indicative of a continuous ( second - order ) phase transition , is at variance with the observation of a loop ( hysteresis ) of the pressure in its variation with the compression .
such loops normally would be associated with a normal pressure @xmath143 plotted vs. @xmath30 for the case @xmath63 , @xmath72 and @xmath173 , comparing runs where the pressure is stepwise increased and decreased again . in a short run , at each value of @xmath30 the system is held @xmath174 md time units whereas in a long run , the equilibration time at every state was @xmath175 times longer .
the variance of measured values is also plotted as error bars . ]
discontinuous ( first order ) phase transition . but a more careful consideration of equilibration reveals , ( fig .
[ fig8 ] ) , that although some hysteresis between runs , where the compression is stepwise increased or decreased , is actually observed , it must be interpreted as an observation time effect : the amount of the hysteresis is almost completely gone , if the observation times are chosen @xmath175 times larger ! of course , we expect that close to a second order transition relaxation times in the system become very long , due to `` critical slowing down '' @xcite .
thus , hysteresis is also known to occur for simulations of second order transitions @xcite , if the observation times at the individual state points near the transition are chosen too short . while for most systems this problem is not important in practice @xcite , here the situation differs because densely arranged stiff polymers are slow objects : the @xmath29-like order of the last bond must be shared by a related order of the inner bonds of each chain as well . figs .
[ fig7 ] , [ fig8 ] focus on the behavior in the vicinity of the phase transition only .
it turns out , ( fig .
[ fig7]b ) , that with strong compression the orientation of the bonds , which in figs .
[ fig7 ] , [ fig8 ] is still mostly along the @xmath1-direction , can be changed to a preference of @xmath176 direction : then , @xmath177 for semi - flexible brushes .
in contrast , flexible brushes at similar compressions , @xmath178 , would exhibit a rather random orientation of bond vectors , @xmath179 .
it is interesting to investigate this transition between a weakly compressed semi - flexible polymer brush ( where @xmath180 ) , and the `` buckling deformation '' of individual stiff chains @xcite ) .
one should bear in mind that this transition depends on the various parameters of the problem : bending stiffness @xmath56 , grafting density @xmath62 , and chain length @xmath61 .
( a ) normal pressure @xmath143 plotted vs. distance @xmath30 for four choices of @xmath26 and two choices of @xmath56 , @xmath181 ( empty symbols ) and @xmath72 ( full symbols ) .
all data are for @xmath63 .
the inset shows the variation of the critical pressure @xmath182 with @xmath62 for @xmath114 .
dashed line denotes a fit @xmath183 .
( b ) normal pressure @xmath143 plotted vs. @xmath137 for the case @xmath184 , and a number of chain lengths from @xmath47 to @xmath55 , as indicated .
the inset shows @xmath182 vs chain length @xmath61 , dashed line is an empirical fit with @xmath185 . ]
[ fig10 ] presents a selection of our data on this issue .
one can see that the onset of compression results in a steeper @xmath143 curve , when @xmath62 increases ( fig .
[ fig7]b ) , and also the almost flat plateau region of @xmath143 , that is reached at compressions beyond the maximum , rises with @xmath26 very distinctly .
the variation with @xmath56 is rather weak , however : the primary effect is a slight increase of the brush height @xmath32 with @xmath56 , while the height of the flat region in the pressure gets slightly smaller with increasing @xmath56 ( presumably the chains order the better the stiffer they become ) .
moreover , the onset of bending occurs at _ smaller _ degree of compression , the denser the brush , @xmath62 , is .
an important detail in this picture are also the observed occasional abrupt _ slides _ of the brush chains , as in the @xmath186 curves for @xmath187 at @xmath188 , whereby some excess elastic energy stored in the system is released .
it is conceivable that this effect is due to the finite ( nonzero ) rate of brush compression in the course of the md simulation .
a very pronounced effect is seen when the chain lengths is varied , however ( fig .
[ fig10]b ) : very stiff short chains exhibit much larger pressures for comparable compressions @xmath137 .
note that chains with @xmath47 , @xmath72 are in the limit where the persistence length @xmath189 ( which equals @xmath56 for our choice of units ) exceeds by far the contour length @xmath190 , while for the chains with length @xmath63 to @xmath191 both lengths are comparable .
interestingly , the data shown in fig . [ fig10]b , imply that the pressure , when plotted versus @xmath137 for @xmath192 ( i.e. , after the onset of the @xmath29-like ordering where chains bend uniformly in a chosen direction ) , depends on the persistence length only , and not on the chain length ( while @xmath32 does depend strongly on @xmath61 , as shown already in fig .
[ fig2 ] ) .
the insets in fig .
[ fig10 ] present the variation of the critical pressure @xmath182 with grafting density .
@xmath182 itself is defined as the pressure @xmath144 where the second - order transition , manifested by the onset of lateral orientation order ( in the @xmath29-plane ) , takes place . for simplicity ,
we have estimated it roughly from the pressure maximum ( a more precise study of this transition , e.g. , by finite - size scaling methods @xcite , must be left to further work ) .
an important objection that could be raised against our findings is the suspicion that the regular square - lattice arrangement of our grafting sites leads to a structure of a dense semi - flexible brush that is much more regular than a real brush , for which the grafting sites are distributed at random . to check this caveat
, we also performed a simulation for a typical choice of parameters ( @xmath193 ) , where we compared the pressure variation with compression for two brushes , one with an ordered arrangement of grafting sites , and another one with an ( almost ) random grafting - fig .
[ fig_ord_rand ] . by `` almost '' random , we mean that the randomly chosen grafting sites were abandoned , if they were closer than a distance @xmath194 to a grafting site that was already present ( so as to account for the fact that the chemical groups that effect the grafting for two different chains are subject to excluded - volume effective repulsion ) .
chains with @xmath63 and @xmath195 .
the inset displays variation of the respective order parameter @xmath155 with height @xmath30 of the compressing flat piston .
[ fig_ord_rand ] ] one sees that the pressure variation with piston height @xmath30 for the random choice of grafting sites hardly differs from that for regular arrangement except when the onset of order starts . in this narrow interval
the @xmath196-relationship in the case of random grafting is slightly less sharp but rather smeared over a slightly broader interval of heights @xmath30 , presumably indicating an effective `` dilution '' of the collective response due to randomness .
we start with the observation that for grafting densities @xmath197 and stiff chains ( @xmath198 ) , the polymers in the uncompressed state ( for our choice of rather small chain length , @xmath199 ) are stretched out to their maximal length , that is , a rod - like conformation prevails .
transverse fluctuations are very small , as the @xmath29-components of the gyration radius show - cf .
entropy plays then relatively little role and the equilibrium properties of a compressed brush in this limit should follow from a simple mechanical description . the bending energy of a chain ( including the uncompressed case ) can be approximated as @xmath200 in the last step one assumes that the distribution of bending energy along the backbone of the chain is approximately uniform .
[ fig_angle ] supports of consecutive bonds along the chain backbone with the normal to the grafting surface at different degrees of compression @xmath137 , as indicated .
here @xmath201 , and @xmath72 .
( b ) the same as in ( a ) for the angle @xmath202 between neighboring bonds .
[ fig_angle ] , title="fig : " ] of consecutive bonds along the chain backbone with the normal to the grafting surface at different degrees of compression @xmath137 , as indicated .
here @xmath201 , and @xmath72 .
( b ) the same as in ( a ) for the angle @xmath202 between neighboring bonds .
[ fig_angle ] , title="fig : " ] this conclusion , and also implies the very interesting result that @xmath203 is independent of the compression @xmath137 , at least for small deformations @xmath204 .
recall from fig .
[ fig10]b that for this choice of parameters the pressure plateau ( associated with the onset of lateral orientational order ) starts at @xmath205 . at zero pressure , and neglecting any interaction among the chains , one would find ( recall that for @xmath146 the azimuthal angle @xmath206 is still random and uniformly distributed , @xmath207 is a polar angle , and @xmath208 for @xmath21 can be used ) @xmath209 for the example shown in fig .
[ fig_angle]b we would obtain @xmath210 , the value that is actually observed is close to this value at @xmath211 . with increasing compression , larger angles for smaller @xmath212
are observed , i.e. , the rod - like chains undergo strong bend in the vicinity of the substrate while staying straight away of this local deformation . the opposite limit that we are considering is the case of perfect orientational ordering .
then the azimuthal angle @xmath213 of all bonds from monomer @xmath212 to monomer @xmath214 remains the same for all bonds of all chains , and the polar angles @xmath215 of the bonds with respect to the @xmath1-axis lie in a plane characterized by this azimuthal angle @xmath206 and containing the @xmath1-axis . in this limit
a plausible assumption is that fluctuations of @xmath202 relative to their mean value @xmath216 are small .
therefore we would conclude that @xmath217 this result follows from the assumption that @xmath218 has a minimum , subject to the constraint that in the case of perfect azimuthal ordering the last angle @xmath219 .then one simply has to minimize the function @xmath220 with respect to all the @xmath202 with @xmath221 being a lagrange multiplier . in this case one could derive a simple relation between @xmath222 and the distance @xmath30 of the topmost monomer from the surface , as far as @xmath223 , @xmath224 ignoring all other interactions among monomers , one then finds for the force @xmath225 exerted by a chain in the brush on the compressing plane @xmath226 the pressure @xmath144
would then be obtained simply by multiplying with the grafting density @xmath62 , to yield ( for small @xmath227 ) the result @xmath228 \label{eq_p}\ ] ] while @xmath229 and @xmath230 .
our simulation data , however , imply that such a description ( intended to hold for the limit @xmath97 ) still lacks an important ingredient which we identify as the repulsive potential energy between the monomers .
this energy clearly increases when we compress the system ( simply due to density increase ) . writing the volume fraction , occupied by the monomers , as @xmath231 , whereby @xmath232 is taken as an estimate of the volume of a monomer , a mean - field estimate of the repulsive energy in the rod - like regime that prevails for @xmath233 yields : @xmath234 where @xmath235 is a constant ( proportional to the second virial coefficient ) . the corresponding contribution to the pressure , @xmath236 ,
has to be added to eq .
( [ eq_p ] ) and then a revised prediction for the critical pressure follows @xmath237 fig .
[ fig_angle]a indicates , that the result @xmath238 holds only for very small degrees of compression whereas for stronger compression , @xmath239 , it does not comply with our data .
snapshots of brush configurations - fig .
[ fig_snap_def ] - manifest instead that the polar angle @xmath215 attains its ultimate value @xmath240 in the narrow interval encompassing the first few bonds only . beyond this region of strong deformation in the vicinity of the grafting surface
, the compressed semi - rigid polymer brush organizes itself into a quasi - crystalline nematic order of parallel bonds with constant inclination angle @xmath240 .
( [ eq_pcr ] ) was the motivation for the fit , used in the inset of fig .
[ fig10]b , in agreement with simulation measurements . in the inset of fig .
[ fig10]a , we found it useful to include a term that could be attributed to the third virial coefficient and turn relevant in the limit of very large grafting density @xmath241 . in what follows we
replace the discrete index @xmath212 of each monomer by a continuous variable : @xmath242 . from figs .
[ fig_angle]a , [ fig_snap_def ] we recognize that the bond orientations @xmath243 along the chain contour , cf .
. [ eq_u ] , describes a nontrivial curve in the plane , singled out by the symmetry breaking of the azimuthal angle @xmath206 due to orientational order .
if one would only take the bending energy into account , one would obtain the trivial linear variation @xmath244 of eq .
[ eq_u_mean ] . in order to derive a reasonable approximation for @xmath243 ,
it is necessary to take into account both the energy due to compression , that is , a pressure times volume term , which is @xmath245 , as well as the energy of monomer - monomer repulsion , @xmath246 . indeed , with growing @xmath144 , as the stiff chains bend more and more , the monomers get closer to one another , unlike the case when the chains are vertically stretched out like stiff rods . a simple geometric argument , cf .
[ fig_theta_s]a , yields thus @xmath247 for @xmath248 , eq . ( [ eq_u_rep ] ) leads back to eq .
( [ eq_virial ] ) , of course . putting all terms together , the free energy that needs to be minimized , becomes ( using the abbreviations @xmath249 ) becomes since the normal distance between rods in an array of rods , tilted by angle @xmath250 scale like @xmath251 with respect to the @xmath1-axis .
@xmath252,\ ] ] which yields upon minimization with respect to @xmath253 the following euler - lagrange equation : @xmath254 in the region where @xmath243 is still small , one can expand eq .
( [ eq_el ] ) keeping only the terms up to order @xmath255 , arriving at @xmath256 where the abbreviations @xmath257 have been introduced .
note that considerable deviations of @xmath250 from zero , and the ensuing increased importance of the repulsion energy , eq .
( [ eq_u_rep ] ) , emerge when the applied compressing pressure @xmath144 is sufficiently large so that the condition @xmath258 , i.e. , @xmath259 in eq .
( [ eq_m ] ) appears physically reasonable .
, the distance @xmath260 between neighboring rods diminishes like @xmath261 , where @xmath262 is the separation between grafting points .
( b ) variation of the polar angle @xmath263 along the chain contour length of deformed polymer chains with @xmath264 beads , @xmath265 , and different values of the parameter @xmath266 , as indicated . apparently , the bending increases with growing pressure @xmath267 , and even more so for softer , @xmath268 , or less dense brushes .
[ fig_theta_s ] , title="fig : " ] , the distance @xmath260 between neighboring rods diminishes like @xmath261 , where @xmath262 is the separation between grafting points .
( b ) variation of the polar angle @xmath263 along the chain contour length of deformed polymer chains with @xmath264 beads , @xmath265 , and different values of the parameter @xmath266 , as indicated . apparently , the bending increases with growing pressure @xmath267 , and even more so for softer , @xmath268 , or less dense brushes . [ fig_theta_s ] , title="fig : " ] the approximation , involved in reducing eq .
( [ eq_el ] ) to eq .
( [ eq_kg ] ) , holds for not too large angles @xmath250 only , and also eq .
( [ eq_f_nl ] ) can only hold at high grafting density @xmath62 ( otherwise the neglect of entropy and the assumption of perfect symmetry breaking of the azimuthal angle can not be true ) .
for sufficiently strong pressure , large angles @xmath269 occur , and one must then solve the full eq .
( [ eq_el ] ) .
the governing eq .
( [ eq_kg ] ) , derived in the present treatment , is familiar from the so called @xmath270-model of statistical mechanics , and has emerged in different contexts . with respect to the boundary condition , @xmath271 , one obtains the well - known soliton ( kink ) solution @xmath272 where @xmath37 and @xmath273 are positive constants , cf .
( [ eq_m ] ) . indeed , fig .
[ fig_theta_s ] demonstrates that the solution , eq .
( [ eq_kink ] ) , nicely reproduces the observed md data , shown in fig .
( [ fig_angle])a , at least qualitatively .
eventually , one should note that going back to the case of small deformation when @xmath274 , and the energy contribution due to chain repulsion , eq .
( [ eq_u_rep ] ) , can be neglected , a solution of our governing equation , eq .
( [ eq_el ] ) , recovers the expression for the critical pressure @xmath182 , namely , eq .
( [ eq_pcr ] ) , as expected @xcite .
it is worth mentioning , that this result , eq . ( [ eq_pcr ] ) , has been first obtain by l. euler and d. bernoulli for the famous beam bending problem in the year @xmath275 .
in polymer brushes one may study relaxation phenomena on many different levels : ( i ) fluctuations of individual monomers and and motion of chains as a whole in thermal equilibrium @xcite , ( ii ) collective relaxation phenomena as described by dynamic correlations of the local monomer density in the brush , ( iii ) the nonlinear dynamic response , associated with compression and shear , e.g. @xcite .
a comprehensive treatment for any of these topics for brushes formed from rather stiff chain molecules is beyond the scope of the present paper relaxation functions of perpendicular and parallel mean square components of the gyration radius of the chains .
the polymer brush recovery was monitored after the compressing piston was instantaneously removed at time @xmath276 .
all cases shown refer to @xmath63 , @xmath109 , and @xmath277 .
case ( a ) shows @xmath152 , while ( b ) compares three grafting densities , as indicated .
the characteristic relaxation time @xmath278 for initial brush recovery is shown in the inset as function of @xmath62 along with a fit by @xmath279 .
a comparison with the recovery kinetics of a polymer brush at @xmath280 , comprised of totally flexible chains , @xmath281 , is presented in ( c ) and ( d ) . here
again @xmath63.,title="fig : " ] relaxation functions of perpendicular and parallel mean square components of the gyration radius of the chains .
the polymer brush recovery was monitored after the compressing piston was instantaneously removed at time @xmath276 .
all cases shown refer to @xmath63 , @xmath109 , and @xmath277 .
case ( a ) shows @xmath152 , while ( b ) compares three grafting densities , as indicated .
the characteristic relaxation time @xmath278 for initial brush recovery is shown in the inset as function of @xmath62 along with a fit by @xmath279 .
a comparison with the recovery kinetics of a polymer brush at @xmath280 , comprised of totally flexible chains , @xmath281 , is presented in ( c ) and ( d ) . here
again @xmath63.,title="fig : " ] relaxation functions of perpendicular and parallel mean square components of the gyration radius of the chains .
the polymer brush recovery was monitored after the compressing piston was instantaneously removed at time @xmath276 .
all cases shown refer to @xmath63 , @xmath109 , and @xmath277 .
case ( a ) shows @xmath152 , while ( b ) compares three grafting densities , as indicated .
the characteristic relaxation time @xmath278 for initial brush recovery is shown in the inset as function of @xmath62 along with a fit by @xmath279 .
a comparison with the recovery kinetics of a polymer brush at @xmath280 , comprised of totally flexible chains , @xmath281 , is presented in ( c ) and ( d ) . here
again @xmath63.,title="fig : " ] relaxation functions of perpendicular and parallel mean square components of the gyration radius of the chains .
the polymer brush recovery was monitored after the compressing piston was instantaneously removed at time @xmath276 .
all cases shown refer to @xmath63 , @xmath109 , and @xmath277 .
case ( a ) shows @xmath152 , while ( b ) compares three grafting densities , as indicated .
the characteristic relaxation time @xmath278 for initial brush recovery is shown in the inset as function of @xmath62 along with a fit by @xmath279 .
a comparison with the recovery kinetics of a polymer brush at @xmath280 , comprised of totally flexible chains , @xmath281 , is presented in ( c ) and ( d ) . here again @xmath63.,title="fig : " ] yet here we present few results which relate to the relaxation ( brush recovery ) , associated with changes of the compression @xmath137 of the brush ( recall the hysteresis observed near the onset of lateral orientational ordering )
. a very straightforward computer experiment , which could also possibly carried out in the laboratory , considers the relaxation of a compressed brush after the compressing piston at time @xmath282 is suddenly removed and the brush relaxes from its height @xmath283 back to its uncompressed state with height @xmath32 .
[ fig_kin ] shows the time dependence of the @xmath1- and @xmath284components of both the end - to - end vector and gyration radius square for few typical cases of such computer experiment .
the data are shown as semi - logarithmic plots which would lead to straight lines , if the relaxation functions plotted were simple exponential ( which obviously is not the case ) .
[ fig_kin]a indicates that the relaxation behavior of the @xmath284 and @xmath1-components is similar , and fig .
[ fig_kin]b shows that the initial decay becomes much faster with increasing grafting density .
[ fig_kin ] includes also an empirical fit to a function @xmath285 where the relaxation time @xmath278 is a second adjustable parameter , @xmath286 .
these measurements are compared to those concerning time recovery of an initially compressed brush comprised of flexible chains , @xmath45 , and otherwise identical parameters - fig .
[ fig_kin]c , d. apparently , in the absence of chain stiffness the relaxation kinetics for the @xmath1 and @xmath284components differs significantly : components perpendicular to the grafting surface undergo short - termed fast recovery followed by a much slower secondary relaxation whereas the relaxation of the parallel components appears to be governed by a power law @xmath287 - fig .
[ fig_kin]d .
in this work molecular dynamics simulations of dense polymer brushes , formed from semi - flexible chains were presented , focusing on the regime where persistence length and contour length of these rather stiff macromolecules are comparable . in addition , the chemical nature of the grafted first bond renders it perpendicular to the underlying surface and , therefore , shapes the overall brush behavior .
we show that this behavior is very different from what is known for brushes comprised of flexible chains .
it turns out that the chains in the brush at high grafting density behave nearly as rigid rods , and the @xmath1-components of the chain end - to - end distance and gyration radius square take their maximal possible values , provided the chemistry of chain grafting fixes the first bond perpendicular to the substrate .
it is shown that fluctuations of the chain in the transverse @xmath2-directions , perpendicular to the @xmath1-axis along which the rods are oriented , are extremely small for large bending stiffness , and decrease like a simple exponential with growing grafting density @xmath62 - cf .
in contrast , the lateral size of a chains , @xmath288 , changes non - monotonically with stiffness and goes through a maximum for a finite degree of rigidity .
this behavior is very different from what is known for flexible chains in brushes in the case of semi - dilute and concentrated monomer densities in the brush . only for the case of melt densities ,
crystallization in polymer brushes has been predicted @xcite which may be related to our observations . in the present work we have focused on the compression of these stiff chain brushes and on the resulting orientational order of the bonds in the @xmath29-directions during bending .
we have demonstrated that the onset of this order is a continuous phase transition where the rotational symmetry in the @xmath29-plane is broken . however , since all bonds of the chains in a perfectly ordered system would have to share this ordering in the simulation box , requiring large lateral monomer displacements in the same direction , there is obviously a commensurability conflict between the lateral deformation of such a brush and the periodic boundary conditions .
this singles out the @xmath289 axes , imposing a cubic symmetry of the bending phenomenon . due to these problems ,
a study of the critical behavior of this phase transition would be premature , and has not been attempted . in order to obtain theoretical guidance on the observed phenomenon ,
some phenomenological considerations were presented .
again we emphasize that these considerations and the resulting interesting predictions , cf .
section [ sec_theory ] , should be taken as a first step only while a more systematic theory even on the mean - filed level remains a challenge for the future .
one interesting feature is the finding that for large compression the resilient pressure of such stiff chain brushes , exerted on the compressing piston , is much smaller that for flexible chain brushes of the same chain length and grafting density .
thus , a study of shear forces between two stiff chain brushes would also be very interesting but must be left as a future task too . in conclusion
, we hope that the present work will stimulate experimental interest in this subject .
we also believe that some related phenomena may take place in dense arrangements of stiff biopolymers in the context of complex biological structures .
we are indebted to friederike schmid , oleg borisov , and burkhard dnweg for instructive discussions of this work .
this investigation was supported by the deutsche forschungsgemeinschaft ( dfg ) under grant no bi 314/23 .
99 s. milner , _ science _ , 1991 , * 251 * , 905 .
a. halperin , m. tirrell and t. p. lodge , _ adv .
sci . , _ 1992 , * 100 * , 31 .
g. s. grest and m. murat , in _
monte carlo and molecular dynamics simulations in polymer science _ , ed . by k. binder , p. 476
, oxford univ
. press , new york , 1995 .
j. klein , _ annu .
_ , 1996 , * 26 * , 581 . i. szleifer and m. carignano , _ adv
phys_. , 1996 , * 94 * , 165 .
g. s. grest , _ adv .
sci_. , 1999 , * 138 * , 149 .
l. leger , e. raphasel , and h. hervet , _ adv .
_ , 1999 , * 138 * , 185 .
r. c. advincula , w. j. brittain , k. c. caster , and j. rhe , ( eds . ) _ _ polymer brushe__s , wiley - vch , weinheim , 2004 .
r. descas , j .- u .
sommer , and a. blumen , _ macromolecular theory simul _ , 2008 , * 17 * , 429 . k. binder , t. kreer , and a. milchev , _ soft matter _
, 2011 , * 7 * , 7159
. k. binder and a. milchev , _ j. polym .
b : polym . phys . _ , 2012 , * 50 * , 1515
. s. alexander , _ j. phys .
( paris ) _ , 1977 , * 38 * , 983 .
p. g. de gennes , _ macromolecules _ , 1980 , * 13 * , 1069 .
grosberg and a. r. khokhlov,_statistical physics of macromolecules _ , aip press , new york , 1994 .
m. carignano and i. szleifer , _ j. chem .
phys_. , 1993 , * 98 * , 5006 . g. g. kim and k. char , bull . _ korean chem .
soc_. 1999 , * 20 * , 1026 .
chen and y .- a .
fwu , _ phys . rev .
_ e , 2000 , * 63 * , 011506 .
a. gopinath and i. mahadevan , _ proc .
london ser .
a _ , 2011 , * 467 * , 1665 .
a. milchev and k. binder , _
phys . _ , 2012 , * 136 * , 104901 .
d. r. m. williams and a. halperin , _ europhys_. lett .
1992 , * 19 * , 963 .
d. r. m. williams and a. halperin , _ macromolecules _ , 1993 , * 28 * , 2025 .
a. halperin and d. r. m. williams , _ europhys_. lett .
1993 , * 21 * , 575 .
d. r. m. williams and a. halperin , _
macromolecules _ , 1993 , * 26 * , 4208 .
d. r. m. williams and a. halperin , _ langmuir _ , 1995 , * 11 * , 3626
. f. m. haas , r. hilfer , and k. binder , _ j. phys
_ 1996 , * 100 * ,
15290 . v. m. kaganer , h. mhwald , and p. dutta , _ rev .
phys_. 1999 , * 71 * , 779 .
f. schmid , d. johannsmann , and a. halperin , _ j. phys .
ii ( paris ) _ , 1996 , * 6 * , 1331 k. kegler , m. salomo , and f. kremer , _ phys .
_ 2007 , * 98 * , 058304 k. kegler , m. konieczny , g. dominguez - espinosa , c. gutsche , m. salomo , f. kremer , and c. n. likos2 , _ phys .
_ 2008 , * 100 * , 118302 g. dominguez - espinosa , a. synytska , a. drechsler , c. gutsche , k. kegler , p. uhlmann , m. stamm , f. kremer , _ polymer _ , 2008 , * 49 * , 4802 g. wong , a. lin , j. tang , y. li , p. janmey , and c. safinya , _ phys .
lett . _ , 2003 , * 91 * , 018103 . j. h. shin , m. l. gardel , l. mahadevan , p. matsudaira , and d. weitz , _ pnas _ 2004 , * 101 * ,
0636 . j. kierfeld , p. gutjahr , t. khne , p. kraikivski , and r. lipowsky , _ j. comput
nanosci_. 2006 , * 3 * , 898 .
m. bathe , c. heussinger , m. m. a. e. claessens , a. r. bausch and e. frey , _ biophys_. j. 2008 , * 94 * , 2955 .
o. lieleg , k. m. schmoller , c. j. cyron , y. luan , w. a. wall and a .r .
bausch , _ soft matter _ , 2009 , * 5 * , 1796 .
j. kierfeld , k. baczynski , p. gutjahr and r. lipowsky , _ aip conf .
proc_. , 2008 , * 1002 * , 151 .
s. k. lakkaraju and w. hwang , _ phys .
lett_. , 2009 , * 102 * , 118102 . j. r. blundell and e. m. terentjew , _ soft matter _ , 2009 , * 5 * , 4015 .
y. higari , r. okazaki , and a. takahara , _ acs macro lett_. , 2012 , * 1 * , 1124 . t. moro , y. takatori , k. ishihara , t. konno , y. takigawa , t. matsushita , u. chung , k. nakamura , and h. kawaguchi , _ nat .
mater_. 2004 , * 3 * , 829 .
j. klein , proc .
, 2006 , * 220 * , 691 .
j. klein , _ science _
, 2009 , * 323 * , 47 .
s. t. milner , t. a. witten , and m. e. cates , _ macromolecules _ 1988 , * 21 * , 2610
. s. t. milner , _ europhys .
lett_. 1988 , 7 , 695 .
e. b. zhulina , o. v. borisov , and v. a. prymatsyn , _
_ 1990 , * 137 * , 495 .
a. galuschko , l. spirin , t. kreer , a. johner , c. pastorino , j. wittmer , and j. baschnagel , _ langmuir _ , 2010 , * 26 * , 6418
. s. m. balko , t. kreer , p. j. costanzo , t. e. patten , a. johner , t. l. kuhl , and c. m. margues , _ plos one _ , 2013 , * 8 * , issue 3 , e58392 .
t. kreer and s. m. balko , _ acs macro lett .
_ 2013 , a. milchev and k. binder , _ epl _ , 2013 , * 102 * , 58003 . m. murat and g. s. grest , _ macromolecules _ 1989 , * 22 * , 4054 . g. s. grest and k. kremer , _ phys .
a _ 1986 , * 33 * , 3028 .
j. d. weeks , d. chandler , h. c. andersen , _ j. chem .
phys . _ , 1971 , * 54 * , 5237
. o. kratky and g. porod , _ j. colloid .
_ , 1949 , * 4 * , 35 .
v. a. ivanov , a. s. rodionova , j. a. martemyanova , m. r. stukan , m. mller , w. paul and k. binder , _ macromolecules _ , 2013 , ( submitted ) h .- p .
hsu , w. paul and k. binder , _ macromolecules _ , 2010 , * 43 * , 3094 .
m. p. allen and d. j. tildesley , _ computer simulation of fluids _
( clarendon press , oxford , 1987 ) .
i. k. snook , _ the langevin and generalized langevin approach of the dynamics of atomic , polymeric , and colloidal systems _
( elsevier , amsterdam , 2007 ) t. odijk , _ macromolecules _ , 1983 , * 16 * , 1340 h .-
hsu and k. binder , _
_ , 2012 , * 136 * , 024901
. k. binder and d. w. heermann , _ monte carlo simulation in statistical physics .
an introduction .
_ ( springer , berlin , 2010 ) p. c. hohenberg and b. i. halperin , _ rev .
_ 1977 , j. r. blundell and e. m. terentjew , _ soft matter _ 2009 , * 5 * , 41015 .
gui - li he , holger merlitz , jens - uwe sommer and chen - xu wu , _ eur .
e _ 2007 * 24 * , 325 d. reith , a. milchev , p. virnau , and k. binder , _ macromolecules _ , 2012 , 45 , 4381 g. arya , _ j. phys . chem .
b _ , 2010 ,
* 114 * , 15886 . | using a coarse - grained bead - spring model for semi - flexible macromolecules forming a polymer brush , structure and dynamics of the polymers is investigated , varying chain stiffness and grafting density . the anchoring condition for the grafted chains
is chosen such that their first bonds are oriented along the normal to the substrate plane .
compression of such a semi - flexible brush by a planar piston is observed to be a two - stage process : for small compressions the chains contract by `` buckling '' deformation whereas for larger compression the chains exhibit a collective ( almost uniform ) bending deformation .
thus , the stiff polymer brush undergoes a @xmath0-nd order phase transition of collective bond reorientation .
the pressure , required to keep the stiff brush at a given degree of compression , is thereby significantly _ smaller _ than for an otherwise identical brush made of entirely flexible polymer chains ! while both the brush height and the chain linear dimension in the @xmath1-direction perpendicular to the substrate increase monotonically with increasing chain stiffness , lateral @xmath2 chain linear dimensions exhibit a maximum at intermediate chain stiffness .
increasing the grafting density leads to a strong decrease of these lateral dimensions , compatible with an exponential decay .
also the recovery kinetics after removal of the compressing piston is studied , and found to follow a power - law / exponential decay with time . a simple mean - field theoretical consideration , accounting for the buckling / bending behavior of semi - flexible polymer brushes under compression , is suggested . |
in a recent experimental work , sistla , et al .
@xcite observed a novel granular circulation occurring in a flat - bottomed cylindrical container ( or `` pan '' ) subjected to coupled horizontal and vertical vibrations .
the apparatus studied was a commonly available industrial device used mainly for sieving and polishing of granular particles .
of particular interest was the appearance of a `` toroidal '' circulation , in which granular flow occurred simultaneously in the tangential direction ( i.e. a circulation in the horizontal plane about the vertical axis ) , as well as radially ( i.e. particles flowed outward along the bottom surface and then back to the center of the pan along the top surface of the granular bed ) . in this mode ,
the trajectory of a single particle would therefore approximate a helix on the surface of a torus . among several other phenomena
, it was noted that the direction of the tangential flow was , under most of the conditions explored , opposite to that of the orbital motion of the center of the pan . however , under some conditions , the tangential flow would reverse direction , and move in the same direction as the pan center .
the authors noted that the phase shift relating the vertical and horizontal vibrations seemed to control this reversal .
whatever their origin , these phenomena indicate a non - trivial interplay of the horizontal and vertical vibrations , and their interaction with the granular bed . since the bottom surface of the pan in the apparatus was a rough wire screen , the authors of ref .
@xcite suggested that entrainment of particles to the motion of the bottom surface might play an important role in establishing the direction of the observed granular circulations .
motivated by the complexity of phenomena observed in ref .
@xcite , we present here a computer simulation study of a granular matter system enclosed in the same container geometry as in ref .
@xcite , and subjected to a similar form of horizontal and vertical vibration . as we will show below , we are able to reproduce the major dynamical modes of the granular bed observed experimentally .
we also explore the dynamical behavior beyond the range studied experimentally , and discover an entire family of circulation reversal transitions , affecting both the tangential and radial circulations . by analyzing the interaction of the particle bed with the rough pan bottom , we are able to show how particle entrainment , along with the phase shift relating the horizontal and vertical vibrations , together conspire to control the direction of the observed circulations .
although most previous work on granular circulation has focussed on either purely vertical or purely horizontal vibrations ( see e.g. refs .
@xcite ) , a growing body of research is emerging related to granular flows generated by simultaneous horizontal and vertical vibrations and/or shear @xcite .
this is significant since most industrial devices ( e.g. seives , grinders and vibratory conveyors ) generate both vertical and horizontal excitations .
understanding how the fundamental physical mechanisms of granular flow relate to practical applications therefore depends on the detailed examination of systems , such as the one studied here , having excitations in more than one dimension . as in our findings
, these earlier works on simultaneous vertical and horizontal vibrations also point out the significance of the phase shift between the two types of excitation as being an important control parameter .
the cylindical geometry of our system ( with the cylinder axis oriented vertically ) is also a distinguishing aspect of the present work .
a few recent works have studied granular flow in annular troughs , and have also observed tangential circulations similar to those studied here @xcite
. however , ref .
@xcite and the present study are , to our knowledge , the first works to address radial circulation in this geometry .
the shape of the flat - bottomed cylindrical pan and its motion are illustrated in fig .
[ fig : pan_model ] .
the forcing motion we impose on the pan in our simulations is based on the motion observed in the experimental apparatus used in ref .
this motion is complex and is described in detail in this section .
we assume that the pan is a rigid body .
let @xmath0 be the normal vector to the bottom of the pan , rooted at the center of the bottom surface .
the motion of the pan can then be described as a superposition of two motions : ( i ) a precession of @xmath0 about the vertical axis with angular frequency @xmath1 , and with @xmath0 tilted at a constant angle @xmath2 with respect to the vertical ; and ( ii ) , a circular motion of the center of the pan in the horizontal plane , of amplitude @xmath3 and also at frequency @xmath1 .
note that the direction of rotation of both motions is always the same , and in our simulations , chosen to be counter - clockwise . in this motion
, the time evolution of the point @xmath4 at the center of the pan bottom has only horizontal components which are non - zero , and thus can be described in the lab frame by , @xmath5 also in the lab frame , the components of the vector @xmath0 are given by , @xmath6 where @xmath7 is the phase shift between the circular motion of the pan center , and the precessional motion of @xmath0 . in order to analyze the behavior of the granular particles in the moving pan
, we also need to derive expressions describing the motion , in the lab frame , of an arbitrary point @xmath8 fixed on the pan bottom .
let @xmath9 be the cartesian coordinates of @xmath8 in a frame of reference fixed to , and lying in the plane of , the pan bottom .
the coordinates of @xmath8 in the lab frame , @xmath10 , are then given by , @xmath11 the expression for @xmath12 above is determined from the condition that the pan is a rigid body , and from the orthogonality of the vectors @xmath0 and @xmath13 , where @xmath13 is the vector from the center of the pan bottom to the point @xmath8 , in the pan frame . if we express @xmath9 in terms of a polar coordinate system @xmath14 that is also fixed in the pan frame , then @xmath15 and @xmath16 , where @xmath17 .
using the trigonometric identity @xmath18 , eq .
[ eq : xy3 ] can be rewritten as , @xmath19 note that in ref .
@xcite , @xmath3 is referred to as @xmath20 and the maximum vertical amplitude of vibration @xmath21 , where @xmath22 is the radius of the pan , is referred to as @xmath23 . with these changes of notation , the representation of the pan motion presented in eqs .
[ eq : xy1 ] , [ eq : xy2 ] and [ eq : z ] is consistent with eqs .
1 - 3 of reference @xcite .
the event driven algorithm @xcite used in our simulations has proved very effective in simulating agitated granular matter .
for example , simulations of standing wave patterns in 3d vertically oscillated granular layers @xcite show remarkable agreement with experiments .
the algorithm itself is well described elsewhere ( see e.g. refs .
@xcite ) . in short ,
it involves five steps : ( i ) the calculation of collisions times between particles and between particles and walls ; ( ii ) determining which pair of objects are next to collide by finding the minimum of the collision times ; ( iii ) evolving the system forward in time to the next - to - occur collision by changing the coordinates and velocities of all particles according to the equations of motion of free particles in a gravitational field ; ( iv ) changing the velocities of the two colliding particles ( or of a particle hitting a wall ) according to the conservation of linear and angular momentum and a specified rule for the dissipation of energy in the collision ; and ( v ) , returning to step ( i ) .
the algorithm assumes that interactions occur only through instantaneous binary collisions .
we have adapted this algorithm to our specific system by implementing the interactions between particles and moving walls of a cylindrical flat - bottomed container in a manner that is consistent with hard sphere dynamics .
as described below , we give special care to the implementation of roughness of the bottom surface of the container .
we have tested several models for the interaction between particles and a rough bottom plate and decide on the one presented below because of its simplicity combined with its efficiency . in all our simulations
the particles are monodisperse spheres of diameter @xmath24 possessing both linear and angular momentum .
otherwise , our numerical model has a fairly large number of parameters that have to be set so that the model can be validated against the experimental results of ref .
@xcite . in the following
we organize these into three groups : dissipative , system size , and forcing parameters . _ dissipative parameters : _ these parameters control the way energy is dissipated in the system .
we implement inelastic collisions between particles in the manner described in @xcite .
the collisions between particles are thus dependent on the coefficient of restitution @xmath25 , defined as the ratio of the normal components of the relative velocities of two colliding particles before and after the collision ; the rotational coefficient of restitution @xmath26 , defined as the ratio of tangential components of relative surface velocities of colliding particles before and after the collision ; and the coefficient of friction @xmath27 for the material from which the particles are made .
we set @xmath25 to be velocity dependent in the following way : @xmath28 here @xmath29 is the component of relative velocity along the line joining particle centers , @xmath30 is a characteristic velocity of the particles , @xmath31 is the acceleration due to gravity , and @xmath32 is a tunable parameter of the simulations . decreasing @xmath32 increases the energy dissipated during each collision .
the advantages of this model for the coefficient of restitution are discussed and demonstrated in refs .
we have set @xmath33 , because this value gives the best agreement in terms of particle compaction observed in similarly agitated states in experiments with nylon balls and in our simulations .
the dissipation of rotational kinetic energy due to colliding particles is controlled by @xmath26 . as validated in ref .
@xcite , the dependence of @xmath26 on the relative particle velocities and on @xmath27 depends on the value of the threshold parameter @xmath34 that controls the choice between a rolling or sliding solution in a given particle - particle collisions .
@xcite suggests that @xmath35 is a reasonable choice for plastic sphere collisions based on comparison with experiments .
we use this value in all our simulations .
we have set @xmath36 in all our simulations based on the findings of ref .
@xcite .
the dissipation of energy in collisions between particles and the side walls and the bottom of the container is determined by the same set of parameters ( @xmath32 , @xmath27 , and @xmath34 ) as in particle - particle collisions . for the case of collisions between particles and the side walls
we have observed , however , that the simulated system s behavior is not detectably changed when using @xmath37 ( elastic collisions ) instead of @xmath38 ( inelastic collisions ) .
thus , in all our simulations we set @xmath37 for collisions between particles and the side walls .
special attention is given to the implementation of momentum transfer and dissipation of energy in collisions between particles and the rough bottom of the container .
instead of using three dissipative parameters as in particle - particle collisions , we implement a one - parameter model as follows . in the reference frame of the moving pan
, the particle strikes the bottom plate with a relative incident velocity @xmath39 at the contact point where the surface normal is @xmath40 .
the particle is reflected from the surface with relative velocity @xmath41 $ ] where @xmath42 has the same form as eq .
[ eq : restitution ] , except that it is a function of the absolute value of the incident velocity and not just the normal component .
thus , if not for the factor of @xmath42 , the particle - bottom collision would be elastic .
the decrease of all three components of the reflected velocity by the factor @xmath42 is essential here for modeling the entrainment of particles by the rough bottom surface . indeed , the effect of this decrease is that the particle s velocity ( in the lab frame ) after the collision tends to be more similar to the velocity of the rough bottom surface at the point of contact , than before the collision .
this , in turn , makes the choice of the reflection direction of little importance .
statistically the average direction of reflection must be in the plane of the vectors @xmath39 and @xmath40 , as it is in our model .
thus the only parameter of our model for the rough bottom surface is @xmath32 in eq .
[ eq : restitution ] .
we have tested several values of this parameter in the range between @xmath43 and @xmath44 .
lower values of @xmath32 provide better entrainment of particles by the bottom surface , but at the same time make the average collision time close to the surface so small that roundoff errors render the simulations unphysical .
thus , we have adopted a value of @xmath45 as a compromise that realizes particle entrainment at the bottom surface as well as well - behaved numerical simulations .
_ system size parameters : _ these parameters include the particle diameter @xmath24 , number of particles @xmath46 and the radius of pan @xmath22 .
the layer depth @xmath47 can be estimated assuming the volume fraction @xmath48 is equal to the experimentally determined @xcite value of @xmath49 .
for example , in the laboratory experiments of ref .
@xcite using nylon balls , a 30 kg load of particles ( each weighing 0.9 g and 11.5 mm in diameter ) in a pan of radius 0.381 m gave @xmath50 for the layer depth .
in the present simulations we set the number of particles to either @xmath51 or @xmath52 . in the first case we have @xmath53 and @xmath54 , and in the second case we have @xmath55 and also @xmath54 .
although the layer depth in our simulations is smaller than in the experiments , as we will show below , we are able to reproduce the main features of the dynamical behavior observed in experiments . in this work
we have chosen to present the results for only the smaller system size ( @xmath51 ) because the shorter computing times allow for larger statistical sampling .
our tests of the system behavior using the larger number of particles indicates that the results are qualitatively the same as for the smaller system . _ forcing parameters : _ the frequency of vibrations @xmath56 , the phase shift @xmath7 , the amplitude of horizontal vibrations @xmath3 , and the tilt angle @xmath2 , were introduced in section [ sec : pan_model ] .
for all results presented here we consider the set @xmath57 as the independent variables and we study the system behavior as a function of this set , with all other parameters fixed .
table [ tab : parameters ] shows the values ( or range of values ) of the forcing and system size parameters studied here , compared to those examined in the experiments of ref .
@xcite .
as stated above , we conduct simulations on a system that is smaller than that of the experiments in order to allow us to study as wide a range of dynamical states as possible .
since the pan radius @xmath22 in our simulations is therefore smaller than in the experiments , we have set the value of the tilt angle @xmath2 to a larger value , such that the maximum amplitude of vertical motion at the edge of the pan @xmath58 is approximately the same as in the experiments .
we also find that we must set the value of @xmath3 , the amplitude of the horizontal motion , to a larger value ( compared to experiments ) in order to achieve a robust entrainment of the particles near the rough bottom surface .
this suggests that the model of surface roughness used in our simulations is less effective in entraining particle motion than the real system .
.comparison of simulation parameters used in the present work , and experimental parameters from ref .
[ tab : parameters ] [ cols="^,^,^,^,^,^,^",options="header " , ]
velocity field for toroidal motion at @xmath59 hz and @xmath60 .
top panel : the velocities of particles in the layer nearest the bottom surface , viewed from above .
each arrow length is proportional to the magnitude of the average velocity corresponding to each grid location .
note that the pan center always rotates in the counter - clockwise direction in our simulations , but that under the conditions of the state shown here , the tangential flow is clockwise .
bottom panel : the velocity field of particles of the same system in a vertical cross section passing through the pan center . here
, all arrow lengths are fixed to a constant , and do not represent the magnitude of the average velocities . @xmath61 and @xmath62 approximately locate the center of circulation of the radial flow . ]
using the set of parameters described in the previous section , we study the behavior of the system as a function of @xmath63 and @xmath7 .
we conduct simulations for 15 different frequencies , in the range between 5 and 55 hz .
for each @xmath63 we set the phase angle @xmath7 to at least 20 different values in the range @xmath64 .
we thereby analyze a total of more then 300 state points . in the range of @xmath63 considered here ,
the motion settles into a steady state in less than 1.0 s of physical simulation time , which corresponds to approximately 10 - 50 revolutions of the pan , depending on @xmath63 . to ensure equilibration
, we therefore run each simulation for at least 2 s before starting to accumulate averages . from a visual examination of animations of the steady - state behavior
, we find that at fixed @xmath7 , three main regimes of dynamical behavior occur as @xmath63 increases : ( i ) at small @xmath63 , we observe a solid - like heap state that undergoes a collective tangential rotation about the vertical axis .
this is qualitatively the same behavior observed at small @xmath63 in ref .
( ii ) at larger @xmath63 , typically in the range of 15 - 20 hz , a transition occurs to a `` toroidal '' motion , having simultaneous tangential and radial circulations , which is also very similar to that observed experimentally .
( iii ) at large @xmath63 , the toroidal state crosses over to a gas - like state in which the particles have a sufficiently large kinetic energy that gravity can not maintain the particles in a dense bed at the bottom of the pan ; in this regime , the particles spread out over the entire system volume .
this crossover occurs in the range of @xmath63 between 30 and 48 hz , depending on @xmath7 .
the simultaneous tangential and radial circulations occurring in the toroidal regime are shown in fig .
[ fig : vfield_01 ] . the top panel of fig .
[ fig : vfield_01 ] shows the velocity field for the particles in a horizontal layer that is adjacent to the pan bottom .
the bottom panel shows the velocity field for particles in a vertical plane passing through the center of the pan . in these vector field plots we display the velocity vectors within the corresponding volume of space averaged over a local volume of @xmath65 and over 400 snapshots of the system taken 0.001 s apart .
therefore , the total averaging time ( 0.4 s ) is much longer than the period of oscillation of the pan in the range of @xmath63 studied here .
@xmath66 ( triangles ) and @xmath67 ( circles ) as a function of @xmath7 at various @xmath63 . to facilitate comparison , @xmath66 values
are multiplied by 30 , and @xmath67 values by 1.5.,width=249 ] phase diagram of dynamical modes found in simulations .
the solid - like heap state occurs at low @xmath63 , in the diagonally striped areas .
the domain of the gas - like state at high @xmath63 is denoted by the square lattice pattern .
the remaining white areas , bounded by triangles , indicate the domains in which the toroidal circulation is observed .
the label `` inward '' indicates the region where the radial flow is such that particles travel toward the center along the top surface of the bed .
the label `` outward '' indicates the region where the radial flow is such that particles travel away from the center along the top surface of the bed . for both the heap and toroidal states ,
the line that connects the open circles indicates a change of direction of the tangential flow .
left ( right ) arrows indicate that the direction of the tangential flow is against ( with ) the direction of the pan rotation .
the dashed rectangle shows the approximate region corresponding to the experiments of ref . .
error bars for the boundaries of the toroidal domains ( triangles ) are of the size of the symbol for low @xmath63 and increase to up to three times the size of the symbol for high @xmath63 . ]
velocity field in a vertical cross section passing through the pan center , illustrating _ outward _ radial flow , occurring at @xmath68 hz and @xmath69 . ]
we next seek to create a dynamical phase diagram of the system , i.e. to subdivide the plane of @xmath63 and @xmath7 into domains according to where each of the observed dynamical modes occurs .
to do so , we quantify the state of circulation in the granular bed using two measures of the average velocity of the particles .
the first measure , @xmath67 , quantifies the tangential circulation and is the average angular velocity of particles in the horizontal plane around the vertical line of symmetry : @xmath70 here , @xmath71 is the tangential component of @xmath72 , the velocity of particle @xmath73 , about the vertical axis .
@xmath74 is the perpendicular distance from a particle @xmath73 to the @xmath75-axis of the lab frame ; @xmath76 is the unit vector pointing along the @xmath75-axis of the lab frame ; and @xmath77 is the radial unit vector at the position of particle @xmath73 in the lab frame .
the second measure quantifies the radial circulation .
we define this measure , @xmath66 , as the average angular velocity of particles about the curved axis that forms a horizontal circle of radius @xmath61 and elevated above the bottom of the pan at a distance @xmath62 . as illustrated in the lower panel of fig .
[ fig : vfield_01 ] , if we choose @xmath78 and @xmath79 , we find that this curved axis approximately coincides with the center of the radial circulation observed in our simulations .
@xmath66 is then given by , @xmath80 here @xmath81 is the tangential component of @xmath72 , in a vertical plane , with respect the circulation center defined by @xmath61 and @xmath62 ; and @xmath82 .
@xmath83 and @xmath84 are respectively the direction and magnitude of the vector @xmath85 which defines the position of particle @xmath73 with respect to the center of circulation in the vertical plane
. examples of the behavior of @xmath67 and @xmath66 , as a function of @xmath7 at various fixed @xmath63 , are shown in fig .
[ fig : plot_w ] . in fig .
[ fig : plot_w](a ) , we see that @xmath86 for @xmath87 , becomes non - zero for higher @xmath7 , but then returns to zero as @xmath7 approaches @xmath88 . based on our visual observations of the computer animations , the heap state exhibits no radial circulation , and so the progression of behavior of @xmath66 in fig .
[ fig : plot_w](a ) reflects the transition first from the heap to the toroidal mode , and then back again , as a function of @xmath7 . on the other hand
, we see that @xmath67 varies smoothly as a function of @xmath7 in fig .
[ fig : plot_w](a ) , reflecting the fact that tangential circulation is common to both the heap and toroidal modes . based on these observations , for the purpose of constructing a phase diagram
, we define the toroidal state as that in which @xmath66 is clearly different from zero .
we also note that the variation of @xmath67 and @xmath66 with @xmath7 is approximately sinusoidal , with the behavior of @xmath66 lagging that of @xmath67 by about @xmath89 .
we will return to this behavior in the next section . the resulting phase diagram is shown in fig .
[ fig : phase_d_n2.175k ] . the most notable behavior of the observations summarized in figs .
[ fig : plot_w ] and [ fig : phase_d_n2.175k ] is that we find that both the radial and tangential circulations reverse their direction as a function ( primarily ) of @xmath7 , as indicated by the occurrence of both positive and negative values of @xmath67 and @xmath66 in fig .
[ fig : plot_w ] .
the result is a rich and highly controllable spectrum of dynamical behavior : at low @xmath63 , we see heap states that rotate either clockwise or counter - clockwise ( viewed from above ) ; at higher @xmath63 , toroidal states with any combination of clockwise or counter - clockwise tangential circulation , and inward or outward radial circulation ( based on the flow direction on the upper free surface ) can be realized .
the velocity field of a system with an outward radial circulation is shown in fig .
[ fig : vfield_02 ] .
in addition , the magnitudes of both @xmath67 and @xmath66 can be readily tuned through their dependence on @xmath63 and @xmath7 . the phase behavior shown in fig .
[ fig : phase_d_n2.175k ] is in quantitative agreement with the experimental results of ref .
@xcite for the boundary between the heap and toriodal modes .
the forcing conditions studied in ref .
@xcite correspond approximately to @xmath90 and @xmath91 , in the notation of this work . in this range ,
the tangential circulation is ( almost always ) in the same direction for both the heap and toroidal modes , and , as in experiments , is opposite to the direction the pan center rotates .
the exception , in both simulations and experiments , occurs for small values of both @xmath63 and @xmath7 , where the tangential circulation occurs in the same direction as the pan center rotates .
since the experiments only studied a small fraction of the phase diagram explored here , the reversal of the radial circulation was not observed in experiments , but is revealed here , along with the complete demarcation of the reversal boundary for the tangential circulation .
we note that the more subtle modifications of the toroidal mode observed experimentally ( the `` surface waves '' and `` sectors '' described in ref .
@xcite ) are not observed here , suggesting that these phenomena depend on aspects of the motion and/or particle interactions that are not modeled in the present simulations .
one possibility is that these modifications of the toroidal mode are due to variations of the pan motion from a perfect sinusoid .
as indicated in ref .
@xcite , the experimental apparatus was not perfectly symmetrical , and the measured motion indicated the presence of a weak pattern of beats superimposed on the principal motion . while it is possible that the experimentally observed surface waves and sectors had there origin in these additional components of the motion , more research , beyond the scope of that described here , would be required to clarify this issue .
in this section , we discuss some of the physical mechanisms that may underlie the behavior noted so far .
we focus in turn on two questions : ( i ) why do the horizontal and radial circulations reverse their direction as a function of @xmath7 ?
( ii ) what parameters determine the onset of heap - to - toroidal flow ? interrelationship of the zone of contact between the granular bed and the pan bottom , and the motion of the pan , as viewed from above , for @xmath92 hz , @xmath93 .
the contours are lines along which the average momentum change of particles colliding with the bottom surface is constant .
the vector labeled @xmath0 is the projection of the pan normal vector into the horizontal plane , and indicates the direction along which the vertical displacement of the pan is most negative .
the vector labelled @xmath94 therefore indicates the direction along which the vertical displacement of the pan is most positive .
since the pan center orbits counter - clockwise in the horizontal plane , the half circle above and to the left of the bisector formed by @xmath0 and @xmath94 is that half of the bottom surface that is rising .
the zone of contact , located in the region around the center of the momentum - change contours , is located on the rising side of the pan bottom , and has its center near the direction @xmath95 . the horizontal `` pushing velocity '' of the bottom pan surface , @xmath96 , and its polar component vectors ,
are also shown . ] in order to clarify the nature of the interaction between the granular bed and the moving pan , we first examine the nature of the contact between the two as a function of time . to determine where and when the bed is in contact with the pan as a function of time
, we subdivide the pan bottom into a grid of squares of area @xmath97 . over a time interval of 0.05 of the period of revolution of the pan
, we then find the average change of momentum of particles colliding with the pan bottom in each grid square .
this provides a measure of the instantaneous pressure exerted by the bed on each area element of the pan bottom .
when the system is in a dynamical steady state , we can average over `` snapshots '' of the pressure distribution on the pan bottom at different times by rotating each snapshot around the pan center such that the direction given by @xmath98 always points along the positive @xmath99-axis .
pressure contours resulting from this averaging process over a total time of 300 revolutions of the pan are shown in fig .
[ fig : pan - model - top ] for the case @xmath92 hz , @xmath93 .
the pressure contours show that a well - defined `` zone of contact '' exists between the granular bed and the pan bottom at a particular angular position .
similar results are obtained at other values of @xmath63 and @xmath7 , and differ only in the angular position of the zone of contact .
we find that @xmath100 , the angular position of the center of the zone of contact is , in all cases examined , close to the value , @xmath101 a relatively simple rationale supports this expression : we expect that the contact between the bed and the pan occurs on that side of the pan which is rising from its maximum negative displacement in the @xmath75 direction to its maximum positive displacement . as shown in fig .
[ fig : pan - model - top ] , when the normal vector of the pan @xmath102 is precessing counter - clockwise , the `` rising '' half of the pan bottom is identified by those values of @xmath103 such that , @xmath104 we see from fig .
[ fig : pan - model - top ] that the maximum pressure of the zone of contact is approximately in the middle of this half of the pan bottom , the angular position of which is given approximately by eq .
[ eq : theta_c ] .
next we note that we have modeled the pan bottom as a rough surface that entrains the motion of particles that collide with it ; i.e. particles tend to have a post - collision velocity with a direction similar to that of the surface element of the pan bottom that they strike . since the contact between the granular bed and the pan bottom is localized to the zone of contact , we expect that the velocity of the surface elements at the angular position given by eq .
[ eq : theta_c ] will have a significant influence on the direction of flow of the bed .
let us define the `` pushing velocity '' @xmath96 as the component of the pan bottom velocity in the horizontal plane at @xmath105 .
the time derivatives of eqs .
[ eq : xy1 ] and [ eq : xy2 ] give us the @xmath99 and @xmath106 components of @xmath96 : @xmath107 and @xmath108 .
converting these cartesian components to those in a polar coordinate system gives the radial and tangential components of @xmath96 .
the desired radial unit vector is @xmath109 and the tangential unit vector is @xmath110 $ ] .
the radial and tangential components of @xmath111 are therefore given by , @xmath112 comparing these expressions with the dependence of @xmath67 and @xmath66 on @xmath7 shown in fig . [ fig : plot_w ] shows that there is a good correspondence between the polar components of @xmath111 and the observed angular velocity of both the radial and tangential flows .
@xmath67 shows an especially good match to a function proportional to @xmath113 .
the correspondence of @xmath66 to @xmath114 is complicated by the fact that @xmath86 in the heap state .
however the observed trend for @xmath66 to lag @xmath67 by @xmath89 is consistent with the implications of eqs .
[ eq : theory_overs ] and [ eq : theory_laps ] .
overall , we therefore find that the assumption that the particles are entrained by the pan bottom in a specific zone of contact explains well the observed circulation .
if the polar components of @xmath96 drive the motion observed via @xmath67 and @xmath66 , then eqs .
[ eq : theory_overs ] and [ eq : theory_laps ] predict the occurrence of both clockwise and counter - clockwise tangential circulations , as well as both inward and outward radial flows .
in particular , eqs .
[ eq : theory_overs ] and [ eq : theory_laps ] also provide a basis for understanding why the tangential flow can be both the same as , or opposite to , the direction of rotation of the pan center .
most importantly , the dominant role of the phase shift @xmath7 in controlling the direction of the observed flows is confirmed .
model phase diagram based on eqs .
[ eq : theory_overs ] , [ eq : theory_laps ] and [ eq : wc ] .
solid lines indicate the onset of radial flow .
dotted lines indicate the change of direction of the tangential flow .
arrows and region labels have the same meaning as in fig . .
] since both the heap and toroidal circulations exhibit tangential flow , understanding the onset of the toroidal mode requires us to understand the conditions required for the appearance of the radial circulation .
this is a complex question , since the heap - to - toroidal boundary in fig .
[ fig : phase_d_n2.175k ] is clearly dependent both on @xmath7 and @xmath63 . to make a preliminary attempt to understand the dependence of the heap - to - toroidal boundary on @xmath7 and @xmath63 , let us consider @xmath115 , the maximum radial acceleration experienced by the particle bed , which drives the radial circulation observed in the toroidal mode . by analogy with many other systems exhibiting granular circulation
, we expect that the onset of the toroidal mode occurs when @xmath115 exceeds a critical threshold large enough to sustain the radial circulation .
to estimate the scaling of @xmath115 with @xmath7 and @xmath63 , we first assume that @xmath116 , where @xmath117 is the average displacement of the particle bed in the radial direction as it is pushed by the rough pan bottom . given the correspondence between the components of the pushing velocity @xmath96 and the circulation speeds @xmath67 and @xmath66 established in the previous section , we propose that @xmath117 is proportional to the product of @xmath118 and @xmath119 , the time interval during which particles are exposed to the zone of contact .
we also assume that @xmath119 is simply some fraction of the period of the pan motion , and so is proportional to @xmath120 . combining these assumptions with the expression for @xmath121 in eq .
[ eq : theory_overs ] , we therefore can write , @xmath122 if the onset of the heap - to - toroidal transition occurs at a critical value of @xmath123 then the corresponding critical angular frequency @xmath124 for the onset of radial circulation ( and therefore toroidal flow ) is given by , @xmath125 fig . [
fig : phd_theory ] shows a plot of the critical frequency @xmath126 in the @xmath63-@xmath7 plane , based on the proportionality expressed in eq .
[ eq : wc ] , with the constant of proportionality chosen so that the mimimum onset frequency is @xmath127 hz , as found in our simulations . comparing figs .
[ fig : phase_d_n2.175k ] and [ fig : phd_theory ] shows that the model expressed in eq .
[ eq : wc ] captures several qualitative features of the observed variation of the heap - to - toroidal transition : e.g. the occurrence of maxima in the transition boundary as a function of @xmath7 ; and the separation of the toroidal regime into two distinct domains , one exhibiting an inward radial flow , and one outward .
clearly the approximations of this simple model should be refined if it is to have any quantitative predictive value .
however , it does support the view that entrainment of the particle bed by the rough bottom is a key factor in determining the overall behavior of the system .
three types of boundary condition occurring in this system : ( a ) the periodic boundary condition involving the tangential flow ; ( b ) the vertical wall obstacle encountered by the inward radial flow ; ( c ) central `` plume '' obstacle encountered by the outward radial flow . ] in summary , we have established the following picture for the dynamical behavior observed in our simulations , and to a large degree , in the experiments of ref .
@xcite : the granular bed is excited by a rough horizontal surface that is undergoing both vertical and horizontal vibrations .
the vertical vibrations impose a periodicity on the contact between the bed and the surface , and the horizontal vibrations push the granular bed only during the interval of contact .
the phase shift relating the vertical and the horizontal components of the vibration determines the direction and strength of the horizontal pushing force applied by the rough surface on the bed .
the result is a highly tunable spectrum of behavior , in which both tangential and radial circulations , in arbitrary combinations , can be generated .
the cylindrical geometry studied here plays an important underlying role in the granular flows we observe .
the tangential circulation is in many ways similar to the kind of linear flow observed in vibratory conveyors @xcite , but wrapped around into a circle , creating in effect a periodic boundary condition .
indeed , a recent work studied the flow of granular matter in an annular vibratory conveyor and , as in our work , noted both the occurrence of a reversing tangential flow , and the controlling influence on this flow of the phase shift between vertical and horizontal vibrations @xcite .
the fact that the granular bed in our system encounters no obstacles as it moves about the center accounts for the ease with which the tangential flow occurs not only in the toroidal mode , but also the heap mode at very small @xmath63 .
in contrast , the granular flow in the radial direction encounters a vertical barrier that it must surmount : either the outer wall in the case of the inward radial flow , or the central `` plume '' that is created when the granular particles are pushed toward the center in the outward radial flow ( see fig . [ fig : boundary3 ] ) . here , the mechanism that drives the flow along the bottom surface is also similar to that of a vibratory conveyor , but with the complication that this driving force must be large enough to overcome the gravitational barrier imposed at the vertical boundary .
this largely explains why the radial flow only occurs above a critical value of @xmath63 .
our results also demonstrate that complex dynamical modes observed in real ( i.e. industrial ) granular matter devices can , on the one hand , be understood using computer simulations ; and on the other hand , the behavior observed in these real devices can reveal important fundamental principles ( such as the role of the phase factor established here ) relating to the creation and control of highly specialized granular flows .
we acknowledge useful discussions with k. bevan , s. fohanno , p. sistla and e.b .
we thank materials and manufacturing ontario and nserc ( canada ) for financial support ; and sharcnet for computing resources .
php acknowledges the support of the crc program .
jaeger , s.r .
nagel , and r p. behringer , rev .
phys . * 68 * , 1259 ( 1996 ) .
h. m. jaeger , s.r .
nagel , and r.p .
behringer , physics today * 49 * , 32 ( 1996 ) .
g . de gennes , physica a * 261 * , 267 ( 1998 ) .
s.g.k . tennakoon and r.p .
behringer , phys .
* 81 * , 794 ( 1998 ) .
, , , , * * , ( ) . | granular flows due to simultaneous vertical and horizontal excitations of a flat - bottomed cylindrical pan are investigated using event - driven molecular dynamics simulations . in agreement with recent experimental results ,
we observe a transition from a solid - like state , to a fluidized state in which circulatory flow occurs simultaneously in the radial and tangential directions . by going beyond the range of conditions explored experimentally , we find that each of these circulations reverse their direction as a function of the control parameters of the motion .
we numerically evaluate the dynamical phase diagram for this system and show , using a simple model , that the solid - fluid transition can be understood in terms of a critical value of the radial acceleration of the pan bottom ; and that the circulation reversals are controlled by the phase shift relating the horizontal and vertical components of the vibrations .
we also discuss the crucial role played by the geometry of the boundary conditions , and point out a relationship of the circulation observed here and the flows generated in vibratory conveyors . |
in @xcite , v.i . arnold introduced a theorem stating that the family of graphs of solutions of any linear second - order differential equation ( lsode ) , with arbitrary time dependent coefficients , is locally diffeomorphic to the family of graphs of solutions of the simplest one - dimensional equation of motion , i.e. , the equation of motion of a free particle .
this important result allows to establish a direct relationship between the symmetries of an arbitrary classical linear system , whose equation of motion is a lsode , with that of the free particle , which is known to have the largest possible symmetry group , @xmath0 ) .
the diffeomorphism was explicitly given in @xcite , namely the classical arnold transformation ( cat ) , and can be seen as a particular class of more general lie transformations relating systems defined by second order differential equations with the free particle @xcite .
in @xcite , the authors , in collaboration with v. aldaya and f. cosso , introduced the quantum version of the cat , the quantum arnold transformation ( henceforth qat ) , as a unitary map that relates the hilbert space of solutions of the time - dependent schrdinger equation for a generalized caldirola - kanai oscillator ( the quantum version of a classical system whose equation of motion is a lsode ) into the corresponding one for the free particle . in the present paper
we revise the qat and extend it in two respects .
first , we give the explicit transformation connecting quantum physical systems with arbitrary time - dependent quadratic hamiltonians with the free particle system , here called the gauged qat .
second and more importantly , we describe the transformation that links any lsode - system to any other lsode - system in terms of cat s , giving rise to a generalization of the ermakov - pinney equation , and for that reason we shall call it the arnold - ermakov - pinney transformation , providing also its quantum version .
although the qat is very recent , the cat is more than thirty years old and , since it was introduced , many authors have used it ( even without realizing it ) and have built something similar to the qat . even before that , lewis and riesenfeld ( 1969 ) @xcite introduced a technique to obtain solutions of the time - dependent schr " odinger equation ( tdse ) for a time - dependent quadratic hamiltonian ( tdqh ) as eigenfunctions of quadratic invariants . for that purpose
they wrote the solutions in terms of auxiliary variables that satisfy the classical equations of motion ( something that resembles the cat ) .
dodonov & manko ( 1979 ) @xcite constructed invariant operators for the damped harmonic oscillator and introduced coherent states , using a method similar to that of lewis and riesenfeld .
jackiw ( 1980 ) @xcite gave ( implicitly ) the quantum transformation from the harmonic oscillator ( even with a @xmath1 term ) to the free particle when studying the symmetries of the magnetic monopole .
junker & inomata ( 1985 ) @xcite gave the transformation of the propagator , in a path integral approach , for a tdqh , into the free one ( the equivalent of the qat , but in terms of propagators ) .
takagi ( 1990 ) @xcite gave the quantum transformation from the harmonic oscillator to the free particle , interpreted as the change to comoving coordinates .
bluman & shtelen ( 1996 ) @xcite gave the ( non - local ) transformation of the tdse for a tdqh plus a non - linear term into the free particle one , in the context of transformations of pdes .
kagan et al .
@xcite and independently castin & dum @xcite ( 1996 ) introduced a scaling transformation in the gross - pitaevskii equation describing bose - einstein condensates ( bec ) which is related to the qat .
suslov et al .
( 2010 ) @xcite computed the propagator for a tdqh using the classical equations . to the best of the authors knowledge
, the quantum arnold transformation can be not only very useful to perform certain analytic calculations , but key to unify different concepts scattered through the mentioned literature in a simplified framework .
the paper is organized as follows .
first , in section [ sec : cat ] we revise the classical arnold transformation . after that we go on with the quantum arnold transformation in section [ sec : qat ] .
finally two extensions are described in section [ sec : ext ] : the gauged qat and the arnold - ermakov - pinney transformation .
to set up the framework in which the classical arnold transformation acts , let us briefly remind the subject of lie symmetries of ordinary differential equations .
a lie symmetry of an ordinary differential equation ( ode ) is a coordinate transformation that sends solutions into solutions .
the problem of determining the lie symmetries of an ode is rather old , and s. lie gave the main results at the end of the nineteenth century @xcite .
one of these results was that a second order differential equation ( sode ) @xmath2 has the maximal number of lie symmetries ( @xmath3 ) if it can be transformed to the free equation by a point transformation : @xmath4 this linearization is possible if the ode is of the form : @xmath5 with @xmath6 satisfying some _ integrability conditions _
( see , for instance , @xcite ) .
there is a nice geometric interpretation of this condition in terms of projective geometry .
the non - linear sode ( [ edo ] ) is obtained by projection from the geodesic equations in a 2-dim riemannian manifold .
the coefficients @xmath6 are in one - to - one correspondence with thomas projective parameters @xmath7 , and the integrability conditions that they satisfy are the conditions for the riemann tensor to be zero ( see @xcite ) .
arnold named this process _
rectification _ or _ straightening _ of the trajectories , and studied the case of linear sode ( lsode ) , giving explicitly the point transformation for this case @xcite . in the next subsection we describe it in detail
. a general linear second order differential equation ( lsode ) is given by the differential equation : @xmath8 where @xmath9 and @xmath10 are functions of @xmath11 .
the classical arnold transformation ( cat ) is a point transformation that is a local diffeomorphism : @xmath12 where @xmath13 and @xmath14 are , in general , open intervals containing @xmath15 and @xmath16 , respectively , @xmath17 and @xmath18 are independent solutions of the homogeneous lsode satisfying the canonicity conditions : @xmath19 @xmath20 is a particular solution of the inhomogeneous lsode satisfying @xmath21 , and @xmath22 is the wronskian of the two solutions . here @xmath15 is an arbitrary time , conveniently chosen to be @xmath23 ( see @xcite for details ) .
the cat transforms the original lsode ( [ lsode ] ) into that of the free particle , up to a factor : @xmath24 the presence of this factor implies that patches of trajectories of ( [ lsode ] ) are transformed into patches of straight ( free ) trajectories . in fact , an arbitrary trajectory solution of ( [ lsode ] ) can be written as @xmath25 , and the cat sends it to @xmath26 .
while @xmath11 varies in the interval @xmath13 defined by two consecutive zeros of @xmath27 ( containing @xmath15 ) , @xmath28 varies in the range of the map defined by @xmath29 . in the case
in which @xmath27 has one zero , @xmath13 is ( left- or right- ) unbounded , and , if it has no zeros , @xmath13 is @xmath30 .
even though the cat is a local diffeomorphism , it can be defined for an arbitrary time @xmath15 .
thus different cats can be defined for different times @xmath15 and cover in this way a complete trajectory of ( [ lsode ] ) .
we shall show with the example of the harmonic oscillator how this can be done .
the harmonic oscillator ( ho ) is the best example to understand how the cat works . for this case , and considering @xmath31 , the two solutions are @xmath32 and @xmath33 .
the open interval @xmath13 defined by two consecutive zeros of @xmath27 , and containing @xmath23 , is @xmath34 , and the cat @xmath35 and its inverse @xmath36 are then written as : @xmath37 @xmath38 in this case @xmath39 .
pictorially , the cat for the ho can be represented as in figure [ figura - cat ] , where velocities have also been included in the graphic for clarity . here
@xmath35 maps the solid part of the helix ( half a period of a harmonic oscillator trajectory ) into the whole line ( a free particle trajectory ) .
the horizontal plane represents the space of all possible initial conditions at @xmath40 .
note that both trajectories are tangent when projected onto this plane , due to the conditions ( [ canonicity ] ) .
see @xcite for more details in this case .
for the cat to map other patches of the ho trajectories into the free particle trajectories , different branches of the @xmath41 function in the inverse cat ( [ cat - ho - inv ] ) should be used ( and a different @xmath42 for the cat ) . for each integer @xmath43 ,
let us take @xmath44 and @xmath45 .
the solutions verifying conditions ( [ canonicity ] ) at @xmath46 are @xmath47 .
define a pair of cat and inverse cat from @xmath48 into @xmath49 of the form : @xmath50 and @xmath51 , where the @xmath43-th branch of the @xmath41 function has been used in @xmath52 .
an _ unfolded _ version of the cat , @xmath53 , can be built by joining all the patches @xmath54 , defining an application that maps a complete trajectory @xmath55 of the harmonic oscillator into a trajectory @xmath56 of the free particle .
@xmath53 is periodic on @xmath11 with period @xmath57 , although discontinuous .
other simple examples where this construction can be done are the damped particle and the damped harmonic oscillator , see @xcite for details .
an arbitrary lsode system ( [ lsode ] ) can be derived from the lagrangian ( we shall take @xmath31 for simplicity but the whole formalism can be developed with @xmath58 , see @xcite ) : @xmath59 and from this the hamiltonian @xmath60 is derived , which is known as the generalized caldirola - kanai ( gck ) hamiltonian for a damped oscillator ( see @xcite and references therein ) . the case in which @xmath61 and @xmath62 are
constants corresponds to the original caldirola - kanai hamiltonian for a damped harmonic oscillator @xcite , and whose corresponding lagrangian was given for the first time by bateman @xcite , after a constraint to the bateman dual system is imposed @xcite .
canonical quantization of the gck hamiltonian leads to the time - dependent schrdinger equation : @xmath63 the cat @xmath35 is a local ( in time ) diffeomorphism between the space of solutions of the lsode system ( [ lsode ] ) and the space of solutions of the free particle .
we would like to extend it to a unitary transformation @xmath64 , the quantum arnold transformation ( qat ) , between the hilbert space of solutions @xmath65 of the time - dependent schrdinger equation for the gck oscillator ( [ ckeq ] ) at time @xmath11 , @xmath66 , into the hilbert space of solutions @xmath67 of the time - dependent schrdinger equation for the galilean free particle @xmath68 at time @xmath28 , @xmath69 .
the desired extension is given by : @xmath70 here @xmath71 is the pullback of the cat @xmath35 , acting on functions ( i.e. @xmath72 ) . the qat can be diagrammatically represented as : @xmath73 where @xmath74 is the common hilbert space of solutions of the schrdinger equation for both systems at @xmath75 ( we shall take , for simplicity , @xmath23 , as before ) , @xmath76 is the unitary time - evolution operator for the gck oscillator and @xmath77 is the corresponding one for the galilean free particle .
the map at the bottom of the diagram is the identity due to conditions ( [ canonicity ] ) ; otherwise a non - trivial unitary transformation appears ( see @xcite ) . from the commutative diagram ,
it is clear that @xmath78 is unitary given the unitarity of the evolution operators .
however , it can also be checked explicitly that the scalar product of two states in @xmath79 at a given time @xmath28 is the same than that of the transformed states by @xmath78 in @xmath80 at the corresponding time @xmath11 : @xmath81 where @xmath28 , @xmath82 , @xmath83 and the integration limits have been transformed according to the cat .
the qat inherits from the cat the local character in time , in the sense that it is valid only for @xmath84 and @xmath85 , although it can be defined for an arbitrary initial time @xmath15 . to extend the qat beyond @xmath13 , we can proceed as in the classical case for the harmonic oscillator , considering the different branches of the inverse function of @xmath86 , defining an _ unfolded _
qat , @xmath87
. it should be stressed that if in the different branches of the _ unfolded _ cat proper solutions verifying ( [ canonicity ] ) are not used , changes in signs can appear , which result in changes in phases in the different branches of the _ unfolded _ qat .
this phenomenon is related to the maslov correction ( see for instance @xcite ) .
in fact , it can be checked that , for the case of the harmonic oscillator previously considered , @xmath88 . by the results of sec .
[ sec : lie ] , any classical lsode system has the maximal number of lie symmetries , the ones of the free particle ( @xmath3 ) .
the vector fields generating these symmetries can be computed by means of the cat from the symmetries of the free particle . even though the cat is a local diffeomorphism ,
these vector fields are defined for all times @xcite ( the reason for this is that the only function appearing in the denominator of the vector fields is the wronskian , which is never vanishing ) .
the quantum symmetries of the free schrdinger equation ( [ free - eq ] ) are smaller than @xmath3 , only the schrdinger group preserves ( [ free - eq ] ) @xcite .
the schrdinger group is a semidirect product of the ( centrally extended ) galilei group by the group @xmath89 .
it is generated by the basic operators : @xmath90 in addition to their quadratic combinations ( representing scale and non - relativistic _ conformal _ transformations ) .
the basic operators , together with their quadratic combinations , are constant of motion operators=0 $ ] , where @xmath91 is the hamiltonian operator for the free galilean particle .
] , they are infinitesimal generators of symmetries of the free schrdinger equation ( [ free - eq ] ) , and preserve the hilbert space @xmath79 of solutions ( [ free - eq ] ) .
we can import the symmetries acting on the free particle hilbert space @xmath92 ( the schrdinger group ) into the gck hilbert space @xmath66 by means of the commutative diagram ( [ diagrama - qat ] ) .
the infinitesimal generators of these symmetries are : @xmath93 and their quadratic combinations ( see @xcite for the complete list of operators ) .
all these operators expand a representation of the schrdinger group , are constant of motion operators for ( [ ckeq ] ) and preserve the hilbert space @xmath80 .
despite of the local character of the qat , and similarly to the classical case , these operators are well defined for all @xmath94 , since the wronskian @xmath95 never vanishes .
the hamiltonian for the gkc harmonic oscillator can be expressed in terms of the basic , conserved operators : @xmath96 where the time - dependent functions @xmath97 and @xmath98 are given by : @xmath99 again , since the wronskian never vanishes and the basic operators @xmath100 and @xmath101 are also well - defined for all times , this construction remains valid for all times .
note that , since the basic operators @xmath100 and @xmath101 are conserved , @xmath102 is not a conserved operator ( unless @xmath103 is a constant and @xmath104 ) , and therefore an eigenvalue equation for it makes no sense ( i.e. , there is no time - independent schrdinger equation ) . instead , any linear combination of @xmath105 , @xmath106 and @xmath107 can be used to define an eigenvalue problem , and the corresponding eigenvectors would constitute a basis for the solution of the generalized caldirola - kanai schrdinger equation ( see @xcite and references therein ) . from the commutative diagram ( [ diagrama - qat ] ) and from ( [ unitarityqat ] )
it is clear that the qat is a unitary operator , and this has interesting and far - reaching consequences . among them
we can mention the possibility of importing operators ( symmetries ) from one system to the other , importing wave functions , scalar product , computing the time evolution operator , etc .
we shall not discuss them here , referring the reader to @xcite for details .
the qat have many more and very interesting applications and generalizations .
one of them is the application of the qat to density matrices , transforming the quantum liouville equation of an arbitrary time - dependent quadratic hamiltonian into the free one @xcite , with the possibility of extending it and to wigner functions and other quasi - probability distributions , or even the extension of the qat to lindblad type equations , in order to study dissipation and decoherence under the qat point of view .
it can also be applied to bose - einstein condensates , to transform the time - dependent potential ( oscillator traps with time - dependent frequencies ) into a time - independent potential .
the qat can also be relevant in quantum inflationary cosmological models , where quantum fluctuations of the scalar field are governed , under some assumptions , by a lsode with a dissipative term . among the possible generalizations of the qat , we mention the relativistic case , as the quantum version of a generalized cat for geodesic equations in a fixed background and with external forces , or as the quantum version of _ geodesic mappings _ , that transform geodesics of a metric into geodesics of a different metric ( beltrami theorem ) .
other possible generalizations refer to non - linear potentials ( the quantum lie transformation ) , to second order riccati equations ( see for instance @xcite ) , or even to non - local potentials like in the gross - pitaevskii equation .
in this section we shall introduce the main results of the paper : two immediate , although non - trivial , extensions of the qat .
the first one is to allow for more general schrdinger equations than the gkc one .
the second one is the transformation relating two lsode - systems , denoted as the arnold - ermakov - pinney transformation , obtained as the composition of the cat corresponding to the first lsode - system with the inverse of the cat corresponding to the second lsode - system , as well as the corresponding quantum version .
the qat , as it has been designed , transforms the solutions of the schrdinger equation of the gkc oscillator into that of the free particle .
it can not transform , as it stands , the solutions of the schrdinger equation associated with an arbitrary time - dependent quadratic hamiltonian .
the reason is the choice of solutions @xmath108 satisfying the canonicity condition ( [ canonicity ] ) and the fact that the classical counterpart , the cat , transforms the lsode ( [ lsode ] ) into that of the free particle .
suppose that the quantum hamiltonian is an arbitrary , time - dependent self - adjoint quadratic hamiltonian : @xmath109 where @xmath110 and @xmath111 are real and @xmath112 ( to have a positive - definite kinetic energy ) .
then the transformation that maps solutions of the schrdinger equation for this hamiltonian into that of the free particle is given by : @xmath113 where @xmath114 and @xmath27 are canonical solutions ( in the sense of ( [ canonicity ] ) ) verifying the lsode ( [ lsode ] ) with @xmath115 and @xmath116 .
the wronskian of the two solutions is @xmath117 .
note that the transformation ( [ gaugedqat ] ) is an ordinary qat times an extra _ gauge _ term @xmath118 : @xmath119 for this reason we shall denote this transformation gauged qat ( gqat ) . in the case where @xmath120 , the gqat turns into an ordinary qat .
note that @xmath121 can be interpreted as a time - dependent mass , and therefore the hamiltonian can be written as @xmath122 where @xmath123 , and the _ classical _ damping coefficient and frequency are written now as @xmath124 and @xmath125 , respectively .
although the cat relates any lsode system with the free particle , the dynamics of both systems can be very different .
thus , it could be interesting to relate directly two arbitrary lsode systems with similar behaviour , and this can be achieved by composing a cat and an inverse cat .
more precisely , let @xmath126 and @xmath127 denote the cats relating the lsode - system 1 and lsode - system 2 to the free particle , respectively , then @xmath128 relates lsode - system 2 to lsode - system 1 .
@xmath129 can be written as : @xmath130 the explicit form of the transformation can be easily computed by composing the two cats , resulting in : @xmath131 where @xmath132 satisfies the non - linear sode : @xmath133\,,\ ] ] and where @xmath134 refers to the @xmath135-th particular solution for system @xmath136 ; @xmath137 , @xmath138 and @xmath139 stand for the wronskian and the lsode coefficients for system @xmath136 ; and the dot means derivation with respect to the corresponding time function .
for the particular case where lsode - system 1 is a harmonic oscillator ( @xmath140 and @xmath141 ) , this expression simplifies to : @xmath142 resulting in a generalization of the ermakov - pinney equation . for @xmath143 the ermakov - pinney equation ( also known as milne - pinney )
is recovered @xcite : @xmath144 representing a harmonic oscillator of frequency @xmath145 with an extra inverse squared potential @xmath146 . for @xmath147 ,
the arnold - ermakov - pinney transformation reduces to the ordinary cat , i.e. @xmath148 . the ermakov - pinney equation is related to the ermakov invariant @xcite and appears in many branches of physics , such as cosmology @xcite , bec @xcite , etc .
its generalization to higher dimensions , know as ermakov pairs ( or system ) , appears in bec @xcite and what is known as kepler - ermakov systems @xcite .
the ermakov - pinney equation entails a kind of nonlinear superposition principle , in the sense that its solutions can be written in terms of the solutions @xmath149 of the corresponding linear equation ( with @xmath147 ) : @xmath150 the other way round , the general solution @xmath151 of the linear equation can be writen in terms of a paricular solution @xmath152 of the ermakov - pinney equation ( [ ermakov - pinney ] ) as : @xmath153 where @xmath154 are arbitrary constantss and @xmath155 .
note that this equation is just ( [ arnold - ermakov - pinney ] ) for @xmath156 , @xmath157 , @xmath158 , @xmath159 and @xmath160 . as a result , the general solution of ( [ ermakov - pinney ] ) can be determined from a particular solution @xmath152 using ( [ pinney - oscilador ] ) and ( [ oscilador - pinney ] ) .
the quantum version of the arnold - ermakov - pinney transformation , @xmath161 , can be obtained computing the composition of a qat and an inverse qat , to give : @xmath162 the quantum arnold - ermakov - pinney transformation is a unitary map importing solutions of a gck schrdinger equation from solutions of a different , auxiliary gck schrdinger equation which , in particular , can be the one corresponding to a harmonic oscillator .
in that case the transformation is very similar to the one used in bec , known as scaling transformation to transform the time - dependent potential ( oscillator traps with time - dependent frequencies ) into a time - independent harmonic oscillator potential @xcite .
also , in that case ( i.e. for @xmath163 ) equation ( [ aep ] ) reduces to the transformation given by hartley and ray @xcite ( this was already given by lewis and riesenfeld in @xcite ) .
however , the quantum arnold - ermakov - pinney transformation allows to choose in a suitable way the auxiliary system from which the solutions may be imported .
work partially supported by the fundacin sneca , spanish micinn and junta de andaluca under projects 08814/pi/08 , fis2011 - 29813-c02 - 01 and fqm219-fqm1951 , respectively .
arnold v i , _ supplementary chapters to the theory of ordinary differential equations _
( nauka , moscow , 1978 ) ; english transl . , _ geometrical methods in the theory of ordinary differential equations _ , ( springer - verlag , new york - berlin , 1983 ) . | the previously introduced quantum arnold transformation , a unitary operator mapping the solutions of the schrdinger equation for time dependent quadratic hamiltonians into the solutions for the free particle , is revised and some interesting extensions are introduced , providing in particular a generalization of the ermakov - pinney equation . |
w43-mm1 is a large and young high mass star forming clump located within the w43 region and at 5.5 kpc from the sun @xcite .
the clump is near @xmath0 and at an interface with an extended region powered by a cluster of o type and wolf - rayet stars . @xcite .
w43-mm1 has been well studied in continuum from 1.3 mm to 70 @xmath1 m @xcite .
these studies identified a large sample of clumps in w43 , from which w43-mm1 is the most massive with an estimated mass of 2128 and a deconvolved size of 0.09 pc @xcite .
infalling motions have been detected towards w43-mm1 @xcite suggesting that the clump is undergoing gravitational collapse .
@xcite made interferometric observations of polarized dust emission with bima , finding an ordered pattern for the magnetic field and estimating an on - the - plane of the sky field strength of about 1 mg .
recent sma results from polarized dust emission and ch@xmath2cn line emission at higher angular resolution ( @xmath3 pc scales ) updated the magnetic field estimate to 6 mg also computing a mass to magnetic flux ratio of about the critical value @xcite . additionally from the ch@xmath2cn emission , evidence for an embedded hot core of @xmath4 k was found in w43-mm1 main clump . in this letter
we report the first alma observations of polarized dust emission towards w43-mm1 . here , section [ se : obs ] report the observations , section [ se : ce ] the continuum emission and source extraction , section [ se : mf ] the magnetic field properties , and section [ se : sum ] is the summary and discussion .
alma observations at 1 mm ( band 6 ) were done on may 30th , 2015 over w43-mm1 using @xmath518:47:47.0 and @xmath6 - 01:54:28.0 as phase center .
an array of 35 antennas was used reaching an angular resolution of 0.5@xmath7 ( @xmath8 0.01 pc scales ) .
the spectral configuration was set to single spectral windows per baseband in continuum mode with 64 channels giving 31.250 mhz as the spectral resolution in full polarization mode ( @xmath9 , and @xmath10 ) .
each spectral window was centered at the standard alma band 6 polarization frequencies ( 224.884 , 226.884 , 238.915 , and 240.915 ghz ) , where two successful executions were done as part of the session scheme @xcite .
calibration and imaging was done using the common astronomical software applications ( casa ) version 4.5 .
the alma antennas are equipped with receivers sensitive to linear polarization .
after the incoming radiation has gone through the feed - horn , the wave is divided into two orthogonal components ( @xmath11 and @xmath12 ) by a wave splitting device @xcite .
this operation is not perfect and there is always a residual , or projection , from one polarization onto the other which is known as the instrumental polarization , or d - terms @xcite . given that an antenna uses azimuth and elevation coordinates , the frame of the sky rotates with respect to the antenna introducing an angular dependence which is parameterized by the parallactic angle .
additionally to the d - terms , the @xmath11 and @xmath12 polarizations have different signal paths which introduces a relative delay between both polarizations .
also , the interferometric calibration scheme for amplitude and phase requires the use of a reference .
this reference breaks the degeneracy intrinsic to the array and thus ; we do not measure absolute phase values but relative ones with respect to the reference ( where phases are set to zero in both polarizations ) . by doing this ,
we introduce an additional phase bandpass between the _ xy _ and _ yx _ cross correlations . to calibrate all these quantities
, an alma polarization observation samples a strong , un - resolved , polarized source over a certain range of parallactic angle .
the polarization calibrator is sampled for 5 minutes every 35 minutes or so ( the precise time cadence is calculated by the online software at run time ) . for our observations , we obtained about 100@xmath13 of parallactic angle coverage for j1924 - 292 which was selected as polarization calibrator . using this source , we derived solutions for the cross polarization delay , the _
xy_-phase , and the d - terms .
these solutions were applied to w43-mm1 data also using j1751 + 0939 to calibrate the bandpass , j1851 + 0035 to calibrate the phase , and titan to calibrate the flux . after applying the calibration tables , we imaged the data using the @xmath14 casa task with the _ briggs _ weighting scheme , robust number 0.5 , for sidelobe robustness and the clark deconvolution algorithm to produce the stokes images .
the final images were produced after three self - calibration iterations using a final solution interval of 90 seconds .
figure [ stokesi ] presents the stokes i image from w43-mm1 .
the continuum emission shows a fragmented filament extending from the north to the south - west .
two bright sources at the center , a and b1 , completely dominate the energy budget in w43-mm1 ( with integrated fluxes of @xmath8 2.0 and 0.5 jy , which correspond to @xmath15 of the total flux recovered ) , over a number of additional fragments extending to the south - west . also , additional sources to the east and west have been detected . comparing with the sma results from @xcite and the pdbi 1 mm results from @xcite ,
the alma observations reproduce quite well the overall morphology of w43-mm1 , but with better resolution .
the noise in the alma map is @xmath16 mjy @xmath17 with a peak of 503 mjy @xmath17 obtained from gaussian fitting .
we used the _ getsources _ algorithm @xcite to successfully extract 14 sources from our alma data ( see table [ tab1 ] and figure [ stokesi ] ) .
the extraction was later compared to other methods such as _ clumpfind _
@xcite , _ fellwalker _ , and _ reinhold _
@xcite obtaining a good agreement with the selection produced by _
getsources_. we used the source positions and sizes derived using _ getsources _ as initial guess for a gaussian 2-d fitting ( using casa _ imfit _
algorithm ) in order to derived accurate fluxes from the extracted sources .
also , we kept the same source nomenclature used by @xcite , but adding numbers where higher multiplicity was discovered with respect to the sma map . using the standard procedure to calculate masses from dust emission
@xcite we computed masses for all 15 sources in our catalog assuming a dust opacity of @xmath18 @xmath19g@xmath20 @xcite , a gas to dust ratio of 1:100 , and a dust temperature of @xmath21 k @xcite , with the exception of the hot core , source a , where we used a range between @xmath22 k. although the sma detected ch@xmath2cn emission towards b1 and c , it was unresolved and not sufficient to derived temperatures ; hence , we used @xmath21 k for these sources .
@xcite modeled an sed and derived a temperature profile for source a using all the publicly available data on w43-mm1 to date .
their model suggests a temperature of @xmath8 150 k at 2500 au distance ( 0.5@xmath7 radial ; 1@xmath7 size ) and @xmath23 k at about 8000 au distance ( 0.7@xmath7 radial ; 1.4@xmath7 size ) , which are the length - scales sampled by alma .
however if larger spatial scales than our source size are considered , the temperature might be lower and in the order of 30 k ( as suggested by * ? ? ?
these scales ( @xmath24 ) are consistent with herschel primary beam at 160 @xmath1 m and , off course , larger than our deconvolved source sizes and synthesized beam . therefore and using this temperature range ,
our derived mass for source a is between 312 and 146 ( with the exception of the 30 k temperature derived mass of 728 ) .
these estimates put source a below other massive clumps such as the sdc335-mm1 mass estimate of 545 @xcite and g31.41 + 0.31 with a mass estimate of 577 @xcite)=@xmath25 @xmath26 .
however , these mass estimates where derived from observations sampling larger length scales than our alma w43 observations . the deconvolved size of sdc335-mm1 is about 0.054 which is 2 times the size of source a or 4 times the area
. a simple estimate assuming a @xmath27 density profile , will give about 272 per source a size for sdc335-mm1 and about 205 per source a size for g31.41 + 0.31 .
the polarized emission from w43-mm1 shows fractional polarization levels between 0.03% to 22% where the lowest values are seen over the peaks in stokes i , which corresponds to the well known polarization - hole ( * ? ? ? * and references therein ) . assuming perfect grain alignment , the magnetic field morphology onto the plane of the sky is inferred from the polarized emission by rotating the electric vector position angle ( evpa ) by 90@xmath13 and shown in figure [ w43_pol ] .
the field morphology shows a smooth and ordered pattern over the filament on angular scales @xmath28 , .
however , recent results from alma commissioning @xcite showed that the systematic error within -3 db level ( @xmath29 ) of the primary beam is less than 0.5% in band 6 , which is larger than the size of the w43-mm1 filament .
in fact , our observations are consistent to the sma data beyond the 1/3 of fwhm or 10@xmath7 limit . ] where ordered magnetic fields are found in massive dense cores when observed at similar spatial scales , as it has been shown from a relatively large sample by @xcite .
we found overall agreement with the sma results , but now given our higher resolution and sensitivity , we can see the field morphology in greater detail .
in fact for source a , there is a @xmath890@xmath13 change in orientation to the west of the clump , where there is also a decrease in polarized intensity , and towards d1 , the field changes about @xmath30 in orientation with respect to the main filament . for analysis
, we divided the filament into four regions according to the boundaries in the field pattern .
although , the field morphology is continuous from a to b1 , we set the boundary to the north of b2 where we plotted the lowest contour in the polarized intensity .
we did that in order to analyze the field locally to source a and sources b1 , b2 , b3 , b4 , and e. to understand the dynamical importance of the magnetic field over w43-mm1 , we estimated the strength of the field using the chandrasekhar & fermi technique ( here after cf * ? ? ?
* ) as follows from @xcite .
@xmath31 where @xmath32 is the molecular hydrogen number density calculated as an average over the region sampled by the polarized dust emission , @xmath33 is the fwhm from a line tracing the gas motions in w43 which in this case was taken to be 3.0 km / s from h@xmath34co@xmath35(4 - 3 ) and dco@xmath35(5 - 4 ) single dish emission @xcite , and @xmath36 is the evpa dispersion , which we calculated as the standard deviation of the evpas for each region .
thus , the derived field strengths are local to all the sources in a particular region .
table [ tab2 ] summarizes our polarization results with the corresponding estimations for the field .
the usage of the cf method have been debated given the small angle dispersion approximation required and the assumption of energy equipartition . from table
[ tab2 ] we see that 2 out of 4 of our regions have evpa dispersions larger than the @xmath37 limit suggested by @xcite .
given this , we included modified versions of the cf method @xcite , to calculate the field strength on each region independently .
@xcite attempt to address the limitation of the small angle approximation by replacing @xmath38 by @xmath39 which is calculated locally and by adding a geometric correction to avoid underestimating the field in the super - alfvenic case .
in contrast , @xcite assumed that the field perturbation is a global property and thus , they replaced @xmath36 by @xmath40 in the denominator of equation [ cf ] . by using the 3 versions of the cf method
, we obtained estimations of the magnetic field between 0.2 to 7 mg for our 4 regions , where for source a we obtained values between 1 and 4 mg .
@xcite plotted the most up - to - date field profile from zeeman measurements .
figures 6 and 7 in that work , suggests that our estimates for b@xmath41 are consistent with the curve if we extrapolate the profile as , unfortunately , the plot lacks field values for the @xmath42 @xmath43 range in column density .
the alma results on w43_mm1 suggest that we are seeing a highly fragmented filament where the emission is dominated by a single clump , source a. indeed , from table [ tab1 ] we see that the bulk of fragments ( 13 ) have fluxes between 32 and 200 mjy ; while source a alone is about 2 jy . the high degree of fragmentation seen in the alma data pose the question about the gravitational stability of these sources .
although we do not have high resolution line data to address the kinematics and determine if these sources are self - gravitating , the infalling motions detected by @xcite ( covering an area of @xmath44 ) are likely showing accretion from larger scales onto the whole filament and suggesting gravitational collapse . to test the gravitational stability of each fragment , we calculated the thermal jeans length as @xmath45 , where @xmath46 the sound speed , @xmath47 is the volume density , and @xmath48 is the gravitational constant .
we found that our deconvolved sizes are larger than the estimated jeans lengths by roughly an order of magnitude for most fragments .
thus , we can not conclude what fraction of these fragments are gravitationally unstable from these data alone .
interestingly is the case of source a which has been determined to be gravitationally bound ( * ? ? ?
* ; * ? ? ?
* and references therein ) but its deconvolved size is between a factor of 10 and 6 larger than its jeans length . if the jeans length suggests further fragmentation , it is possible that source a has a larger multiplicity of fragments which requires higher angular resolution to resolve . where n@xmath50 is molecular hydrogen column density in @xmath43 calculated for each source independently , b@xmath41 is the magnetic field strength in @xmath51 assumed to be unique for a given region , and factor of 3 in the denominator corresponds to an statistical geometrical correction @xcite .
this statistical geometrical factor of 3 is for a uniform density slab and will be smaller than 3 for more physically realistic cases , such as a centrally condensed disk .
we found highly super - critical values for all fragments in the filament ( see table [ tab2 ] ) , showing that the field is not strong enough to support them against gravity . also , it is worth noting that polarized emission decreases significantly towards the dust peaks in w43-mm1 , which has already been observed towards dr21(oh ) by the sma @xcite .
this is particularly evident towards source a where the polarized intensity emission goes below the 5@xmath52 level close the dust peak .
additionally , the b@xmath41 morphology in source a seems to indicate the field weakness as the lines appear to be bent towards the dust peak .
this additional evidence , suggests that gravity is pulling the field lines as the gas is infalling due to gravitational collapse .
however , what fraction of these fragments are bound by gravity is still open ; and thus , further fragmentation can not be ruled out . here , we have presented alma polarized dust emission observations of w43-mm1 .
we have found a fragmented filament threaded by a smooth , ordered , and highly detailed magnetic field .
we have derived mass estimates for 15 fragments extracted from the continuum map , showing that source a dominates the flux distribution with a 51% of the total flux .
using temperature modeling by others , we derived a mass range for source a between 146 and 312 under the length - scales sampled by alma .
however , it is not clear if further fragmentation is ongoing inside source a and only higher resolution alma observations can reveal this .
we have derived magnetic field strengths for all fragments with sufficient polarization data and found field estimations in range of 0.2 - 7 mg using 3 versions of the cf method .
derivation of the mass to magnetic flux ratio indicate that the fragments are super - critical suggesting that the field is dominated by gravity at this stage of the w43-mm1 evolution . .
overlaid are the sources extracted as ellipses , in black , representing the deconvolved sized obtained from the gaussian fits . with an rms noise of @xmath53 mjy @xmath17 ,
the dynamic range of the self - calibrated image is 1220 . ] . the magnetic field morphology is shows as pseudo - vectors at a significance of 3@xmath52 in green and 5@xmath52 in blue , where @xmath54jy @xmath17 corresponds to the noise in the polarized intensity image .
the length of each pseudo - vectors is normalized .
also , each pseudo - vector is plotted every half - beam , i.e. in steps of 8 and 4 pixels , where the beam is @xmath55 pixels . ]
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
+ p.c.c . would like to thank ed fomalont for very helpful and enlighten discussion about polarization data analysis .
+ j.m.g . acknowledges support from micinn aya2014 - 57369-c3-p and the mecd prx15/00435 grants ( spain ) .
is supported in part by nasa nnx14ab38 g and nsf ast-1313083 .
thanks the support of the ministry of science and technology ( most ) of taiwan through grants nsc 98 - 2112-m-007 - 007-my3 , nsc 101 - 2119-m-007 - 004 , and most 102 - 2119-m-007 - 004- my3 .
+ this research made use of aplpy , an open - source plotting package for python hosted at http://aplpy.github.com , j. , anderson , l. d. , battersby , c. , calzoletti , l. , digiorgio , a. m. , faustini , f. , ginsburg , a. , li , j. z. , nguyen luong , q. , molinari , s. , motte , f. , pestalozzi , m. , plume , r. , rodon , j. , schilke , p. , schlingman , w. , schneider - bontemps , n. , shirley , y. , stringfellow , g. s. , testi , l. , traficante , a. , veneziani , m. , & zavagno , a. 2010 , , 518 , l90 , d. s. , reinhold , k. , jenness , t. , & economou , f. 2007 , in astronomical society of the pacific conference series , vol .
376 , astronomical data analysis software and systems xvi , ed .
r. a. shaw , f. hill , & d. j. bell , 425 , c. l. h. , plambeck , r. l. , kwon , w. , bower , g. c. , carpenter , j. m. , crutcher , r. m. , fiege , j. d. , franzmann , e. , hakobian , n. s. , heiles , c. , houde , m. , hughes , a. m. , lamb , j. w. , looney , l. w. , marrone , d. p. , matthews , b. c. , pillai , t. , pound , m. w. , rahman , n. , sandell , g. , stephens , i. w. , tobin , j. j. , vaillancourt , j. e. , volgenau , n. h. , & wright , m. c. h. 2014 , , 213 , 13 , f. , motte , f. , hennebelle , p. , maury , a. , bonnell , i. , bontemps , s. , gusdorf , a. , hill , t. , gueth , f. , peretto , n. , duarte - cabral , a. , stephan , g. , schilke , p. , csengeri , t. , nguyen luong , q. , & lis , d. c. 2014 , , 570 , a15 , q. , qiu , k. , girart , j. m. , ( baobab liu , h. , tang , y .- w . , koch , p. m. , li , z .- y . , keto , e. , ho , p. t. p. , rao , r. , lai , s .-
, ching , t .- c . , frau , p. , chen , h .- h . ,
li , h .- b .
, padovani , m. , bontemps , s. , csengeri , t. , & jurez , c. 2014 , , 792 , 116 c c c c c c c c c c c c a & 18:47:47.0 & -01:54:26.7 & 503.1 @xmath56 26 & 1988.2 & 1.30 & 0.80 & 131.4 & 38,81.3,190 & 146,312,729 & 27.09 & 5,3.4,2.2 + b1 & 18:47:46.8 & -01:54:29.3 & 271.4 @xmath56 15 & 504.1 & 0.58 & 0.51 & 138 & 384.0 & 222 & 14.4 & 1.5 + c & 18:47:46.4 & -01:54:33.4 & 95.2 @xmath56 3 & 190.5 & 0.67 & 0.46 & 33 & 133.9 & 84 & 14.8 & 2.6 + h & 18:47:46.8 & -01:54:31.2 & 73.5 @xmath56 4 & 151.9 & 0.72 & 0.50 & 60 & 85.0 & 67 & 16.0 & 3.3 + b4 & 18:47:46.9 & -01:54:30.1 & 49.8 @xmath56 8 & 121.1 & 1.00 & 0.60 & 72 & 31.3 & 53 & 20.7 & 5.4 + g & 18:47:47.3 & -01:54:29.6 & 48.1 @xmath56 2 & 48.4 & 0.77 & 0.47 & -81 & 26.6 & 21 & 16.1 & 5.9 + b2 & 18:47:46.9 & -01:54:28.6 & 46.1 @xmath56 5 & 201.3 & 1.48 & 0.82 & 103 & 17.9 & 89 & 29.5 & 7.1 + f1 & 18:47:46.5 & -01:54:23.1 & 44.7 @xmath56 1 & 71.9 & 0.77 & 0.47 & -81 & 39.4 & 32 & 16.1 & 4.8 + b3 & 18:47:47.0 & -01:54:29.6 & 43.7 @xmath56 4 & 133.3 & 1.10 & 0.68 & 101 & 24.9 & 59 & 23.0 & 6.1 + d2 & 18:47:46.5 & -01:54:32.4 & 42.6 @xmath56 4 & 153.9 & 1.26 & 0.74 & 75 & 20.8 & 68 & 25.7 & 6.6 + e & 18:47:47.0 & -01:54:30.8 & 38.6 @xmath56 1 & 84.0 & 0.73 & 0.51 & 158 & 43.8 & 37 & 16.3 & 4.6 + d1 & 18:47:46.6 & -01:54:32.0 & 36.7 @xmath56 2 & 137.1 & 1.61 & 0.58 & 79 & 18.5 & 60 & 25.7 & 7.0 + k & 18:47:46.2 & -01:54:33.3 & 25.3 @xmath56 1 & 40.4 & 0.77 & 0.47 & -81 & 22.2 & 18 & 16.1 & 6.4 + f2 & 18:47:46.5 & -01:54:24.1 & 24.3 @xmath56 2 & 32.4 & 0.77 & 0.47 & -81 & 17.8 & 14 & 16.1 & 7.2 + j & 18:47:46.3 & -01:54:33.4 & 15.6 @xmath56 1 & 37.2 & 0.93 & 0.54 & 111 & 12.7 & 16 & 18.8 & 8.5 + c c c c c c c c c c c c c c c a & 1 & 1.4 & 68.0 - 31.7 & -30.5 & 36.2 & 0.41 & 13.9 & 4.9 & 3 & 4 & 1 & 59 - 27 & 40 - 19 & 145 - 68 + b1 & 2 & 1.5 & 170.7 & 42.5 & 21.9 & 0.46 & 19.6 & 6.2 & 5 & 8 & 2 & 89 & 55 & 225 + e & 2 & 1.5 & 22.1 & 42.5 & 21.9 & 0.46 & 19.6 & 6.2 & 5 & 8 & 2 & 12 & 7 & 29 + b2 & 2 & 1.5 & 16.3 & 42.5 & 21.9 & 0.46 & 19.6 & 6.2 & 5 & 8 & 2 & 9 & 5 & 21 + b3 & 2 & 1.5 & 17.7 & 42.5 & 21.9 & 0.46 & 19.6 & 6.2 & 5 & 8 & 2 & 9 & 6 & 23 + b4 & 2 & 1.5 & 20.0 & 42.5 & 21.9 & 0.46 & 19.6 & 6.2 & 5 & 8 & 2 & 10 & 6 & 26 + c & 3 & 1.1 & 61.2 & 12.8 & 49.7 & 0.87 & 9.4 & 4.1 & 2 & 2 & 0.2 & 84 & 67 & 701 + h & 3 & 1.1 & 41.9 & 12.8 & 49.7 & 0.87 & 9.4 & 4.1 & 2 & 2 & 0.2 & 57 & 46 & 480 + d2 & 3 & 1.1 & 16.4 & 12.8 & 49.7 & 0.87 & 9.4 & 4.1 & 2 & 2 & 0.2 & 22 & 18 & 189 + d1 & 3 & 1.1 & 14.7 & 12.8 & 49.7 & 0.87 & 9.4 & 4.1 & 2 & 2 & 0.2 & 20 & 16 & 168 + f1 & 4 & 0.4 & 19.6 & 54.3 & 22.0 & 2.30 & 22.4 & 6.8 & 3 & 4 & 1 & 19 & 12 & 53 + f2 & 4 & 0.4 & 8.8 & 54.3 & 22.0 & 2.30 & 22.4 & 6.8 & 3 & 4 & 1 & 8 & 5 & 24 + | here we present the first results from alma observations of 1 mm polarized dust emission towards the w43-mm1 high mass star forming clump .
we have detected a highly fragmented filament with source masses ranging from 14 to 312 , where the largest fragment , source a , is believed to be one of the most massive in our galaxy .
we found a smooth , ordered , and detailed polarization pattern throughout the filament which we used to derived magnetic field morphologies and strengths for 12 out of the 15 fragments detected ranging from 0.2 to 9 mg .
the dynamical equilibrium of each fragment was evaluated finding that all the fragments are in a super - critical state which is consistent with previously detected infalling motions towards w43-mm1
. moreover , there are indications suggesting that the field is being dragged by gravity as the whole filament is collapsing . |
the knudsen number is defined as @xmath69 , where @xmath70 is the density - dependent mean - free path within the gas and @xmath71 is the typical length scale associated with the diffusing particle @xcite . for the cs atoms ,
the mean free path is given by @xmath72 , where @xmath73 is the interspecies scattering cross - section and @xmath74 @xcite the @xmath75-wave scattering length between two rb atoms ( @xmath76 is the bohr radius ) . on the other hand , in the absence of the rb cloud ,
the dynamics is dominated by collisions with the trapping potential in the radial directions @xmath77 .
we thus identify @xmath78 , the radial @xmath33 width of the thermal rb cloud with @xmath79 the radial angular trap frequency . despite using ultracold temperatures in the micro kelvin range ,
the dynamics of the atoms in our experiments can be described classically .
quantum effects are in general relevant when the condition @xmath80 , with @xmath81 the thermal de broglie wavelength , is satified @xcite . for the highest rb density measured
, we have instead @xmath82 for rb and @xmath83 for cs . in addition , the mean - field contribution of the cold rb cloud to the effective potential on cs @xcite , @xmath84 with @xmath85 the reduced mass , can be used as an order - of - magnitude estimate for the highest rb density .
we have @xmath86 , a value which is negligible compared to the potential depth of @xmath87 for cs and the bath s thermal energy scale @xmath88 .
furthermore , the dynamics of the cs atoms in the cloud are not dominated by the trapping potential , but by the rb density distribution which experiences a six times lower mean field shift due to the lower rb - rb scattering length @xcite .
the lifetime of cs atoms in our experiment is limited by three - body losses .
this influences the evolution of the cs atom s spatial distribution .
therefore , we have measured the inverse three - body loss rate as a function of rb density and account for three - body loss in our simulations .
the spatially averaged lifetime is @xcite , @xmath89 where @xmath90 is the three - body loss coefficient and @xmath91 denotes the spatial average over the cloud s volume @xmath92 .
the local lifetime of a cs atom is proportional to @xmath93 , so cs loss predominantly happens at the center of the rb cloud . to determine the coefficient @xmath90
, we measure @xmath94 for various rb densities by inserting cs atoms into the cold cloud and recording the spatial distribution at various interaction times .
fitting eq .
( [ eq : lossrate ] ) to the experimental data , as shown in fig .
[ fig : lossrate ] , yields @xmath95 . here , cs is prepared in @xmath96 and rb in @xmath97 and the loss coefficient is roughly one order of magnitude larger than for cs in the @xmath98 and rb in the @xmath99 state @xcite .
using the local lifetime @xmath100 as a function of density , allows accounting for three - body recombination in our simulations . for cs in @xmath96 and rb in @xmath97 circles : measured lifetimes .
blue solid line : fit of eq . to the data yielding @xmath101 .
the error marks the @xmath102 confidence interval originating from statistical uncertainties and is visualized by the blue shaded area . ] in this section , we present details of the hard - sphere collision simulations .
we describe cs tracer atoms as point - like masses that evolve in the absence of a collision according to newton s equation of motion , @xmath103 , during each time step @xmath40 , where @xmath104 is the confining potential including gravity .
we model the rb cloud as a thermalized classical gas with time - independent density @xmath50 and temperature @xmath11 .
this assumption is justified in view of the extreme number ratio between cs and rb atoms . at each time step , the local collision probability @xmath41 with the position and velocity dependent collision rate @xmath42
is computed .
comparing @xmath105 to a uniform random number determines if the collision occurs . for each collision event ,
a rb collision partner is randomly generated and the cs atom is assigned a new velocity @xmath106 by solving the two - body collision in the center - of - mass system under the assumption of isotropic @xmath75-wave scattering .
this assumption is compatible with our hard - sphere model and well justified for the kinetic energies involved .
the rb atom does not play any further role in the simulation as a new , statistically representative rb atom is drawn for each collision .
such a procedure is justified because the gas is dilute and correlations due to multiple scatterings with the same rb atom can be neglected .
we account for three - body losses with the probability @xmath48 , with the local lifetime @xmath49 . in view of the initial nonthermal kinetic energy of the tagged atoms
, the velocity @xmath107 of the rb collision particle before the collision should be determined with care .
drawing the velocity of the rb atom from a maxwell - boltzmann distribution is flawed , as the speed - dependence of @xmath42 would depopulate high - velocity classes of @xmath106 due to more frequent scattering and prevent thermalization .
this is the same reason , why the fluctuating force in the langevin equation becomes speed dependent .
the rb velocity should instead be drawn from a modified ( not normalized ) maxwell - boltzmann distribution given by ( see ref . @xcite , chapter 5.4 ) , @xmath108 where @xmath109 , @xmath110 and @xmath111 are polar coordinates of the rb speed vector about @xmath112 as an axis with probability distributions @xmath113
, @xmath114 , @xmath115 ; @xmath116 is the relative velocity of the two collision partners and @xmath117 the usual maxwell - boltzmann - distribution @xmath118 with collision simulated distributions ( red histogram , solid border ) , normalized to the diffused fraction . in green ( solid border ) , the fraction of atoms that have not underwent any collision are shown .
] we numerically solve the langevin equation using a 3rd - order runge - kutta integrator with a typical time step of @xmath119 and 1500 trajectories , corresponding to the number of observed cs atoms .
the fluctuation - dissipation relation for an equilibrium bath implies a direct connection between damping and noise terms . for the velocity - dependent friction coefficient,@xmath120 the detailed - balance condition and the gibbs form of the thermal equilibrium condition lead to the speed - dependent random force @xmath121 , where @xmath122 is a normally distributed random force with zero mean and unit variance , and @xcite @xmath123 equation corresponds to multiplicative noise @xcite . on the other hand , for the usual langevin equation
the friction coefficient is velocity independent and reads @xcite , @xmath124 the random force is accordingly characterized by @xmath125 equation corresponds to additive noise @xcite .
from ( a)-(e ) with collision simulated distributions ( red histogram , solid border ) , normalized to the diffused fraction . in green ( solid border ) , the fraction of atoms that have not underwent any collision are shown . ]
our numerical simulations are conducted without free parameters .
the simulation parameters are obtained from independent measurements .
the trapping potential consists of two gaussian beams at a wavelength of @xmath126 and the potential is precisely known from beam profile measurements , consistent with trap - frequency measurements with rb and measured single cs atom motion in absence of rb .
the vertical beam along the @xmath127 axis has a power of @xmath128 .
its focus is located at @xmath129 and its beam waist is @xmath130 , where the intersection of the two beams defines the coordinate system s origin .
the horizontal beam along @xmath20 is elliptical with beam waists of @xmath131 and @xmath132 , a power of @xmath133 , and its focus at @xmath134 .
the beams produce a potential per intensity @xmath135 of @xmath136 for cs and @xmath137 for rb . for all tracer diffusion measurements at peak densities @xmath22 and interaction times @xmath19 .
overlaps for the hard - sphere collision simulation are shown in ( a ) , overlaps for the langevin simulation with velocity - dependent friction , computed for the diffused fraction , are shown in ( b ) .
all overlaps for both models are @xmath139 , whereas the langevin simulation with velocity - independent friction shown in fig .
4 of the main article yields overlaps down to @xmath140 . ] while @xmath141 and the rb atom number @xmath142 are determined with time - of - flight thermometry and absorption imaging @xcite .
it is important to note that a harmonic approximation of the trapping potential does not hold .
we consider instead the exact gaussian beam potential throughout the simulation .
the interspecies scattering length @xmath74 @xcite for rb in @xmath99 and cs in @xmath98 is used as an approximation for the different hyperfine states employed in this experiment .
the full measurement of the cs spatial distribution when the rb density is varied at a fixed interaction time corresponding to figs . [ fig : density_variation_coll]-[fig : density_variation_langevin ] is shown in figs .
[ fig : density_variation_langevin_all]-[fig : density_variation_coll_all ] , together with langevin and discrete hard - sphere collision simulations .
spatial distributions for various interaction times at a fixed rb density , corresponding to fig .
[ fig : sketch ] , are shown in figs .
[ fig : time_variation_langevin]-[fig : time_variation_coll ] .
a comparison of the overlaps for a full , consistent measurement of different densities and interaction times is depicted in fig .
[ fig : overlaps_2d ] , where the overlap of measured data with both the collision simulation ( fig . [ fig : overlaps_2d](a ) ) and the langevin model with velocity - dependent friction coefficient ( fig .
[ fig : overlaps_2d](b ) ) yields values @xmath143 .
by contrast , a langevin model with velocity - independent friction coefficient yields typical overlaps of @xmath144 .
we emphasize that the simulations have been performed with a single , consistent set of independently measured parameters for all data , without free parameter .
99 g. a. bird , _ molecular gas dynamics and the direct simulation of gas flows _ , ( oxford university press , 1994 ) . c. cercignani , _ rarified gas dynamics _ , ( cambridge university press , 2000 ) .
e. p. muntz , ann .
fluid mech .
* 21 * , 387 ( 1989 ) .
f. sharipov , _ rarefied gas dynamics : fundamentals for research and practice _ ( wiley , 2016 ) .
a. j. masters and t. keyes , phys . rev .
a * 27 * , 2603 ( 1983 ) .
w. coffey , y. kalmykov , and j. waldron , _ the langevin equation _
( world scientific , 1996 ) .
d. gillespie and e. seitaridou , _ simple brownian diffusion _
( oxford university press , 2012 ) .
p. mazur and i. oppenheim , physica * 50 * , 241 ( 1970 ) .
j. blum , s. bruns , d. rademacher , a.voss , b. willenberg , and m. krause , phys .
lett . * 97 * , 230601 ( 2006 ) .
t. li , s. kheifets , d. medellin , and m.g .
raizen , science * 328 * , 1673 ( 2010 ) .
r. huang , i. chavez , k. m. taute , b. lukic , s. jeney , m. g. raizen , and e .-
florin , nature phys .
* 7 * , 576 ( 2011 ) .
s. kheifets , a. simha , k. melin , t. li , and m. g. raizen , science * 343 * , 1493 ( 2014 ) . l. ferrari , chem . phys . * 336 * , 27 ( 2007 ) .
l. ferrari , chem .
phys . * 428 * , 144 ( 2014 ) .
j. metcalf and p. van der straten , _ laser cooling and trapping _ ( springer , 2002 ) .
r. mazo , _ brownian motion _
( oxford university press , 2002 ) .
m. hohmann , f. kindermann , b. gnger , t. lausch , d. mayer , f. schmidt , and a. widera , epj quantum technology * 2 * , 1 ( 2015 ) . m. hohmann , f. kindermann , t. lausch , d. mayer , f. schmidt , and a. widera , phys . rev .
a * 93 * , 043607 ( 2016 ) .
see supplementary material for details .
s. chapman and t. cowling , _ the mathematical theory of non - uniform gases _ ( cambridge university press , 1970 ) .
a. a. dubkov , p. hnggi , and i. goychuk , j. stat . mech .
* p01034 * ( 2009 ) .
t. takekoshi , m. debatin , r. rameshan , f. ferlaino , r. grimm , h .- c .
ngerl , c. r. le sueur , j. m. hutson , p. s. julienne , s. kotochigova , e. tiemann , phys .
a * 85 * , 032506 ( 2012 ) .
pethik and h. smith , _ bose - einstein condensation in dilutes gases _ ( cambridge university press , 2002 ) .
e. g. m. van kempen , s. j. j. m. f. kokkelmans , d. j. heinzen , and b. j. verhaar , phys .
lett . * 88 * , 093201 ( 2002 ) .
n. spethmann , f. kindermann , s. john , c. weber , d. meschede , and a. widera , phys .
lett . * 109 * , 235301 ( 2012 ) j. m. sancho , m. san miguel , s. l. katz , and j. d. gunton , phys .
a * 26 * , 1589 ( 1982 ) .
l. ferrari , chem . phys . * 257 * , 63 ( 2000 ) .
inguscio , m. , stringari , s. , wieman , c.e.,proceedings of the international school of physics `` enrico fermi '' , course cxl ( 1998 ) . | understanding the motion of a tracer particle in a rarefied gas is of fundamental and practical importance .
we report the experimental investigation of individual cs atoms impinging on a dilute cloud of ultracold rb atoms with variable density .
we study the nonequilibrium relaxation of the initial nonthermal state and detect the effect of single collisions which has eluded observation so far .
we show that after few collisions , the measured spatial distribution of the light tracer atoms is correctly described by a generalized langevin equation with a velocity - dependent friction coefficient , over a large range of knudsen numbers .
diffusion is an essential and omnipresent transport phenomenon in nature .
the motion of a tagged particle in a fluid is determined by its mass @xmath0 and by the density of the fluid via the knudsen number @xmath1 @xcite . in the regime of large densities
@xmath2 , the interparticle collision frequency is high and the fluid may be treated as a continuum medium . on the other hand , for low densities @xmath3 , individual collisions matter and
the discrete nature of the fluid is apparent .
the only closed equation applicable to all values of @xmath1 is the boltzmann equation for the phase - space distribution of the particles @xcite
. solutions of this nonlinear kinetic equation have been obtained in the extreme situations of vanishing and infinite knudsen numbers , and in the brownian limit of a heavy tracer particle , @xmath4 , where @xmath5 is the mass of the fluid particles @xcite . however , despite its central importance for the foundations of statistical physics and the study of e.g. fluid flows in the upper atmosphere and aerosols dynamics @xcite , much less is known , both theoretically and experimentally , about the long - standing problem of light tagged particles diffusing in a dilute gas at intermediate @xmath1 @xcite .
an alternative and highly successful description of a tagged particle is offered by the langevin equation @xcite . in this stochastic approach , newton s equation of motion for the particle
is extended by a friction force and a fluctuating force that account for the interaction with the surrounding fluid .
the langevin equation enables simple evaluation of the macroscopic properties of the diffusing particle , without the need to compute complicated collision integrals as in the boltzmann equation .
it is valid in the brownian limit of a heavy tracer particle , where the friction coefficient is independent of the velocity of the particle @xcite .
recent experimental studies of brownian motion have been reported in gases @xcite and liquids @xcite .
however , the usual langevin equation with constant friction does not hold for light tracer particles in dilute gases @xcite . here
, we experimentally observe the motion of individual @xmath6cs atoms impinging on a dilute cloud of ultracold @xmath7rb atoms as shown in fig .
[ fig : sketch ] .
we exploit the variable density of the cloud to explore a wide range of knudsen numbers , from @xmath8 at the center to arbitrarily large values at the edges , for the light cs tracer atom with @xmath9 . by initially accelerating
the laser cooled cs atoms to a nonthermal kinetic energy of about @xmath10 , we are able to investigate in detail the nonequilibrium relaxation induced by the collisions with the cloud particles , from ballistic to diffusive motion ( @xmath11 is here the temperature of the thermal rb cloud and @xmath12 the boltzmann constant ) .
the high resolution of our setup further allows to detect the effect of a single collision on the dynamics of the tracer atom .
discrete hard - sphere collision simulations reveal that only few collisions suffice to thermalize a cs atom to the cloud temperature .
we additionally demonstrate that the measured spatial distribution of tagged cs atoms is well described , without any free parameters , by a generalized langevin equation with a velocity - dependent friction coefficient @xcite . to our knowledge , this extension of the langevin equation has never been experimentally verified so far . in our experiment ,
laser cooled @xmath6cs atoms are released from a magneto - optical trap ( mot ) into a crossed dipole trap containing a dilute gas of @xmath7rb atoms ( fig .
[ fig : sketch](a , b , c ) ) @xcite .
after entering the cloud , a cs atom undergoes collisions with the cold rb atoms and thermalizes .
ultracold temperatures in the micro kelvin range lead to slow dynamics , both for the gas and the tracer atoms , with thermal velocities around @xmath13 mm s@xmath14 . while at room temperature , heavy tracer particles are bombarded at an extremely high rate ( of the order of @xmath15 hz in air @xcite ) ,
ultra low temperatures , a small mass ratio of the order of unity and low gas densities reduce this collision rate to values between @xmath16 hz at the center and @xmath17 at the edges of the gas cloud .
hence the effective mean free time between two collisions is in the experimentally accessible range of @xmath18ms .
for the finite size cloud , there is a non - zero probability that a cs atom does not collide at all and moves unperturbed through the cloud ( fig .
[ fig : sketch](f ) ) .
in order to observe the spatial distribution of the tagged cs atoms , we freeze their positions after a given ( interaction ) time @xmath19 after release from the mot by turning on a strong 1d optical lattice in the @xmath20-direction .
we subsequently remove the rb cloud from the trap and record the atomic position with fluorescence imaging ( fig .
[ fig : sketch](b ) ) @xcite .
the spatial distribution for any fixed time is determined by accumulating atomic positions over about 600 realizations .
the time evolution of the distribution is obtained by varying the interaction time @xmath19 ( figs .
[ fig : sketch ] ( d)-(f ) ) .
we use only a few cs atoms ( @xmath21 on average ) per experimental run .
interspecies effects such as self - diffusion and self - thermalization are therefore negligible . moreover ,
the thermal de broglie wavelength is smaller than the interparticle distance for the densities considered @xcite and the dynamics of the atoms can hence be described classically .
figures 2(a)-(c ) show the measured cs spatial distributions for three increasing values of the peak rb density @xmath22 after an interaction time @xmath23 ms . for low rb density ( fig .
2(a ) ) , we observe a bimodal distribution with a fraction of unperturbed cs atoms ( right ) that has passed through the rb cloud ( grey gaussian curve ) and a larger fraction of diffused atoms that has thermalized inside the cloud ( left ) .
the existence of the unperturbed fraction reveals the discrete nature of the dilute rb gas . when the rb density is increased ( figs .
2(b)-(c ) ) , the unperturbed fraction shrinks , as the collision probability grows with @xmath22 , and the diffused fraction does not fully penetrate the cloud .
this may be qualitatively understood by noting that diffusion slows down with increasing density so that cs tracer atoms need more time to reach the center of the cloud .
in addition , three - body losses become important close to the center , where the density is highest , further decreasing the number of cs atoms .
a measurement of the three - body loss rate is presented in ref .
@xcite .
the lifetime of a cs atom at the peak of the spatial distribution in fig .
2(c ) is @xmath24 ms .
three - body - losses are therefore accounted for in all simulations .
ms for rb center densities of ( a ) @xmath25@xmath26 , ( b ) @xmath27@xmath26 , and ( c ) @xmath28@xmath26 .
the simulations are performed with 1500 atoms and normalized to the measured distribution .
the agreement is excellent for both diffused ( left ) and unperturbed ( right ) fractions with an overlap larger than @xmath29 . the unperturbed fraction is mostly composed of tracer atoms that have not suffered any collision ( green , simulated ) . ] mm ) for various maximum rb densities @xmath30 .
good agreement with simulated collision free cs atoms ( green hexagons ) is observed .
( b ) simulated exponential decay of the difference @xmath31 between the cs kinetic temperature and the rb temperature , as a function of the average number of collisions @xmath32 in a homogeneous cloud .
the solid line is an exponential fit to the data with a @xmath33 number of collisions of @xmath34 . ] in order to quantitatively describe the measured spatial distribution and the thermalization process , we first employ discrete hard - sphere collision simulations @xcite .
we numerically solve newton s equation of motion for individual cs tracer atoms elastically colliding with the thermal gas of rb atoms in three dimensions .
spatial distributions as shown in fig .
2 were obtained by projecting the atomic positions onto the @xmath20-axis .
a tagged cs atom initially starts with an effective gaussian spatial distribution at @xmath35 with width @xmath36 and @xmath37 at time @xmath38 , and a maxwell - boltzmann velocity distribution with temperature @xmath39 .
these parameters were extracted from an independent time - resolved reference measurement in the absence of the cloud .
the properties of the rb cloud were also obtained by independent measurements , so that there are no free parameters in the simulations . at each time
step @xmath40 , a collision between a tracer atom and a rb atom is described by a local collision probability , @xmath41 , where @xmath42 is the collision rate . for nonthermal test particles with velocity @xmath43
, the collision rate is given by ( eq . 5.4,5 in ref .
@xcite ) @xmath44 , \label{eq : v_dep_collision_rate}\ ] ] where @xmath45 is the scattering cross section , @xmath46 and @xmath47 the error function .
we analogously describe three - body losses by the probability @xmath48 , with the local lifetime @xmath49 @xcite .
since the density @xmath50 changes spatially , the expressions above are evaluated at the current position of a tracer .
the results of the discrete collision simulations are compared to experimental tracer distributions in figs .
2(a)-(c ) .
we obtain excellent agreement between measured ( blue ) and simulated ( red ) spatial distributions , both for the unperturbed ( right ) and the diffused fraction ( left ) , with an overlap , @xmath51 , between the two distributions of more than @xmath29 .
we attribute the slight discrepancies to measurement uncertainties of the independently determined parameters entering the simulations @xcite .
figure 3(a ) depicts the relative number of atoms in the unperturbed fraction ( defined as those atoms with position @xmath52 mm , marked by the vertical dashed line ) , for various values of the maximum rb density @xmath30 .
the simulations show that the unperturbed fraction is mostly composed of tracer atoms that do not undergo any collision ( green , simulated ) .
the small difference between the unperturbed and non - collided fractions is due to the sharp cut at @xmath53 .
in addition , we find that the computed temperature difference @xmath31 between the kinetic temperature of the tagged cs atoms and the cloud temperature decays exponentially with the number of average collisions @xmath32 with a @xmath33 decay constant of 1.3 collisions ( fig . 3 ( b ) ) .
this indicates thermalization after a few collisions .
we estimate , for example , a relative temperature difference of @xmath54 after @xmath55 collisions .
[ t ] .
the simulations are performed with 1500 atoms and normalized to the diffused fraction .
excellent agreement is found for the thermalized diffused fraction ( left ) between measured data and velocity dependent friction langevin description , with overlap larger than @xmath29 . by contrast , the velocity - independent description yields only poor agreement with measured data with overlaps as low as . in both cases ,
the langevin equation ( 2 ) does not capture the unperturbed fraction ( right ) , as it assumes that all tracer atoms experience damping and noise . ]
we next test the validity of the stochastic langevin approach by simulating single particle langevin trajectories @xcite . for discrete time steps @xmath40 ,
the langevin equation for the velocity @xmath56 of a tagged particle takes the form @xcite @xmath57 here , @xmath58 is a drag force with friction coefficient @xmath59 , @xmath60 a fluctuating force and @xmath61 the confining force including gravity .
while for large , heavy test particles the friction coefficient @xmath59 is velocity independent , it acquires an explicit velocity dependence for small , light particles @xcite .
it was theoretically shown that @xcite @xmath62 to order @xmath63 and for values of @xmath64 not much larger than unity .
to our knowledge , the predictions of the langevin equation with speed - dependent damping coefficient have never been verified experimentally so far .
we also note that owing to the detailed - balance condition , the nonlinear friction implies a multiplicative , velocity - dependent fluctuating force @xcite .
the results of the langevin simulations are presented in fig
. 4 @xcite .
we have again accounted for three - body losses by removing tracer atoms at each time step with the probability @xmath65 .
as before , the cloud density @xmath50 , and thus the friction coefficient @xmath59 , were evaluated at the position of the tagged atoms .
we observe that the diffused fraction of cs atoms ( left ) is well described by the langevin simulations ( blue ) , without free parameters , for all values of the rb density . however , as expected , the unperturbed fraction ( right ) is not captured by the langevin approach which assumes that all test particles experience damping and fluctuations .
the overlap between the simulated and measured spatial distributions for the diffused fraction is larger than @xmath66 .
we once more ascribe the small deviation to the experimental uncertainties of the reference measurements .
by contrast , langevin simulations with the usual velocity - independent friction coefficient ( brown ) yield overlaps as low as @xcite .
the following physical picture emerges from our experimental and theoretical investigations .
our ability to accurately identify the effect of a single collision allows the distinction of two different types of dynamics : thus , cs atoms that do not experience any collision pass ballistically through the dilute rb cloud ( unperturbed fraction ) . on the other hand ,
a single collision event dissipates on average more than half of the initial nonthermal kinetic energy and suffices to trap a tagged cs atom inside the cloud , leading to additional collisions ( diffused fraction ) .
after few more collisions , a tracer atom will be thermalized with the rb cloud .
its dynamics is correctly described , over a wide range of knudsen numbers , by a generalized langevin equation with a velocity - dependent friction coefficient .
the origin of this unfamiliar speed dependence may be understood by noting that the velocity @xmath43 of a heavy brownian particle ( @xmath67 ) is much smaller than that of the gas atoms . as a result , both the collision rate and the friction coefficient are independent of @xmath43 in this limit . by contrast , for an atomic tracer with a mass @xmath68 , both velocities are of the same order . here , an explicit velocity dependence of the collision rate and of the friction coefficient can no longer be neglected . however , as we have shown , in both cases , a highly successful continuous langevin description over a coarse - grained timescale larger than the mean free time between impacts is possible .
we thank axel pelster and james anglin for helpful discussions .
this work was partially funded by the erc starting grant nr .
278208 , the collaborative project thermiq ( grant agreement 618074 ) and the sfb / trr49 .
t. l. acknowledges funding by carl zeiss stiftung , d.m .
is a recipient of a dfg - fellowship through the excellence initiative by the graduate school materials science in mainz ( gsc 266 ) , f.s .
acknowledges funding by the studienstiftung des deutschen volkes . |
the concept of a `` field '' population of galaxies as distinct from the group / cluster populations has existed since the earliest days of extragalactic astronomy ( hubble 1936 ) and it is used recurrently in studies aimed to explore the effects of large scale environment on galaxy properties . however , the definition of `` field '' is fuzzy .
the distribution of galaxies in space is actually strongly clustered and a large fraction of them is prone to form gravitationally - bound multiple systems , from very populated clusters to loose groups , the majority being in normal groups ( tully 1987 ) . _
isolation _ is an important requirement beyond the concept of `` field '' galaxies .
a galaxy is isolated if it has not suffered any interaction with another normal galaxy or with a group environment over a hubble time or at least since approximately one half of its mass was assembled .
this makes the observational finding and study of isolated galaxies important because , among other reasons , ( i ) they can be used as comparison objects in studies of the environmental effects on galaxies belonging to groups and clusters , and ( ii ) they are ideal for confronting with theoretical and model predictions of galaxy evolution .
the fact that properties of galaxies change with environment has been known for a long time .
the main observable dependencies with environment from cluster centers to the sparse `` field'' are seen for the morphological mix , the global colors , and the specific star formation ( sf ) rate ; for recent reviews see park et al .
( 2007 ) , avila - reese et al .
( 2005 ) , and the references therein .
these properties are inferred mainly from photometric observations .
in fact , to study observationally the effects of high density environments on the galaxy properties , a good understanding of the isolated , non perturbed , galaxies is necessary .
these galaxies are also required as a control sample for studying interacting galaxies .
for example , in hernndez - toledo et al .
( 2005 ; see also conselice 2003 ) we have compared the photometric concentration , asymmetry and clumpiness parameters of local interacting and `` field '' disk galaxies with the aim to establish a relatively easy way for identifying interacting disk galaxies in high redshfit samples . among isolated galaxies ,
the observational study of disk galaxies is of special interest because , on one hand , they are expected to be more drastically affected by environmental and interaction effects ; on the other hand , according to the current paradigm of cosmic structure formation , the formation of disks inside hierarchically growing cold dark matter ( cdm ) haloes is a generic process .
a large amount of work has been done in modeling the evolution of ( isolated ) disk galaxies .
well observed samples of isolated disk galaxies are hence important in order to confront with model predictions ( e.g. , avila reese & firmani 2000 ; cole et al .
2000 ; boissier & prantzos 2000 ; firmani & avila
reese 2000 ; yang , mo & van den bosch 2003 ; zavala et al .
2003 ; pizagno et al . 2005 ; dutton et al . 2006
; kassin , de jong & weiner 2006 ; gnedin et al .
2006 ) .
a uniformly selected and observed sample of isolated disk galaxies is also crucial for studying intrinsic secular processes able to affect the structure , morphology and dynamics of galaxies , for instance , the formation and evolution of bars , circular rings , lopsidedness , and bulges .
however , chances are that isolated galaxies may show evidence of disturbances not associated with intrinsic processes , which opens the necessity to explore other alternatives . on this line , cosmological numerical simulations within the cdm model show that inside the galaxy size haloes there survives a large population of subhaloes ( klypin et al .
1999 ; moore et al .
1999 ) , but reionization and feedback could inhibit the formation of luminous ( satellite ) galaxies inside most of these subhaloes ( e.g. , bullock , kravtsov & weinberg 2000 ; benson et al .
the subhalos and the associated gas clouds could produce signs of distortion on isolated galaxies ( threntam , moller & ramirez - ruiz 2001 ; pisano , wilcots & liu 2002 ) .
if these signs are clear , then observations in radio could reveal the presence of 21cm hydrogen line emission associated to a pure gas companion galaxy .
but if such an emission is absent , there is still the possibility that the perturber is just a dark matter ( sub)halo without any baryonic component ( threntam et al .
recently , karachentsev , karachentseva & huchtmeier ( 2006 ) , by analyzing a large sample of isolated galaxies , have reported the finding of four such possible cases .
homogeneous observational data samples of isolated disk galaxies are crucial for obtaining transparent scaling relationships and correlations that can be appropriately confronted with model predictions ( see , e.g. zavala et al .
2003 ) . in recent years
, some groups have been working on the compilation and observation of such samples ( e.g. , pisano et al .
2002 ; allam et al . 2005 ; koopmann & kenney 2006 ) . for galaxies in voids , which are the most likely
to be isolated , see rojas et al .
( 2004,2005 ) .
it is worth to mention that the important requirements for all these samples are : well defined and strong isolation criteria , a uniform - quality data acquisition in several wavelengths , and completeness when possible . we have carried out optical ccd photometry for a representative set of galaxies in the northern catalogue of isolated galaxies ( hereafter cig , karachentseva 1973 ) .
this is one of the best defined and most complete catalogues of isolated galaxies .
the aim of this paper is to present a global @xmath0 photometric and morphological analysis for a subsample of 44 spiral galaxies from the cig catalog .
it is strongly emphasized that all the observations were done with the same ccd detector . after applying uniform reduction and analysis procedures , an homogeneous set of photometric and morphological data
is guaranteed .
cig galaxies cover a wide range in luminosities , surface brightnesses , morphological types and colors .
their relative simplicity and closeness , as compared with galaxies in other environments , offer a unique opportunity to have a more detailed and less confused interpretation of their structural , photometric and morphological properties .
the outline of the paper is as follows .
section [ datasample ] summarizes the selection criteria applied to the isolated galaxy sample that are relevant to our photometric study , and describes the observations and reduction techniques used here .
section [ magnitudes ] presents a comparison of our estimated total magnitudes against those in the literature . in section [ morphology ] , we discuss the observed morphology based on mosaic @xmath7band and @xmath7band sharp / filtered images , @xmath8 color index maps , and composed near infrared ( nir ) @xmath3 images extracted from the two micron survey archives . emphasis is put on the presence of disturbed morphology .
section [ physmorphology ] presents our estimates of the optical ( @xmath0 ) and nir @xmath9band concentration , asymmetry , and clumpiness ( @xmath4 ) structural parameters . in section [ s4 ]
we explore and discuss some basic correlations among the @xmath10 photometric and structural parameters in this sample that could be useful for comparative studies involving galaxies in other environments .
section [ s5 ] provides a summary of the paper .
finally , an appendix is devoted to the presentation of @xmath0 magnitudes at two other concentric circular apertures .
we have carried out an observational program at the observatorio astronmico nacional at san pedro mrtir ( oan - spm ) , baja california , mxico , devoted to obtain uniform ccd photometric data for one of the most complete and homogeneous samples of isolated galaxies currently available , the cig catalogue of karachentseva ( 1973 ) .
this sample amounts to more than 1050 galaxies in the northern hemisphere .
the ccd @xmath0 images in the johnson - cousins system were obtained with a site1 detector attached to the 1.5 m and 0.84 m telescopes at oan - spm covering an area of about @xmath11 and @xmath12 , a typical seeing of 1.7 arcsec and a scale of 0.51@xmath13/pixel and 0.85@xmath13/pixel respectively . the original number of galaxies in three observing runs amounts to 52 galaxies . from these ,
6 galaxies obtained under bad observing conditions and 2 ellipticals were eliminated , yielding a final sample of 44 isolated spiral galaxies of the present study .
we applied no special strategy in selecting this current subset .
availability of observing time and weather conditions were the main factors constraining the number of observed galaxies .
some aspects of the selection criteria for the cig sample that are most relevant to the present and further photometric analyses are stated here .
the isolated galaxies in the cig sample were selected from a visual search of the palomar sky survey .
the catalogue samples the sky north of @xmath14 .
the vast majority of objects is found in high galactic latitude regions ( @xmath15 ) and as a sample , it is reasonably complete ( @xmath16% ) in the magnitude range @xmath17 ( hernandez - toledo et al .
the selection criteria used in assembling the cig can be expressed by the following relations : @xmath18 where@xmath19 is the apparent separation between the candidate isolated galaxy of apparent diameter @xmath20 and any other neighbor galaxy of apparent diameter @xmath21 . under these criteria ,
any other galaxy of comparable size ( @xmath22 ) should be at a distance of at least 20 times its diameter ( projected on the sky ) from the isolated galaxy . assuming a typical galaxy diameter d @xmath23 20 kpc and a peculiar velocity relative to the hubble flow v @xmath24 km s@xmath25 ( rivolo & yahil 1981 ) , the time required for an intruder galaxy to traverse 20 diameters is @xmath26 yr .
this is a first - order estimate of the time since the last equal - size galaxy - galaxy interaction for a cig system and it suggests that cig galaxies are reasonably isolated .
it is therefore expected that only the intrinsic properties of the individual galaxies in the cig should influence the observed photometric and morphological properties .
karachentseva ( 1973 ) included other isolation criteria ( coded as 1 and 2 in her original catalogue ) depending on whether or not other galaxies , within a factor of 4 in size , were found near a 20 diameter boundary or even if a galaxy is definitely not isolated according to the cig criteria .
less isolated galaxies account for less than @xmath27 of our cig sample and have been excluded from the present study .
a journal of the photometric observations is given in table 1 .
column ( 1 ) gives the original catalogue number , columns ( 2)-(9 ) give the number of frames per filter , the integration time ( in seconds ) , and seeing conditions ( in arcsec ) .
table 2 reports some relevant information on the observed isolated galaxies obtained from the literature .
column ( 1 ) is the cig catalogue number , column ( 2 ) reports other identifications , column ( 3 ) the apparent total @xmath28 magnitude from the lyon extragalactic database ( leda ) , column ( 4 ) the hubble type from leda , column ( 5 ) the apparent total @xmath28 magnitude from the nasa extragalactic database ( ned ) , column ( 6 ) the radial velocity in km s@xmath25 corrected for virgocentric infall from leda .
images were debiased , trimmed , and flat - fielded using standard iraf procedures .
first , the bias level of the ccd was subtracted from all exposures .
a run of 10 bias images was obtained per night , and those were combined into a single bias frame which was then applied to the object frames .
the images were flat - fielded using sky flats taken in each filter at the beginning and/or end of each night .
photometric calibration was achieved by nightly observations of standard stars of known magnitudes from the pg0231 + 051 field of stars ( landolt 1992 ) with a color range @xmath29 and @xmath30 .
once the principal extinction coefficients in @xmath28 , @xmath31 , @xmath1 and @xmath32 were estimated , transformations of the instrumental magnitudes to a standard system were calculated according to the following equations : @xmath33 where @xmath28 , @xmath31 , @xmath1 and @xmath32 are the standard magnitudes , @xmath34 , @xmath35 , @xmath36 and @xmath37 are the instrumental ( and airmass - corrected ) magnitudes , and @xmath38 and @xmath39 are the transformation coefficients for each filter . a constant value associated to the sky background was subtracted using an interactive procedure that allows the user to select regions on the frame free of galaxies and bright stars .
errors in determining the sky background , are , in fact , the dominant source of errors in the estimation of total magnitudes .
the most energetic cosmic - ray events were automatically masked using the cosmicrays task , and field stars were removed using the imedit task when necessary . within the galaxy itself ,
care was taken to identify superposed stars . a final step in the basic reduction involved registration of all available frames for each galaxy and in each filter to within @xmath40 pixel .
this step was performed by measuring centroids for foreground stars on the images and then performing geometric transformations using geomap and geotran tasks in iraf .
apparent magnitudes for each galaxy were estimated in three concentric circular apertures .
this was achieved in @xmath0 bands by using the phot routines in iraf . here
we report the total apparent magnitudes while in the appendix we report apparent magnitudes at two other circular apertures , also in @xmath0 bands .
see table a1 .
an estimation of the errors in our photometry involves two parts : ( 1 ) the procedures to obtain instrumental magnitudes and ( 2 ) the uncertainty when such instrumental magnitudes are transformed to the standard system . for item ( 1 ) ,
notice that the magnitudes produced at the output of the iraf routines ( phot ) have a small error that is internal to those procedures . since we have also applied extinction corrections to the instrumental magnitudes in this step , our estimation of the errors
are mainly concerned with these corrections and the estimation of the airmass .
after a least square fitting , the associated errors with the slope for each principal extinction coefficient are : @xmath41 , @xmath42 , @xmath43 and @xmath44 .
an additional error @xmath45 from the airmass routines in iraf was also considered . for item
( 2 ) , the zero point and first order color terms are the most important to consider . after the transformation to the standard system by adopting our best
fit coefficients , the errors from the assumed relations for @xmath38 were respectively 0.02 , 0.04 , 0.02 and 0.02 in @xmath28 , @xmath31 , @xmath1 and @xmath32 , and 0.02 , 0.03 , 0.02 and 0.02 for @xmath39 . to estimate the total error in each band ,
it is necessary to propagate the errors , after considering the corresponding transformation equations .
an estimate of the sky contribution is necessary for quoting the total uncertainties .
this was achieved by estimating the total magnitudes for all galaxies before and after sky subtraction .
typical values @xmath46 , @xmath47 , @xmath48 and @xmath49 are obtained .
total typical uncertainties are 0.1 , 0.12 , 0.12 and 0.15 in @xmath28 , @xmath31 , @xmath1 and @xmath32 bands , respectively .
the estimated total magnitudes in this work were compared with other external estimations when available in the literature .
this has been done for : 1 ) the standard stars and 2 ) the isolated spiral galaxies . for the standard stars , a comparison of our ccd magnitudes with those reported in landolt ( 1992 ) for stars in the field of pg0231 + 051 and the dipper asterism m67 star cluster (
chevallier & ilovaisky 1991 ) are shown in figure 1 .
figure 1 shows no significant deviations between our ccd magnitudes and those reported for the standard stars .
a linear fit to this plot indicates a @xmath50 as the typical internal error for our magnitude estimations .
figure 2 shows a comparison of the estimated magnitudes for the isolated galaxies in @xmath28 , @xmath31 and @xmath32 bands versus the available total magnitudes in the nasa extragalactic database ( hereafter ned ) and aperture photometry in hyperleda databases .
discrepant cases such as cig 80 , 168 , 188 , 434 , and 691 in the different bands are emphasized in the figure .
we find a reasonable agreement with the available values from the literature , except for a few discrepant cases shown in the figure .
hyperleda reports detailed aperture photometry in the b - band for cig 80 . from a plot of the available data
, the reported value should correspond to an aperture size log ( a ) @xmath23 2 while our magnitude corresponds to log ( a ) = 1.58 .
the corresponding magnitude for an aperture similar to ours is 11.60 mag , in complete agreement with our estimation .
hyperleda does not report any aperture photometry in the b - band for cig 168 .
a detailed aperture photometry in the v and i ( cousins ) bands is found instead .
the v - band value in hyperleda corresponds to log ( a ) = 1.44 while ours is log ( a ) = 1.50 .
an extrapolation of the available v and i magnitudes to our aperture size is consistent with our photometry .
this makes us confident that our b - band value is well estimated and suggests that the b - band magnitude reported in hyperleda should correspond to a smaller log ( a ) value . in the case of cig 188 ,
both hyperleda and ned databases report an homogenized magnitude from previously published data , assuming standard johnson ubvri filters .
notice however that ned magnitude is 14.10 @xmath51 with a large error bar . given the absence of detailed aperture photometry and the diffuse nature of this galaxy
( see corresponding images below ) , we suggest that the observed discrepancy is explained by the different aperture sizes .
hyperleda reports a couple of aperture data points for cig 434 ( corresponding to log ( a ) = 1.12 and 1.3 , see gallagher and hunter 1986 ) . by assuming a linear curve of growth , an extrapolation to log ( a ) = 1.5 ( our aperture size ) yields a magnitude value consistent with our data .
ned reports an homogenized magnitude from previously published data , assuming standard johnson ubvri filters . for cig 691 , hyperleda reports a b band magnitude of @xmath52 while ned reports a magnitude of @xmath53 which is significantly closer to our reported value . even more
, the only aperture data point in hyperleda suggests that the reported value corresponds to a smaller aperture size than ours .
finally , the internal accuracy of our photometry was evaluated by comparing the total magnitudes derived from individual exposures .
we find rms differences between individual measurements of @xmath54 , @xmath55 , @xmath56 and @xmath57 .
additional magnitudes at two other concentric circular apertures in @xmath28 , @xmath31 , @xmath1 and @xmath32 for all the isolated galaxies in this study are reported in the appendix .
the estimated apparent magnitudes and the colors of the galaxies in the sample are presented in table 3 .
entries are as follows : column ( 1 ) gives the cig number ; column ( 2 ) gives the logarithmic aperture size in 0.1 arcmin units , according to the hyperleda convention , columns ( 3 ) to ( 6 ) give the observed integrated apparent magnitudes in @xmath28 , @xmath31 , @xmath1 and @xmath32 bands .
finally , columns ( 7 ) to ( 9 ) give the observed @xmath58 , @xmath59 and @xmath8 color indices .
total typical uncertainties in our photometry are 0.10 , 0.12 , 0.11 and 0.16 for @xmath28 , @xmath31 , @xmath1 and @xmath32 bands , respectively .
the @xmath58 corrected colors span the range of @xmath60 mag .
this is comparable to that reported in other samples of non interacting galaxies ( e.g. , de jong 1996 ; verheijen 1997 ) .
we emphasize important differences in blue galactic absorption values between burstein & heiles ( 1982 ) and schlegel et al .
( 1998 ) for cig 103 ( 1.485 vs 0.440 mag ) , cig 138 ( 1.95 vs 0.65 mag ) , and cig 144 ( 2.135 vs 0.510 mag ) .
the fraction of dust seems to be larger for bigger galaxies according to empirical ( e.g. , giovanelli et al .
1995 ; wang & heckman 1996 ; tully et al . 1998 ) and theoretical ( e.g. , shustov et al .
1997 ) arguments .
therefore , the internal extinction correction should depend not only on inclination but also on galaxy scale : @xmath61[mag ] = @xmath62log(@xmath63 ) , where @xmath63 is the major to minor axis ratio , and @xmath62 is a scale dependent coefficient in the given passband @xmath64 . from an empirical analysis ,
tully et al .
( 1998 ) inferred the coefficients @xmath62 in the @xmath65 bands as a function of the galaxy maximum circular velocity . from their data
( given in tully & pierce 2000 ) we have carried out linear correlations of these coefficients with the corresponding magnitudes _ not corrected _ for internal extinction : @xmath66}= -6.30 -0.40 m_b , \ \ \ \ m_b < -16.7 \cr \gamma_r { \rm [ mag]}= -4.20 -0.26 m_b , \ \ \ \ m_r < -17.7 \cr \gamma_i { \rm [ mag]}= -3.40 -0.20 m_i , \ \ \ \ m_i < -18.0 \cr \gamma_k { \rm [ mag]}= -0.85 -0.05 m_k , \ \ \ \ m_k < -19.7
. \label{extin}\end{aligned}\ ] ] for values of the magnitudes larger than the limits given in eq .
( [ extin ] ) , @xmath62 is assumed to be 0 ( no extinction correction ) .
for the band @xmath31 , the line coefficients of @xmath67 were obtained by a simple interpolation of those in the bands @xmath28 , @xmath1 , @xmath32 , and @xmath68 : @xmath69 } = -4.67 -0.29 m_v$ ] , @xmath70 .
the @xmath63 ratios estimated at the @xmath7125mag / arcsec@xmath72 isophote were taken from the hyperleda database .
table 4 shows foreground and internal extinction corrected color indices and absolute magnitudes .
corrections are based on data generated from the dust galaxy maps given in schlegel et al .
( 1998 ) and available in ned database .
entries are as follows : column ( 1 ) gives the identification cig number ; columns ( 2 ) to ( 4 ) give the corrected @xmath58 , @xmath59 and @xmath8 color indices . finally , columns ( 5 ) to ( 8) report the corrected absolute magnitudes in @xmath28 , @xmath31 , @xmath1 and @xmath32 bands . a hubble constant value of 70 km s@xmath25 mpc@xmath25 was adopted .
a more physical correction applied to the luminosities yields a b - band luminosity range ( @xmath73 ) indicating no faint spirals in this sample , except for the case of cig 434 . in figure 3 we plot different color
magnitude diagrams [ @xmath74 vs @xmath75 and @xmath8 colors ] for our sample of isolated galaxies .
there is a mild correlation of colors with magnitude , although the sample is still small .
in fact , a significant dependence of color on luminosity for normal isolated disk galaxies is not expected ( e.g. , avila reese & firmani 2000 ) . the dependence that several papers reported in the past was mainly due to the dependence of the internal extinction on luminosity ( or circular velocity , tully et al .
1998 ) , which we take into account accordingly here .
in order to discuss the optical morphology ( that could be modified by the presence of bars , rings , etc . or external factors ) and its relationship to the global photometrical properties , the images for each isolated galaxy are presented in the form of a mosaic in figure 4 including , from upper - left to lower - right panel : 1 ) a gray scale @xmath1-band image displayed at full intensity to look for faint external details ; 2 ) an @xmath1 band sharp / filtered image to look for internal structure in the form of star forming regions and/or structure embedded into dusty regions ; the filtered / enhancing techniques ( sofue 1993 ) allow the subtraction of the diffuse background in a convenient way for discussing the different morphological details ; 3 ) a ( @xmath2 ) color index map to visualize the spatial distribution of the sf ( light gray is for blue colors while dark
gray is for red colors ) ; 4 ) a composed ( sharp / filtered ) nir @xmath3 image which is a combination of the archive @xmath76 and @xmath77band images from the two - micron survey ( skrutskie et al .
2006 ) to complement the structural and morphological analysis ; and finally , 5 ) the ellipticity @xmath78 and position angle @xmath79 radial profiles from the @xmath32 and composed @xmath3 images to provide evidence of the presence of bars and other structural details .
( figures 4.1 - 4.43 are available in the electronic edition of the journal ) we identify a bar signature if the ellipticity radial profile @xmath78 rises to a maximum @xmath80 required to be above that of the outer disk , while the @xmath79 radial profile shows a plateau ( within @xmath81 ) along the bar ( wozniak et al .
1995 ) .
all the images are oriented according to the standard ( north - east ) astronomical convention .
the nir images are approximately at the same scale as the optical images . for the sake of not crowding , the major diameter ( arcmin ) of the optical images
is specified in the caption text for each galaxy . in some cases ,
not all the foreground stars in each field have been removed .
we use in addition the fact that the median value of the @xmath58 color declines systematically as the morphological type t increases along the morphological sequence .
median integrated total @xmath58 colors of galaxies according to morphological class are given by roberts & haynes ( 1994 ) .
the ugc and the local supercluster ( lsc ) samples in roberts & haynes ( 1994 ) are rather inhomogeneous in terms of environment , but the interacting objects were excluded from their analysis .
the median colors of these samples will be used as a reference in the following discussion . *
cig 1*. the galaxy was classified as sabbc ( ned ) .
the @xmath7band and sharp / filtered images show a conspicuous fan - like structure at the end of the arms resembling a tidally disturbed galaxy .
the optical images show three sharp - defined arms and a bar also visible in the composed jhk image .
we classify this galaxy as sbbc .
the total @xmath58 color is representative of sc types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy . *
cig 4*. the galaxy was classified as sbc ( ned ) .
the gray - scale @xmath7band image shows an inclined galaxy through a series of dust lanes .
the optical r - band sharp / filtered and the @xmath2 color map images show two multiple and knotty arms and a strongly reddened central region .
the composed jhk image shows two main arms and an elongated central barred - like region .
we classify this galaxy as sabc .
the total @xmath58 color is strongly reddened and not representative of sc types .
the photometric ( @xmath78 and @xmath79 ) profiles show evidence of a weak bar within the first 15 arcsec of this galaxy . *
cig 33*. this galaxy was classified as sab(rs)cd ( ned ) .
the gray - scale @xmath7band image shows a faint outer arm in the east emphasizing its asymmetric appearance .
its internal s - shaped structure surrounded by a series of dust lanes and two prominent blobs in the north and south - east of the inner spiral arms are also apparent .
the @xmath2 color index map shows a ring - like structure .
the @xmath82band image clearly shows an inner barred structure from which two arms emerge .
we classify this galaxy as sb(rs)c .
the total @xmath58 color is representative of sc types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy .
* cig 53*. the galaxy was classified as a sb(rs)c ( ned ) .
the gray - scale @xmath7band and the sharp / filtered images show knotty features along multiple arms , a prominent bar encircled by a ring elongated in the direction perpendicular to the bar , and the presence of strong dusty structures .
the @xmath2 color map permits us to see a red central region and bluer arms .
the jkh - band image confirms the barred and ringed structures and traces of the arms .
we classify this galaxy as sb(r)bc .
the @xmath58 color is representative of sbc types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy . *
cig 56*. the galaxy was classified as a sb(rs)b ( ned ) .
the gray - scale @xmath7band image shows a barred galaxy with two dominant arms and a third faint arm to the north - west . at the outskirts , the two arms become diffuse .
the sharp / filtered image and @xmath8 color map show a blue ring elongated in direction of a red bar .
the @xmath3 composed image shows a prominent bar and two dominant arms .
we classify this galaxy as sb(r)b .
the @xmath58 color is representative of sb types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy . *
cig 68*. the galaxy was classified as sab(s)a ( ned ) .
the gray - scale @xmath7band image shows a central elongated structure from which two arms emerge , while the sharp / filtered image shows traces of a bar .
the @xmath8 color map shows a bluer ring elongated in direction of the bar and an adjacent dust lane .
the composed @xmath3 image shows a prominent bar and two arms .
we classify this galaxy as sb(r)a .
the @xmath58 color is representative of sa types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy .
* cig 80*. the major axis of this galaxy is longer than our ccd frame .
however we comment on the structure found within 4.3 arcmin .
the galaxy was classified as a sa(s)b ( ned ) .
the gray - scale @xmath7band and sharp / filtered images show a multi - armed and strongly asymmetric pattern enhanced by the presence of dust lanes , resembling a strongly perturbed system .
the @xmath8 color map shows an inner ring and bluer arms .
the composed @xmath3 image shows a barred - like structure from which two prominent arms seem to emerge and confirms the red nature of a ring oriented in the direction of the bar .
we classify this galaxy as sb(r)b .
the @xmath58 color is consistent with sab types .
the photometric ( @xmath78 and @xmath79 ) profiles show weak evidence of a bar within the first 25 arcsec of this galaxy . *
cig 103*. the galaxy was classified as sab(rs)c ( ned ) .
this is another example of an apparently multi - armed galaxy seen through a series of strong dust lanes as shown in our @xmath7band , sharp / filtered and @xmath2 color map images .
the color map also shows an elongated red central region apparently encircled by a pseudo - ring alongside the direction of a bar .
the composed @xmath3 image shows evidence of a weak bar from which two arms emerge .
we classify this galaxy as sb(r)c .
the @xmath58 color is consistent with sc types .
the photometric ( @xmath78 and @xmath79 ) profiles show weak evidence of a bar or an elongated structure within the first 15 arcsec of this inclined galaxy . *
cig 116*. the galaxy was classified as rsb(s)a ( ned ) .
all our images show this galaxy with two inner well - defined arms making an s - shaped structure . these arms become diffuse as they wind out , giving the appearance of an external ring oriented almost perpendicular to the bar .
the internal structure is seen through strong dust lanes .
the composed @xmath3 image shows the presence of a bar and two inner dominant arms .
we classify this galaxy as rsb(s)a .
the @xmath58 color is consistent with sa types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy . *
cig 123*. this galaxy is classified as sb(rs)bc ( ned ) . from the global distribution of optical light
this galaxy appears asymmetric .
our images show an apparently multiple set of faint southern arms and a strong southern dust lane .
the red central region , including the bar as well as the bluer ring and arms are emphasized in the @xmath2 color index map .
the composed @xmath3 image clearly shows a strong bar , and a ring oriented in the direction of the bar .
we classify this galaxy as sb(r)c .
the @xmath58 color is consistent with sbc types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in this galaxy . *
cig 138*. this galaxy was classified as sb(s)d ( ned ) .
this is an inclined galaxy seen through a series of dust lanes , causing the arms to appear multiple in nature .
the @xmath2 color index map clearly emphasizes the reddened nature of the light distribution .
in contrast , the composed @xmath3 image shows mainly two arms and an elongated central structure resembling a bar .
we classify this galaxy as sbc .
the reddened @xmath58 color does not correspond to scsd types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar in this galaxy .
* cig 139*. the galaxy was classified as sb(s)m pec ( ned ) .
the @xmath8 color index map shows , however , a central reddened region and an outer bluer region similar to what is observed in spirals .
there are no archive @xmath3 images for this galaxy .
we preserve neds classification in this case .
notice that the obtained @xmath58 color is representative of sm / irr types .
the photometric i - band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar . * cig 144*. the galaxy was classified as sb ( ned ) .
this is an edge - on galaxy showing a peanut - like - shape red bulge in the sharp / filtered image .
however , the @xmath2 color index image and the composed @xmath3 image show a different structure . we classify this galaxy as sabb .
the reddened @xmath58 color corresponds more to sa types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles are consistent with the presence of a bar in the first 30 arcsec of this inclined galaxy .
* cig 151*. the galaxy was classified as sadm ( ned ) .
this is an inclined and apparently multi - armed spiral galaxy .
the central region appears elongated in the sharp / filtered image and red in the @xmath2 color map .
the composed @xmath3 image suggests a bar structure and two adjacent arms .
we classify this galaxy as sabc .
the @xmath58 color corresponds to sc types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show a weak evidence of a bar in this inclined galaxy .
* cig 154*. the galaxy was classified as sbcd ( ned ) .
this apparently disturbed galaxy shows an elongated and reddened central region .
the winding of the inner arms suggests a ring oriented in the direction of the bar .
the inner arms appear bifurcated at various places and faint outer arms are also perceived .
the composed @xmath3 image shows a bar and a ring .
we classify this galaxy as sb(r)cd .
the @xmath58 color corresponds to scdsd types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in the first 10 arcsec and of a ring at about 12 arcsec in this galaxy . *
cig 168*. the galaxy was classified as sab(s)bc ( ned ) .
the galaxy appears moderately flocculent in the optical images with an elongated and red central region .
the composed @xmath3 image also suggests an elongated central structure and only two prominent arms .
the @xmath58 color corresponds to scdsd types .
we classify this galaxy as sab(s)cd .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar .
* cig 175*. the galaxy was classified as sa(s)a pec ( ned ) .
the optical images show traces of diffuse and faint arm - like features and a prominent central region
. the knotty appearance in the central part may be either intrinsic or caused by some coincident field stars not appearing in the composed @xmath3 image .
the @xmath58 color corresponds to smim types .
we classify this galaxy as sa pec .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles are consistent with smooth precessing arm - like features . *
cig 180*. the galaxy was classified as sa(rs)c ( ned ) .
the optical images show a multiple set of tightly wound arms and a reddened central region apparently encircled by an inner ring .
in contrast , the composed @xmath3 image shows a prominent central region and traces of the adjacent disk .
we classify this galaxy as sa(r)b .
this is an interesting case of a non - barred spiral with a symmetric inner ring that also shows hints of a circumnuclear ring .
the @xmath58 color corresponds to sab types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show weak evidence of a central ring within the first 10 arcsec and emphasize , in contrast with the optical image , a smooth nature of the central near - ir structure in this galaxy .
* cig 188*. the galaxy was classified as sab(s)d ( ned ) .
this galaxy shows multiple arms in the optical images .
the sharp / filter image allows us to see a central bar structure that is red in the ( @xmath2 ) color map .
the low signal in the composed @xmath3 image does not permit us to appreciate the central bar but the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles confirm the presence of a central bar .
the total @xmath58 color is representative of sdsm types .
we classify this galaxy as sb(s)d .
* cig 208*. the galaxy was classified as sb ( ned ) .
this is a highly inclined galaxy .
the sharp / filter and color index images show the central region resembling a long bar and the outer arms .
the total @xmath58 color is representative of scdsd types .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles also resemble a bar within the first 25 arcsec .
we classify this galaxy as sabcd .
* cig 213*. the galaxy was classified as s0 ( ned ) .
we are using the @xmath28 band image to show a bar surrounded by an almost circular ring oriented in the direction of the position angle of the bar .
the total @xmath58 color is representative of es0 types .
we classify this galaxy as rsb0 .
the photometric @xmath28 and @xmath3 band ( @xmath78 and @xmath79 ) profiles confirm a bar within the first 25 arcsec . *
cig 224*. the galaxy was classified as sb(rs)d ( ned ) .
the optical images show a galaxy of flocculent appearance with an outer pseudo - ring and evidence of a bar .
the composed @xmath3 image is of low signal but still shows an elongated central region .
the total @xmath58 color is representative of sdsm types .
we classify this galaxy as rsbd .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show a bar within the first 15 arcsec .
* cig 237*. the galaxy was classified as sc ( ned ) .
this is an edge - on galaxy .
the total @xmath58 color is representative of scd types .
based on the observed prominence of the bulge region in the optical and composed @xmath3 images , we assume a sc classification . * cig 309*. the galaxy was classified as sa(r)ab ( ned ) .
the optical and @xmath3 images show an inner and outer ring .
the optical images show an intermediate region of strongly wound and knotty arms ? .
the total @xmath58 color is representative of s0sa types .
we classify this galaxy as rs(r)a .
the smooth behavior of the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) is consistent with an early - type galaxy .
* cig 314*. the galaxy was classified as sab(rs)c ( ned ) .
this is a multi - armed knotty spiral showing a red central region .
the composed @xmath3 image barely shows part of two arms emanating from an elongated central region .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show evidence of a weak bar .
the total @xmath58 color is representative of scsd types .
we classify this galaxy as sab(rs)c .
* cig 434*. the galaxy was classified as i m ( ned ) .
the optical images show a magellanic - type irregular with an elongated and red region resembling a bar .
there are no detected images available in the 2mass survey .
the total @xmath58 color is representative of i m types .
we keep the i m classification . the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show evidence of a bar . *
cig 472*. the galaxy was classified as sab(rs)c ( ned ) .
the optical images show a pattern of two arms that become bifurcated at the outer regions .
in contrast , the composed @xmath3 image shows a single prominent northern arm extending to the south .
the total @xmath58 color is representative of scsd types .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show no evidence of a bar .
we classify this galaxy as sa(rs)c .
* cig 518*. the galaxy was classified as sa(s)c ( ned ) .
the optical images show a pattern of two arms that become bifurcated at the outer regions and a red elongated central region .
in contrast , the composed @xmath3 image shows only two prominent arms that appear to enclose an elongated central region .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show evidence of a bar .
the total @xmath58 color is representative of sbsc types .
we classify this galaxy as sb(s)bc .
* cig 528*. the galaxy was classified as sabc ( ned ) .
the optical images show a spiral pattern with two faint thin outer arms in the ne and sw resembling tidal tails .
the composed @xmath3 image shows only part of two central arms .
the total @xmath58 color is representative of scsd types .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar .
we classify this galaxy as sa(rs)cd .
* cig 549*. the galaxy was classified as sa(rs)c ( ned ) .
the optical images show a multiple set of arms while the composed @xmath3 image shows two arms at the center and an intriguing feature in the se .
the total @xmath58 color is representative of sbc types .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show evidence of a bar .
we classify this galaxy as sb(rs)c .
* cig 604*. the galaxy was classified as ( r)sb(s)a ( ned ) .
the optical images show a boxy bulge and two arms forming an outer ring .
the sharp / filter and composed @xmath3 images show enhanced emission at the starting region of the arms in the ne and sw .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show a large scale bar .
the total @xmath58 color is representative of sab types .
we classify this galaxy as rsb(s)a pec .
* cig 605*. the galaxy was classified as sb(r)ab ( ned ) .
the optical images show multiple arms emanating from the end points of a bar .
an inner ring oriented along the major axis of the bar is also appreciated .
two bright knots at the opposite sides of the bar are also seen in the composed @xmath3 image .
the total @xmath58 color is representative of sabsb types .
we classify this galaxy as sb(r)b .
* cig 691*. the galaxy was classified as sb(rs)d ( ned ) .
the optical images show a bar from which a diffuse set of arms emanates . at the composed @xmath3 image
, only the bar can be seen .
the total @xmath58 color is representative of scscd types .
we classify this galaxy as sb(rs)cd .
* cig 710*. the galaxy was classified as sa(s)cd ( ned ) .
the optical and composed @xmath3 images show an elongated central region resembling a bar . the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles also show a weak evidence of a bar .
the total @xmath58 color is representative of sdsm types .
we classify this galaxy as sb(s)cd .
* cig 889*. the galaxy was classified as sa ( ned ) .
this is an edge - on galaxy showing a prominent peanut - shaped bulge from the optical to the nir .
this shape has been interpreted as representing a barred structure seen at high inclinations ( c.f . bureau & athanassoula 2005 ) .
we classify this galaxy as sba .
the @xmath58 color corresponds to sa types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles are consistent with the presence of a bar in the first 30 arcsec of this inclined galaxy . *
cig 906*. the galaxy was classified as sb ( ned ) .
this is an edge - on galaxy .
the sharp / filtered image allow us to see a prominent dust band along the plane and the @xmath2 color index image shows this plane as highly reddened but with a bluer outer region .
a bulge can be appreciated in the composed @xmath3 image .
we preserve ned classification .
the @xmath58 color corresponds to sab types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles may be consistent with the presence of a bar in the first 30 arcsec of this galaxy .
* cig 910*. the galaxy was classified as sbab ( ned ) .
this is another highly inclined galaxy that shows two main diffuse arms in the optical images .
the sharp / filtered and @xmath2 color index images show evidence of a red barred structure and an outer bluer region .
the composed nir image confirms both the barred structure and the presence of two arms .
we classify this galaxy as sbab .
the @xmath58 color corresponds to sa types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles are consistent with the presence of a bar in the first 30 arcsec of this galaxy . *
cig 911*. the galaxy was classified as sbb ( ned ) .
the optical images show multiple arms emanating from a central ringed region that encloses a bar .
the bar is red in color while the arms are bluer .
the composed @xmath3 image permits us to see a barred structure enclosed by a ring oriented in the direction of the bar .
we classify this galaxy as sb(r)b .
the @xmath58 color corresponds to sb types .
the photometric i and jhk band ( @xmath78 and @xmath79 ) profiles show evidence of a bar in the first 15 arcsec of this galaxy . *
cig 935*. the galaxy was classified as sab(rs)cd ( ned ) .
the optical images show a multi - arm pattern with an elongated central region .
in contrast , the composed @xmath3 image shows traces of a two - arm central pattern .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show weak evidence of a bar .
the @xmath58 color corresponds to sbcsc types .
we classify this galaxy as sab(rs)c .
* cig 976*. the galaxy was classified as sab ( ned ) .
the optical images show an apparently disturbed spiral pattern while the composed @xmath3 image shows only a two - arm pattern .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar .
the @xmath58 color corresponds to sabsb types .
we classify this galaxy as sa(s)ab .
* cig 983*. the galaxy was classified as sab(rs)c ( ned ) .
the optical images show a multi - armed spiral pattern with a slightly elongated central region .
the sharp / filter image shows only a two - arm pattern and also a slightly elongated central region .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles show weak evidence of a bar .
the @xmath58 color corresponds to scdsd types .
we classify this galaxy as sb(rs)cd .
* cig 1004*. the galaxy was classified as sb(s)c ( ned ) .
the optical images show a bar and a multi - armed spiral pattern .
the composed @xmath3 image shows only a two - arm spiral pattern and a prominent large - scale bar .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles also show evidence of a large bar .
the @xmath58 color corresponds to sbcsc types .
we classify this galaxy as sb(s)c .
* cig 1009*. the galaxy was classified as sa ( ned ) .
the optical images show a set of tightly wound arms forming an inner ring .
the composed @xmath3 image also shows a central arm pattern .
the photometric @xmath32 and @xmath3 band ( @xmath78 and @xmath79 ) profiles do not show evidence of a bar .
the @xmath58 color corresponds to sabsb types .
we classify this galaxy as s(r)ab .
* cig 1023*. the galaxy was classified as sb(r)b ( ned ) .
the optical images show arms forming an outer ring .
a bar is seen enclosed by an inner ring alongside the major axis of the bar .
the arms appear to start at the end parts of the bar .
the composed @xmath3 image shows prominence of the bar .
the @xmath58 color corresponds to sabsb types .
we classify this galaxy as rsb(r)b .
table 5 is a summary of the morphological results found in this work .
column ( 1 ) gives the original catalogue number , column ( 2 ) gives the hubble type as reported in ned , column ( 3 ) gives the hubble type as estimated in this work , column ( 4 ) remarks the presence of bars / rings , column ( 5 ) remarks the presence of multiple arms from the optical ( @xmath0 ) images , and column ( 6 ) reports the bar ellipticity ( corrected for inclination ) .
ned database contains morphological information on subtypes for almost all these isolated galaxies , except for some of the most inclined ones ( cig 144 , 208 , 237 , 889 and 906 ) .
42@xmath83 of the galaxy sample is earlier than sbc and 52@xmath83 is of sbc type or later .
the catalogue information concerning bars ( confirmed and presumed ) comprised 27 galaxies before this work , and we were able to add this information to 8 other galaxies .
this indicates that up to @xmath84 of the isolated galaxies in this subsample shows evidence of barred structure : for 63.5% the evidence is clear ( sb galaxies ) and for 16% the bars are weak or suspected ( sab galaxies ) .
the bar fraction is roughly the same for early and late types .
we have measured the @xmath85band and @xmath3 isophotal ellipticities associated with a bar and calculated the maximum ellipticity , @xmath86 .
this quantity ( corrected by inclination ) is related to a measure of the bar strength such as the gravitational bar torque ( laurikainen et al .
2002 ) . column 6 in table 5 gives the values of @xmath86 for our sample . among barred galaxies ,
the average value of @xmath86 is @xmath87 . if we include the values of @xmath88 , 7 early type galaxies and 6 late type galaxies have @xmath89 , which is commonly considered as evidence of strong bar .
similarly , the catalogue information for rings in our sample comprised before 16 galaxies ; in this work , it has been added to 8 other galaxies , accounting now to @xmath90 of the sample .
finally , we emphasize the finding of clear morphological signatures of disturbance at least in two galaxies , cig1 and cig80 .
the disturbance is revealed in the former case by a broad fan like shape of the outer arms , while in the latter case it is revealed by a strong global asymmetric pattern of the multi
arms ( see figs . 4 and 10 , and 4.1 ) .
physical morphology has appeared as an alternative to classifying galaxies on the basis of physical properties rather than on visual features ( see morgan & osterbrock 1969 ; abraham et al .
1996 ; conselice 1997 ; bershady , jangren & conselice 2000 , among others ) .
conselice ( 2003 , and more references therein ) has provided a useful framework for classifying galaxies closely tied to underlying physical processes and properties .
conselice ( 2003 ; hereafter c2003 ) argues that the major ongoing and past formation modes of galaxies can be distinguished using three model independent structural ( photometric ) parameters , which allow for a robust classification system .
these parameters are the concentration of stellar light ( c ) , its asymmetric distribution ( a ) , and a measure of its clumpiness ( s ) .
below we present below the @xmath4 parameters measured at various passbands for our observed sample of isolated galaxies .
the @xmath4 characterization of this set of isolated galaxies is also helpful as a comparative sample for interpreting similar results of other surveys that sample galaxies in a wide range of environments ( e.g. , c2003 ; hernandez toledo et al .
2005 ; 2006 ) .
next , we briefly review each one of the @xmath4 parameters
. _ concentration of light @xmath91.- _ the concentration index @xmath91 is defined as the ratio of the 80% to 20% curve of growth radii ( @xmath92 , @xmath93 ) , within 1.5 times the petrosian inverted radius at r(@xmath94 ) ( @xmath95 ) normalized by a logarithm : @xmath96 ( see for more details c2003 ) .
the concentration is related to the galaxy light ( or stellar mass ) distributions .
_ asymmetry @xmath97.- _ the asymmetry index is the number computed when a galaxy is rotated @xmath98 from its center and then subtracted from its pre - rotated image , and the summation of the intensities of the absolute value residuals of this subtraction is compared with the original galaxy flux ( see for more details c2003 ) .
this parameter is also measured within @xmath99 .
the @xmath97 index is sensitive to any feature that produces asymmetric light distributions .
this includes galaxy interactions / mergers , large star - forming regions , and projection effects such as dust lanes ( conselice 1997 ; conselice et al . 2000 ) . _
clumpiness @xmath100.- _ galaxies undergoing sf are very patchy and contain large amounts of light at high spatial frequency . to quantify this , the clumpiness index @xmath100 is defined as the ratio of the amount of light contained in high frequency structures to the total amount of light in the galaxy within @xmath99 ( c2003 ) .
the @xmath100 parameter , because of its morphological nature , is sensitive to dust lanes and inclination ( c2003 ) .
_ measurement of @xmath4 parameters.- _ the measurement of the @xmath4 parameters for the isolated spiral galaxies was carried out in several steps : \(i ) close field and overlapping stars were removed from each image ; ( ii ) sky background was removed from the images ; ( iii ) the center of each galaxy was considered as the barycenter of the light distribution and the starting point for measurements ; ( iv ) the @xmath4 parameters for all the spiral isolated galaxies were estimated directly , i.e. isolated galaxies are not influenced by light contamination from any other galaxy of similar size in the neighborhood ( isolation criteria ) ; ( v ) galaxies with high inclinations or axis ratios could introduce systematic biased trends in the values of the @xmath4 parameters ( c2003 ) .
isolated galaxies whose apparent axial ratios yield `` inclinations '' larger than @xmath101 are represented as open circles on the corresponding plots . since @xmath4 parameters are mostly reported in the @xmath1 band by other authors , in order to compare , we provide here the calculated @xmath4 parameters and their errors , also in the @xmath1 band ( table 6 ) .
the @xmath4 values in the other bands for our observed isolated spirals will be provided by the authors upon request . by sorting the sample in early and
late type spirals ( sasb and sbcsm , respectively ) , the corresponding average and standard deviation values of the @xmath4 parameters are : @xmath102 , @xmath103 , @xmath104 , and @xmath105 , @xmath106 , @xmath107 . our mean values are consistent with those reported in c2003 for the frei et al .
( 1996 ) sample of non interacting galaxies , except in the case of @xmath108 ( sbcsm ) .
notice however that irregulars were included in in sbcsm class while in c2003 these galaxies are separated .
an interesting question is how the @xmath4 parameters do change with wavelength .
figure [ cumdistr ] shows the cumulative distribution function of the @xmath4 parameters at @xmath28 , @xmath1 , @xmath109 and @xmath68 bands .
this comparison let us see visually that there are significant systematic changes in the @xmath4 parameters with wavelength . the concentration @xmath91 becomes higher from bluer to redder bands , specially for those galaxies with low and intermediate values of @xmath91 . in the case of both the asymmetry @xmath97 and clumpiness @xmath100 parameters , their values strongly decrease from bluer to redder bands , more as larger are these parameters .
in figure [ cas_band ] we plot the average and standard deviation values of the @xmath4 parameters vs wavelength ( color band ) for our sample sorted in early and late type spirals ( sasb left panel and sbcsm right panel , respectively ) .
the @xmath4 parameters of later types show on average more dependence ( and scatter ) with wavelength than the early types . among the @xmath4 parameters ,
the clumpiness is the most sensitive to wavelength .
the optical and ir emission in galaxies are dominated by different populations of stars and are subject to dust absorption at different levels .
the structures that are dominated by older stellar populations are more prominent and less affected by extinction in the ir than in the optical . in the ir one
observes a higher bar fraction than in the optical , as well as cases where the bulge appears more prominent , the spiral arms less flocculent and rings less prominent ( eskridge et al . 2000 ) .
the latter points towards an earlier type classification from the ir images than from optical ones .
the same trends mentioned above are observed for our sample of isolated galaxies .
it should be remarked that , in our attempt to describe the morphology of the observed galaxies , we also have made use of the @xmath3 images from the the two micron - survey .
the re classification presented here ( see table 5 ) preserves the optically observed morphology but takes into account the nir bar morphology . in general , previous results concerning the differences in morphology as passing from optical to nir bands ( eskridge et al .
2000 ) agree with the ones seen in our subsample of isolated galaxies , though the fraction of isolated galaxies for which these differences become significant is actually small in our case ( @xmath110 ) . concerning bars in our isolated galaxies , they come into a variety of sizes , shapes and color distributions ; from apparently strong , to small ones confined to the central parts of galaxies and up to the oval - shaped bulges , suggesting a range of strengths , lengths and mass distributions . we have shown that the fraction of galaxies in our sample with clear evidence of optical / ir bars ( sb galaxies ) is 63% , while 16% more show some evidence of weak bars ( sab galaxies ) .
these fractions are in agreement with estimates from larger samples of galaxies . for example , eskridge et al .
( 2000 ) determined the fraction of strongly barred galaxies in the @xmath111band for a sample of 186 spirals _ from different environments _ to be 56% , while another 16% is weakly barred .
for the same sample , the fraction of barred galaxies reported in the optical is almost a factor of two smaller than in the nir .
we do not find such a strong difference with the passband in our sample of isolated galaxies .
we also reported the presence of inner ( r ) and outer ( r ) rings when possible , but a detailed ring morphology ( buta 1986 ; 1995 ) was not attempted .
the fraction of galaxies with rings in our sample is high , 55% .
notice that the observed fraction of bars and rings in the present paper can hardly be a bias of our observing procedure since we simply selected objects according to their availability in the sky .
the high fractions of bars for the isolated galaxies found here , similar to the fractions observed in other environments , could be suggesting that interactions and the global effects of the group / cluster environment are not crucial for the formation / destruction of bars .
how do bars form in isolated environments ?
it is known that the presence and evolution of bars in a hubble time depends on the host galaxy structure , the dark matter halo structure , the disk - to - halo ratio , as well as on the environment ( e.g. , athanassoula 2003 ; berentzen , shlosman & jogee 2006 ; coln et al .
resolution n - body simulations of isolated disks embedded in cdm halos show that extended strong bars form almost always , but they slow down as a result of angular momentum transport to the disk and halo ( debattista & sellwood 2000 ; athanassoula & misiriotis 2002 ; valenzuela & klypin 2003 ) ; eventually , the bars may dissolve forming a pseudobulge ( e.g. , avila - reese et al .
2005 ; berentzen et al .
2006 ) .
the isolated environment may be ensuring the presence of dynamically cold disks that can form a variety of stellar bars .
the observed bar fraction could also be a consequence of long lived bars , or alternatively , of bars that recurrently form , self
destroy , and resurrect due to gas accretion ( bournaud & combes 2002 ) .
constant gas accretion is a condition more viable in isolated environments than in group / clusters .
our data reveal no difference in the relative bar fraction of early ( sasb ) and late ( sbcsm ) type galaxies .
if any , among the barred galaxies , the late type subsample contains a larger fraction of weakly barred ( sab ) galaxies than the early type subsample .
eskridge et al . ( 2000 ) also found that the fraction of barred galaxies almost does not depend on morphological type .
it seems that rather than the fractions , the properties of the bars ( length , strength , surface brightness profile , etc . )
are those that change as a function of the morphology and/or environment ( e.g. , erwin 2005 and references therein ) . for our sample
, we have estimated the bar deprojected maximum ellipticity , @xmath86 .
we do not find significant differences in @xmath86 as a function of morphological type as well as a function of the @xmath4 parameters .
further analysis is necessary to infer the disk and bar properties and compare them with model predictions .
another relevant topic is that of ring formation in isolated environments .
rings of sf are a common phenomenon in disk galaxies .
most rings form by gas accumulation at resonances , usually under the continuous action of gravity torques from a bar pattern , but sometimes in response to a mild tidal interaction with a nearby companion ( buta & combes 1996 ; buta 1999 ) . in either case
a resonance is a very special place in a galaxy where sf can be enhanced and may proceed either as a starburst or continuously over a period of time .
most of the observed rings in our galaxies are of the type encircling the end of the bars and elongated along the bar s position angle .
singular cases of outer rings , inner rings and a probable circumnuclear ring were detected .
contrary to bars , most of the observed rings in these isolated galaxies show bluer color distributions suggesting that their stellar populations are more similar to those in their hosting disks than those in the bars .
however , some of our composed @xmath3 images also show the prevalence of the rings in the nir . the existence of this old population rings underlying star forming rings suggest a strong coupling between the stellar and gaseous components in the resonance regions .
finally , it should be mentioned that in any of the 44 isolated galaxies studied here , we did not find strong signatures of interactions or perturbations .
however , in two cases ( cig1 and cig80 ) moderate morphological distortions can be seen , which could evidence some level of dynamical disturbance , though , these distortions could hardly be produced by strong interactions , as is the case of the isolated disturbed galaxies reported in karachentsev et al .
( 2006 , see introduction ) .
satellite accretion could explain the distortions seen in cig1 and cig80 .
on the other hand , as mentioned above , bars and rings are axi
symmetric structures that can be explained as a product of the internal secular evolution of disks . interactions and perturbations may induce and amplify bar / ring formation but can also contribute to their fast dissolution .
a larger sample is necessary in order to explore whether or not a fraction of isolated galaxies shows evident signatures of interaction , a question of relevance as mentioned in the introduction .
light concentration , asymmetry and clumpiness @xmath4 parameters have been used in alternative galaxy classification schemes ( see for references 5 ) .
the @xmath4 parameters allow also the possibility to classify galaxies according to their interaction state ( c2003 ; hernndez toledo et al .
2005,2006 ) , which is useful in high redshift studies . in this sense , it is important to have a well studied comparative sample of local isolated galaxies . in spite of the small number of galaxies in our current sample , we will introduce below an indicative discussion of the measured @xmath4 parameters in different color bands and of their trends with other galaxy properties .
figure [ casplanes ] shows the loci of the isolated sasb and sbcsm galaxies in the projected planes of the @xmath7band @xmath4 space .
only the averages and their standard deviations are shown ( crosses and continuous error bars ) .
the boxes indicate the amplitude of variation of the @xmath4 parameters ( lower and upper limits ) from @xmath112 , @xmath109 to @xmath68 bands . for comparison , the @xmath7band averages and standard deviations of galaxies in interacting s+s pairs ( hernndez toledo et al .
2005 ) , and starburst and ultra luminous infrared ( ulir ) galaxies ( c2003 ) are also plotted .
visually , the major difference between isolated spirals and interacting , starburst and ulir spirals takes place in the @xmath6 plane .
the quantitative measure of @xmath91 in our isolated spirals span the range @xmath113 , in agreement with other works .
the average and standard deviation values are @xmath114 .
it is known that for lenticular and elliptical galaxies , the concentrations are typically larger than for spirals ( e.g. , c2003 ; hernndez toledo et al .
we have also found that @xmath91 systematically increases with the passband for almost each one of the isolated galaxies ( fig .
[ cumdistr ] ) .
this is in agreement with previous finds that the scalelength of spirals is smaller in nir bands than in the optical ones ( e.g. , de jong 1996 ) , which is probably pointing out probably to an inside out galaxy formation scenario : bluer colors trace younger stellar populations , and if the disk is more extended in the optical than in the nir , then the outer disk might be younger ( more recently assembled ) than inner regions .
metallicity gradients would emulate such an effect , but it seems that this is not the case ( de jong 1996 ) . according to our current understanding of galaxy formation , disks form generically inside out within growing cdm halos ( see for a recent review avila
reese 2006 ) .
their concentrations ( or surface brightnesses ) depend mainly on the spin parameter of the halo .
cdm halos span a wide lognormal distribution of the spin parameter , hence , one expects also a wide range of concentrations for the disks .
probably , the observed distribution of concentrations for isolated disk galaxies is not as wide as we would expect from theory .
it should be also taken into account that internal secular processes after disk formation rearrange the mass ( light ) distribution , and that the presence of a large bulge in early type spirals tends to increase their @xmath91 parameter with respect to galaxies with smaller bulges
. a fair comparison of concentrations observed in isolated spirals with model predictions is certainly a promising avenue of research . in the cases of ellipticals ( and
probably the bulges of early type galaxies ) , theory suggests that they are more concentrated due to the violent and dissipative processes that are at the basis of their formation : major mergers of gaseous disks .
concentration of light has been shown to correlate with properties of galaxies like hubble type , color , and surface brightness ( okamura et al .
1984 ; c2003 ; see below ) .
more recent studies have also shown that concentration of light correlates with internal scaling properties such as velocity dispersion , size , luminosity , and black hole mass ( graham et al . 2001 ) .
concerning asymmetry , the method used here gives a simple quantitative measure of how a galaxy deviates from axi symmetry .
the asymmetry parameter @xmath97 has been shown to be sensitive mainly to galaxy interactions / mergers , but also is influenced by sf clumps , dust lanes , and projection effects .
the quantitative measure of @xmath97 in the present sample of isolated galaxies spans roughly the range @xmath115 , the average and standard deviation being @xmath116 .
the later types are slightly more asymmetric on average than the earlier ones .
the asymmetries reported here are definitively lower than the typical ones of interacting disk galaxies ( see fig .
[ casplanes ] ) .
the @xmath97 parameter decreases significantly as the passband is redder ( fig .
[ cumdistr ] ) . for interacting spirals ,
the same trend was observed , but in a much less extent ( hernndez toledo et al .
this suggests that while the ( high ) asymmetry in interacting spirals is mainly of global / external origin hence is more or less the same in different bands , in the case of isolated spirals the ( low ) asymmetry is in part related to sf effects ; this is why @xmath97 is so sensitive to the passband in which it is measured . the question of whether a disk of a spiral galaxy is intrinsically asymmetric or not is of great interest .
some studies have shown that important deviations from axisymmetry exist in the optical and other wavelengths ( rix & zaritsky 1995 ; richter & sancisi 1994 ; c2003 ) . however , systematic attempts to quantify asymmetry and other measures like the @xmath4 parameters in several wavelengths for well selected local samples of galaxies are either rare or missing in the literature .
galaxies undergoing sf are patchy , specially in the bluer bands , and an important fraction of light must be in high spatial frequency structures .
this is quantified through the clumpiness @xmath100 parameter . for our sample of isolated galaxies ,
@xmath117 ranges roughly from @xmath118 to @xmath119 , being the average and standard deviation values @xmath120 .
the @xmath100 parameter is on average larger and more scattered in later types than in the earlier ones as is seen in fig .
[ casplanes ] ( see also below ) .
it is well known indeed that late type galaxies present more current sf activity than the early type ones .
although , the parameter @xmath100 in our isolated galaxies is typically smaller than in interacting spirals , the differences are actually small , ant not so significant as in the case of the asymmetry parameter ( fig .
[ casplanes ] ) .
it is also interesting to note that the increase of the @xmath100 value as the passband is bluer in our isolated galaxies ( fig .
[ cumdistr ] ) is much more significant than in the case of interacting galaxies ( hernndez toledo et al .
2005 ) .
we next explore how the @xmath4 parameters of isolated galaxies correlate with other properties and whether these correlations are sensitive to the passband or not .
figures [ cas_t ] and [ cas_color ] show the @xmath28 , @xmath1 , @xmath109 and @xmath68 band @xmath4 parameters vs morphological type @xmath121 and corrected total @xmath8 color .
nearly edge on galaxies ( inclination @xmath122 ) are plotted with ( open ) circles .
given that @xmath97 and @xmath100 are particularly sensitive to projection effects , it is important to visualize these galaxies since they may be masking any trend . after a visual inspection of figs .
[ cas_t ] and [ cas_color ] , the general conclusion is that any potential trend of the @xmath4 parameters with @xmath121 and total @xmath8 color tends typically to be more robust in the redder bands .
this emphasizes the merits of ir parameters , which are less contaminated from ( transient ) sf effects and better represent the basic structure of galaxies .
we should note that the scatters in these trends , even for the @xmath123 band , are large .
the images from the two micron survey are unfortunately of low quality , specially in the @xmath77band . besides , for several of our galaxies , there are not images in this survey .
therefore , the @xmath109 and @xmath68 band data discussed here should be taken only as indicative ones .
larger samples are needed in order to quantify better the showed dependences in figs .
[ cas_t ] and [ cas_color ] , and infer from them clues to the physics of disk galaxies . notwithstanding this
, we present below a brief discussions on the observed trends . according to figs .
[ cas_t ] and [ cas_color ] , while the concentration tends to be higher for earlier type and redder galaxies , the asymmetry and clumpiness tend to become smaller .
the morphological type is led mainly by the bulge to disk ratio .
the global color is also affected by this ratio .
therefore , it is expected that earlier types be more concentrated and redder
. however , the @xmath91 parameter and the global color are not too sensitive to the bulge to disk ratio for galaxies with intermediate - to - small values of this ratio ( say sb types and latter ) ; therefore , in these cases , the measured @xmath91 and global color reflect mostly the pure disk concentration and color .
thus , that @xmath91 depends on @xmath121 for late types , implies mainly a connection between the spiral arm properties and the disk concentration .
the dependence of @xmath91 on color would imply mainly that less concentrated disks have a more constant sf history , probably because their gas surface densities are low .
concerning asymmetry , the observed dependence on @xmath121 indicates that most of the asymmetry of our isolated spirals is associated with the natural flocculency in later - type galaxies as well with sf , which is more active for later types ( as is evidenced also by the trend of higher @xmath100 values as the types are later and the colors bluer , see above ) . in this interpretation , the effect of large scale perturbations ( c.f .
interactions ) is neglected .
the @xmath97 parameter could be used as a first approximation indicator of interaction signatures in isolated spirals .
an automatic analysis of images in large samples of galaxies provides easily the @xmath97 parameter .
an even more reliable test for interactions would be to produce the loci of the studied isolated spirals in the @xmath6 diagram , where maximum differences between isolated and interacting galaxies are revealed as was seen above ( fig [ cas_t ] ; see also hernndez toledo et al .
the adding of color information is also valuable as evidenced by fig [ cas_color ] .
we present results of our @xmath0 ccd photometry for a set of 44 isolated galaxies selected from the cig catalogue ( karachentseva 1973 ) .
we have shown that our derived parameters are generally in good agreement with the aperture photometry reported in the hyperleda database and other individual photometric works .
in addition , we present multiaperture photometry ( appendix ) in order to facilitate further comparisons and contribute to the existing databases of aperture photometry ( e.g. , hyperleda ) . in a further step
, we have analyzed the morphology of each one of the galaxies based on our mosaic @xmath7band , sharp / filtered @xmath1 band images , 2d @xmath8 color maps , composed nir @xmath3 images from the two micron survey archives and photometric @xmath78 and @xmath79 radial profiles .
a morphological re classification of the galaxies has been presented with emphasis on structural features as bars and rings and global disturbances .
the sample morphological types range from sa to sm , half of the galaxies being earlier than sbc ( sasb ) , and the other half being sbc or later ( sbcsm ) . after our reclassification
, we have found that @xmath124 of the galaxies is clearly barred ( sb ) , while a @xmath125 more shows some evidence of a weak bar ( sab ) .
there is not any significant difference of the bar fraction with the morphological type .
the average and standard deviation values of the @xmath85band deprojected maximum ellipticity of the bars , @xmath86 , is @xmath87 .
there is not any trend of @xmath86 with the morphological type and the @xmath4 parameters .
we have also found that 55% of the isolated galaxies in our sample shows ring structures . finally , we have calculated the @xmath0 , @xmath109 and @xmath77 band concentration , asymmetry , and clumpiness ( @xmath4 ) parameters for the sample .
the @xmath4 averages and standard deviations in the @xmath7band for the sasb and sbcsm subsamples are : @xmath102 , @xmath103 , @xmath126 and @xmath127 , @xmath106 , @xmath128 , respectively .
these values are in good agreement with previous results for non interacting galaxies .
while @xmath91 systematically increases from bluer to redder bands , both @xmath97 and @xmath100 significantly decrease .
the @xmath4 parameters present more robust trends with the morphological type @xmath121 and the total @xmath8 color in the redder bands , suggesting that the basic structure of galaxies is revealed better in the ir and nir bands .
the @xmath91 parameter tends to be higher for earlier type and redder galaxies , while @xmath97 and @xmath100 tend to become smaller .
the @xmath97 parameter could be an excellent way to detect candidates to isolated spirals with signs of interaction by means of an automatic analysis of images .
the loci of our isolated galaxies in the projected planes of the @xmath4 space depend on the morphological type ( and on the color ) .
the major difference between the isolated and interacting , starburst and ulir spirals takes place in the @xmath6 plane .
after the completion of this paper , a work by taylor - mager et al .
( 2007 ) with some aims and results similar to those presented here was posted in the arxiv database .
taylor - mager et al .
analyzed the @xmath4 parameters for a sample of galaxies ( mainly late - types including peculiars ) as a function of wavelength , from uv to ir ( @xmath129 m ) , with the aim of exploring how the galaxy s appearance changes with rest
frame wavelength .
the result leads to a measure of the morphological k
correction for high redshift galaxies .
their results complement well the ones presented here and both are in qualitative agreement , where comparison is possible .
the building of well defined samples of local isolated galaxies with uniform and detailed photometric information is of great relevance because it provides a fair database for comparison with model predictions as well as with observed samples of galaxies in other environments and at higher redshifts . in this paper
we present a first step in the building of such a sample and discuss some preliminary results .
thanks the staff of the observatorio astronmico nacional in san pedro mrtir , b. c. for the help with the observation runs .
we are grateful to the referee , dr .
michael s. vogeley , for his comments that improved the presentation of the paper .
this work was funded by conacyt grant 42810 to h.m.h.t .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
we acknowledge the usage of the hyperleda database ( http://leda.univ-lyon1.fr ) .
since the birth of galaxy photometry ( whitford ( 1936 ) , the amount of photometric data has increased exponentially ( prugniel 1987 ) .
however , these data are inhomogeneous both in quality and format : photographic , photoelectric or more recently , ccd observations .
the data are usually presented as centered aperture photometry through circular or elliptical apertures or as photometric profiles . in order to take into account the continuously growing amount of photometric data and at the same time , to make different photometric data reports somehow comparable , we present in table a1 our estimations of integrated magnitudes in two additional concentric circular apertures .
columns ( 2 ) and ( 3 ) give the logarithm of the aperture radius ( in units of 0.1 arcmin see hyperleda convention ) for each isolated spiral galaxy .
columns ( 4)-(11 ) give their corresponding magnitudes in @xmath28 , @xmath31 , @xmath1 and @xmath32 bands , respectively .
the contribution of the sky to the errors in the magnitudes is relatively small at these apertures .
typical uncertainties in the magnitudes are 0.11 , 0.12 , 0.11 and 0.12 in @xmath28 , @xmath31 , @xmath1 and @xmath32 bands , respectively .
cig 1 & 1@xmath1301200 & 2.3 & 1@xmath130600 & 2.4 & 1@xmath130300 & 2.1 & 1@xmath130300 & 2.4 + cig 4 & 1@xmath1301200 & 2.3 & 1@xmath130600 & 2.2 & 1@xmath130300 & 2.0 & 2@xmath130150 & 1.8 + cig 33 & 1@xmath1301200 & 2.3 & 1@xmath130600 & 2.2 & 1@xmath130300 & 2.3 & 1@xmath130300 & 2.5 + cig 53 & 2@xmath1301200 & 2.1 & 2@xmath130600 & 2.3 & 2@xmath130300 & 2.0 & 2@xmath130300 & 1.8 + cig 56 & 1@xmath1301200 & 2.4 & 1@xmath130600 & 2.5 & 1@xmath130300 & 2.3 & 1@xmath130300 & 2.2 + cig 68 & 1@xmath1301200 & 2.1 & 2@xmath130300 & 2.0 & 2@xmath130120 & 2.0 & 2@xmath13060 & 2.3 + cig 80 & 1@xmath1301200 & 2.7 & 2@xmath130300 & 2.5 & 1@xmath130300 & 2.6 & 2@xmath130120 & 1.8 + cig 103 & 2@xmath1301200 & 2.0 & 1@xmath130600 & 1.7 & 1@xmath130300 & 1.8 & 1@xmath130150 & 1.4 + cig 116 & 1@xmath1301200 & 2.1 & 1@xmath130600 & 1.9 & 2@xmath130150 & 1.6 & 2@xmath130150 & 1.6 + cig 123 & 1@xmath1301200 & 2.0 & 1@xmath130600 & 2.3 & 2@xmath130150 & 1.9 & 2@xmath130150 & 2.0 + cig 138 & 2@xmath1301200 & 2.2 & 2@xmath130600 & 1.9 & 3@xmath130150 & 1.8 & 3@xmath130150 & 2.1 + cig 139 & 2@xmath1301200 & 2.1 & 1@xmath130600 & 2.0 & 1@xmath130300 & 1.8 & 2@xmath130150 & 1.6 + cig 144 & 2@xmath1301200 & 3.0 & 1@xmath130600 & 3.0 & 1@xmath130300 & 2.8 & 1@xmath130300 & 2.6 + cig 151 & 2@xmath1301200 & 2.1 & 2@xmath130900 & 1.7 & 2@xmath130300 & 1.8 & 2@xmath130300 & 1.9 + cig 154 & 2@xmath1301200 & 1.9 & 1@xmath130600 & 1.5 & 1@xmath130300 & 1.5 & 1@xmath130300 & 1.5 + cig 168 & 1@xmath1301200 & 1.7 & 1@xmath130600 & 1.5 & 1@xmath130300 & 1.5 & 1@xmath130300 & 1.5 + cig 175 & 1@xmath1301200 & 1.8 & 1@xmath130600 & 1.6 & 1@xmath130300 & 1.7 & 1@xmath130300 & 1.7 + cig 180 & 2@xmath1301200 & 1.9 & 2@xmath130600 & 2.0 & 1@xmath130300 & 1.7 & 4@xmath130150 & 1.8 + cig 188 & 2@xmath1301200 & 1.8 & 1@xmath130900 & 1.8 & 2@xmath130300 & 1.7 & 1@xmath130300 & 1.8 + cig 208 & 1@xmath1301200 & 1.8 & 1@xmath130900 & 1.7 & 1@xmath130480 & 1.7 & 1@xmath130480 & 1.8 + cig 213 & 1@xmath1301200 & 1.8 & 1@xmath130600 & 1.7 & 1@xmath130300 & 1.7 & 1@xmath130300 & 1.8 + cig 224 & 1@xmath1301200 & 1.7 & 2@xmath130600 & 1.7 & 1@xmath130300 & 1.8 & 1@xmath130240 & 1.8 + cig 237 & 1@xmath1301200 & 1.8 & 1@xmath130900 & 1.7 & 1@xmath130300 & 1.7 & 1@xmath130300 & 1.8 + cig 309 & 2@xmath130600 & 1.8 & 2@xmath130300 & 1.7 & 2@xmath13060 & 1.7 & 2@xmath13060 & 1.7 + cig 314 & 1@xmath1301200 & 1.8 & 1@xmath130600 & 1.7 & 1@xmath130300 & 1.8 & 1@xmath130300 & 1.8 + cig 434 & 1@xmath1301200 & 1.8 & 1@xmath130900 & 1.7 & 2@xmath130300 & 1.7 & 1@xmath130600 & 1.7 + cig 472 & 1@xmath130600 & 1.8 & 1@xmath130600 & 1.7 & 1@xmath130300 & 1.7 & 1@xmath130300 & 1.7 + cig 518 & 1@xmath130600 & 1.8 & 1@xmath130600 & 1.8 & 1@xmath130300 & 1.8 & 1@xmath130300 & 1.7 + cig 528 & 1@xmath1301200 & 1.9 & 1@xmath130600 & 1.8 & 1@xmath130300 & 1.8 & 1@xmath130300 & 1.9 + cig 549 & 1@xmath130600 & 1.9 & 1@xmath130300 & 1.8 & 2@xmath130120 & 1.7 & 1@xmath130120 & 1.9 + cig 604 & 1@xmath130900 & 1.8 & 1@xmath130600 & 1.8 & 2@xmath130120 & 1.8 & 1@xmath130120 & 1.7 + cig 605 & 1@xmath130900 & 1.9 & 1@xmath130480 & 2.0 & 2@xmath130180 & 2.0 & 1@xmath130180 & 1.9 + cig 691 & 1@xmath130900 & 1.9 & 1@xmath130600 & 2.0 & 1@xmath130300 & 2.0 & 1@xmath130300 & 1.9 + cig 710 & 1@xmath130600 & 1.9 & 1@xmath130600 & 2.0 & 1@xmath130300 & 2.0 & 1@xmath130300 & 1.9 + cig 889 & 1@xmath1301200 & 2.0 & 1@xmath130600 & 2.1 & 1@xmath130300 & 2.0 & 2@xmath130150 & 1.9 + cig 906 & 2@xmath1301200 & 2.3 & 1@xmath130600 & 2.7 & 2@xmath130300 & 2.0 & 2@xmath130300 & 1.8 + cig 910 & 2@xmath1301200 & 2.3 & 1@xmath130600 & 2.0 & 1@xmath130300 & 1.9 & 1@xmath130300 & 1.8 + cig 911 & 2@xmath1301200 & 1.6 & 2@xmath130600 & 1.6 & 1@xmath130300 & 1.5 & 1@xmath130300 & 1.6 + cig 935 & 2@xmath1301200 & 1.6 & 2@xmath130600 & 1.6 & 2@xmath130300 & 1.5 & 2@xmath130300 & 1.6 + cig 976 & 2@xmath1301200 & 1.7 & 2@xmath130600 & 1.8 & 2@xmath130300 & 1.8 & 2@xmath130300 & 1.9 + cig 983 & 2@xmath1301200 & 1.6 & 1@xmath130600 & 1.6 & 1@xmath130300 & 1.5 & 2@xmath130300 & 1.6 + cig 1004 & 2@xmath1301200 & 1.6 & 1@xmath130600 & 1.6 & 1@xmath130300 & 1.5 & 1@xmath130300 & 1.6 + cig 1009 & 2@xmath1301200 & 1.8 & 1@xmath130600 & 1.6 & 1@xmath130300 & 1.9 & 2@xmath130150 & 1.9 + cig 1023 & 1@xmath1301200 & 1.5 & 1@xmath130600 & 1.6 & 1@xmath130300 & 1.6 & 3@xmath130120 & 1.6 + cig 1 & ugc 00005 & 14.08 & sbc & 13.97 & 7243 & + cig 4 & ngc 7817 & 12.74 & sbc & 12.56 & 2391 & + cig 33 & ngc
0237 & 13.72 & sabc & 13.70 & 4136 & + cig 53 & ngc 0575 & 13.76 & sc & 13.45 & 3185 & + cig 56 & ngc 0622 & 14.08 & sb & 13.71 & 5113 & + cig 68 & ngc 0718 & 12.59 & sa & 12.59 & 1683 & + cig 80 & ngc 0772 & 10.29 & sb & 11.09 & 2473 & + cig 103 & ngc 0918 & 13.07 & sc & 13.05 & 1536 & + cig 116 & ngc 1050 & 13.55 & sba & 13.47 & 3989 & + cig 123 & ic 0302 & 13.60 & sbc & 13.59 & 5854 & + cig 138 & ugc 02936 & 14.76 & sc & 15.00 & 3743 & + cig 139 & ngc 1507 & 12.79 & sbd & 12.89 & 769 & + cig 144 & ugc 02988 & 15.52 & sb & 14.90 & 3868 & + cig 151 & ugc 03059 & 15.11 & sd & 14.70 & 4743 & + cig 154 & ugc 03171 & 14.84 & sc & 14.78 & 4475 & + cig 168 & ic 2166 & 13.24 & sabc & 13.20 & 2892 & + cig 175 & ugc 03580 & 13.09 & saba & 12.71 & 1439 & + cig 180 & ngc 2344 & 12.89 & sabb & 12.81 & 1126 & + cig 188 & ugc 03826 & 14.59 & sabc & 14.10 & 1947 & + cig 208 & ugc 04054 & 14.80 & sb & 14.30 & 2175 & + cig 213 & pgc 022141 & 14.79 & sa & 14.88 & 6089 & + cig 224 & ngc 2500 & 12.22 & scd & 12.20 & 693 & + cig 237 & ugc 04277 & 14.82 & sc & 14.90 & 5650 & + cig 309 & ngc 2775 & 11.14 & sab & 11.03 & 1347 & + cig 314 & ngc 2776 & 12.18 & sabc & 12.14 & 2798 & + cig 434 & ugc 05829 & 13.73 & i & 13.73 & 780 & + cig 472 & ngc 3596 & 11.79 & sabc & 11.95 & 1267 & + cig
518 & ngc 4062 & 11.88 & sabc & 11.90 & 939 & + cig 528 & ngc 4357 & 13.25 & sbc & 13.20 & 4367 & + cig 549 & ngc 4651 & 11.38 &
sc & 11.39 & 911 & + cig 604 & ngc 5377 & 12.16 & sa & 12.24 & 2044 & + cig 605 & ngc 5375 & 12.79 & sbab & 12.40 & 2572 & + cig 691 & ngc 5964 & 13.33 & sbcd & 12.60 & 1555 & + cig 710 & ngc 6015 & 11.62 & sc & 11.69 & 1111 & + cig 889 & ngc 6969 & 15.33 & sa & 14.89 & 4778 & + cig 906 & ugc 11723 & 14.74 & sb & 14.70 & 4928 & + cig 910 & ic 5104 & 14.33 & sab & 14.27 & 5110 & + cig 911 & ngc 7056 & 13.67 & sbbc & 13.75 & 5501
& + cig 935 & ngc 7156 & 13.29 & sabc & 13.11 & 4023 & + cig 976 & ngc 7328 & 13.93 & sab & 13.98 & 2886 & + cig 983 & ugc 12173 & 13.56 & sabc & 13.49 & 4960 & + cig 1004 & ngc 7479 & 11.71 & sbbc & 11.60 & 2443 & + cig 1009 & ngc 7514 & 13.55 & sbc & 13.54 & 5005 & + cig 1023 & ugc 12646 & 14.22 & sb & 13.99 & 8143 & + cig1 & 1.59 & 13.92 & 13.32 & 12.40 & 12.08 & 0.60 & 1.51 & 1.83 & + cig4 & 1.62 & 12.86 & 11.94 & 11.30 & 10.27 & 0.92 & 1.56 & 2.59 & + cig33 & 1.53 & 13.65 & 13.06 & 12.53 & 11.85 & 0.59 & 1.12 & 1.80 & + cig53 & 1.53 & 13.65 & 13.02 & 12.51 & 11.81 & 0.63 & 1.14 & 1.83 & + cig56 & 1.59 & 13.96 & 13.30 & 12.71 & 12.00 & 0.66 & 1.26 & 1.96 & + cig68 & 1.58 & 12.52 & 11.67 & 11.10 & 10.29 & 0.84 & 1.42 & 2.23 & + cig80 & 1.59 & 11.65 & 10.80 & 10.23 & 9.38 & 0.85 & 1.42 & 2.27 & + cig103 & 1.58 & 13.28 & 12.37 & 11.68 & 10.75 & 0.92 & 1.60 & 2.53 & + cig116 & 1.55 & 13.71 & 12.81 & 12.25 & 11.48 & 0.90 & 1.46 & 2.23 & + cig123 & 1.59 & 13.81 & 13.00 & 12.33 & 11.60 & 0.81 & 1.47 & 2.20 & + cig138 & 1.59 & 14.60 & 13.33 & 12.51 & 11.28 & 1.27 & 2.09 & 3.31 & + cig139 & 1.60 & 12.89 & 12.36 & 11.90 & 11.45 & 0.52 & 0.98 & 1.43 & + cig144 & 1.59 & 15.31 & 13.95 & 13.03 & 11.63 & 1.36 & 2.28 & 3.68 & + cig151 & 1.57 & 14.96 & 14.07 & 13.34 & 12.37 & 0.89 & 1.62 & 2.59 & + cig154 & 1.52 & 14.62 & 14.05 & 13.48 & 12.93 & 0.58 & 1.15 & 1.69 & + cig168 & 1.50 & 12.68 & 12.01 & 11.45 & 10.86 & 0.66 & 1.22 & 1.82 & + cig175 & 1.50 & 13.01 & 12.57 & 12.08 & 11.50 & 0.44 & 0.92 & 1.51 & + cig180 & 1.53 & 12.99 & 12.14 & 11.57 & 10.80 & 0.85 & 1.41 & 2.19 & + cig188 & 1.50 & 13.20 & 12.70 & 12.25 & 11.99 & 0.50 & 0.94 & 1.20 & + cig208 & 1.50 & 14.58 & 14.01 & 13.55 & 12.86 & 0.58 & 1.03 & 1.72 & + cig213 & 1.55 & 14.74 & 13.80 & 00.00 & 00.00 & 0.94 & 0.00 & 0.00 & + cig224 & 1.50 & 12.06 & 11.62 & 11.34 & 10.85 & 0.44 & 0.72 & 1.20 & + cig237 & 1.50 & 14.39 & 13.53 & 12.95 & 12.27 & 0.86 & 1.44 & 2.11 & + cig309 & 1.50 & 11.31 & 10.39 & 9.72 & 9.09 & 0.91 & 1.58 & 2.30 & + cig314 & 1.50 & 12.26 & 11.69 & 11.26 & 10.77 & 0.56 & 0.99 & 1.49 & + cig434 & 1.50 & 13.31 & 13.09 & 12.84 & 12.85 & 0.21 & 0.46 & 0.45 & + cig472 & 1.50 & 12.01 & 11.49 & 10.99 & 10.45 & 0.51 & 1.01 & 1.55 & + cig518 & 1.50 & 11.90 & 11.21 & 10.62 & 10.03 & 0.69 & 1.28 & 1.90 & + cig528 & 1.50 & 13.36 & 12.68 & 12.14 & 11.84 & 0.67 & 1.21 & 1.51 & + cig549 & 1.50 & 11.65 & 10.92 & 10.42 & 9.84 & 0.73 & 1.22 & 1.80 & + cig604 & 1.50 & 12.39 & 11.46 & 10.87 & 10.25 & 0.92 & 1.51 & 2.13 & + cig605 & 1.50 & 12.78 & 12.08 & 11.65 & 11.65 & 0.70 & 1.13 & 1.14 & + cig691 & 1.50 & 12.45 & 11.82 & 11.48 & 10.89 & 0.63 & 0.97 & 1.56 & + cig710 & 1.50 & 11.69 & 11.14 & 10.61 & 10.08 & 0.56 & 1.08 & 1.61 & + cig889 & 1.57 & 14.47 & 13.52 & 12.84 & 12.07 & 0.95 & 1.63 & 2.40 & + cig906 & 1.43 & 14.89 & 14.09 & 13.44 & 12.61 & 0.81 & 1.46 & 2.29 & + cig910 & 1.43 & 14.81 & 13.82 & 13.25 & 12.49 & 0.99 & 1.57 & 2.32 & + cig911 & 1.58 & 13.86 & 13.09 & 12.48 & 11.75 & 0.77 & 1.38 & 2.11 & + cig935 & 1.56 & 13.41 & 12.72 & 12.24 & 11.53 & 0.69 & 1.17 & 1.88 & + cig976 & 1.53 & 13.99 & 13.10 & 12.55 & 11.46 & 0.89 & 1.44 & 2.53 & + cig983 & 1.57 & 13.81 & 13.03 & 12.28 & 11.64 & 0.78 & 1.53 & 2.18 & + cig1004 & 1.57 & 11.82 & 11.04 & 10.47 & 9.66 & 0.78 & 1.35 & 2.16 & + cig1009 & 1.35 & 13.98 & 13.15 & 12.55 & 11.80 & 0.84 & 1.44 & 2.18 & + cig1023 & 1.54 & 14.08 & 13.26 & 12.71 & 11.73 & 0.82 & 1.37 & 2.35 & + cig1 & 0.36 & 1.28 & 1.39 & -22.11 & -22.47 & -23.39 & -23.50 + cig4 & 0.65 & 1.27 & 2.08 & -21.11 & -21.75 & -22.37 & -23.18 + cig33 & 0.47 & 1.00 & 1.59 & -20.68 & -21.15 & -21.67 & -22.26 + cig53 & 0.54 & 1.01 & 1.64 & -20.03 & -20.57 & -21.04 & -21.67 + cig56 & 0.54 & 1.12 & 1.73 & -20.84 & -21.39 & -21.96 & -22.58 + cig68 & 0.79 & 1.34 & 2.10 & -19.65 & -20.44 & -20.99 & -21.75 + cig80 & 0.64 & 1.15 & 1.84 & -22.00 & -22.64 & -23.16 & -23.85 + cig103 & 0.46 & 0.93 & 1.53 & -20.33 & -20.79 & -21.26 & -21.86 + cig116 & 0.78 & 1.31 & 1.98 & -20.60 & -21.38 & -21.90 & -22.57 + cig123 & 0.54 & 1.10 & 1.62 & -21.95 & -22.49 & -23.04 & -23.57 + cig138 & 0.54 & 1.06 & 1.75 & -22.23 & -22.77 & -23.29 & -23.98 + cig139 & 0.19 & 0.57 & 0.75 & -18.59 & -18.78 & -19.16 & -19.33 + cig144 & 0.53 & 1.11 & 1.89 & -22.05 & -22.58 & -23.16 & -23.95 + cig151 & 0.38 & 0.93 & 1.53 & -21.34 & -21.72 & -22.28 & -22.87 + cig154 & 0.43 & 0.93 & 1.37 & -20.01 & -20.44 & -20.94 & -21.38 + cig168 & 0.44 & 0.93 & 1.35 & -21.33 & -21.77 & -22.26 & -22.68 + cig175 & 0.30 & 0.75 & 1.23 & -19.11 & -19.41 & -19.86 & -20.34 + cig180 & 0.74 & 1.25 & 1.94 & -18.52 & -19.26 & -19.77 & -20.46 + cig188 & 0.40 & 0.80 & 0.97 & -19.45 & -19.84 & -20.25 & -20.42 + cig208 & 0.34 & 0.77 & 1.28 & -18.83 & -19.17 & -19.60 & -20.10 + cig213 & 0.86 & 0.00 & 0.00 & -20.30 & -21.17 & 00.00 & 00.00 + cig224 & 0.40 & 0.65 & 1.10 & -18.12 & -18.52 & -18.77 & -19.22 + cig237 & 0.46 & 1.02 & 1.32 & -21.99 & -22.45 & -23.01 & -23.31 + cig309 & 0.83 & 1.48 & 2.12 & -20.49 & -21.32 & -21.97 & -22.61 + cig314 & 0.42 & 0.83 & 1.21 & -21.33 & -21.75 & -22.16 & -22.54 + cig434 & 0.18 & 0.42 & 0.39 & -17.05 & -17.23 & -17.47 & -17.44 + cig472 & 0.48 & 0.97 & 1.48 & -19.43 & -19.91 & -20.39 & -20.90 + cig518 & 0.56 & 1.16 & 1.65 & -19.32 & -19.88 & -20.48 & -20.97 + cig528 & 0.44 & 0.96 & 1.05 & -21.59 & -22.03 & -22.55 & -22.63 + cig549 & 0.65 & 1.14 & 1.64 & -19.28 & -19.93 & -20.42 & -20.92 + cig604 & 0.79 & 1.38 & 1.86 & -20.59 & -21.39 & -21.98 & -22.45 + cig605 & 0.64 & 1.05 & 0.98 & -20.34 & -20.98 & -21.39 & -21.32 + cig691 & 0.53 & 0.83 & 1.34 & -19.71 & -20.24 & -20.54 & -21.05 + cig710 & 0.41 & 0.94 & 1.33 & -19.91 & -20.32 & -20.85 & -21.24 + cig889 & 0.66 & 1.30 & 1.81 & -21.05 & -21.71 & -22.35 & -22.86 + cig906 & 0.50 & 1.16 & 1.72 & -20.73 & -21.22 & -21.88 & -22.45 + cig910 & 0.68 & 1.17 & 1.65 & -20.97 & -21.64 & -22.14 & -22.62 + cig911 & 0.66 & 1.21 & 1.86 & -21.08 & -21.74 & -22.29 & -22.94 + cig935 & 0.59 & 1.03 & 1.67 & -20.81 & -21.40 & -21.84 & -22.47 + cig976 & 0.67 & 1.17 & 2.09 & -20.12 & -20.79 & -21.30 & -22.22 + cig983 & 0.50 & 1.17 & 1.59 & -21.63 & -22.12 & -22.80 & -23.21 + cig1004 & 0.58 & 1.08 & 1.74 & -21.72 & -22.30 & -22.80 & -23.46 + cig1009 & 0.67 & 1.22 & 1.84 & -20.98 & -21.65 & -22.20 & -22.82 + cig1023 & 0.68 & 1.20 & 2.08 & -21.85 & -22.53 & -23.04 & -23.92 + cig1 & sabbc & sbbc & b & multi & 0.24 + cig4 & sabc & sabc & b & multi & + cig33 & sab(rs)cd & sb(rs)c & b / r & & 0.27 + cig53 & sb(rs)c & sb(r)bc & b / r & multi & 0.59 + cig56 & sb(rs)b & sb(r)b & b / r & & 0.38 + cig68 & sab(s)a & sb(r)a & b / r & & 0.39 + cig80 & sa(s)b & sb(r)b & b / r & multi & 0.20 + cig103 & sab(rs)c & sb(r)c & b / r & multi & 0.19 + cig116 & ( r)sb(s)a & rsb(s)a & b & & 0.42 + cig123 & sb(rs)bc & sb(r)c & b / r & multi & 0.46 + cig138 & sb(s)d & sbc & b & multi & 0 + cig139 & sb(s)m pec & sb(s)m & b & & + cig144 & sb & sabb & b(peanut ? ) & & + cig151 & sadm & sabc & b & multi & + cig154 & sbcd & sb(r)cd & b / r & multi & 0.36 + cig168 & sab(s)bc & sab(s)cd & & multi & + cig175 & sa(s)a pec & sa pec & & & + cig180 & sa(rs)c & sa(r)b & r & multi & + cig188 & sab(s)d & sb(s)d & b & multi & + cig208 & sb & sabcd & b & & + cig213 & s0 & rsb0 & b / r & & 0.39 + cig224 & sb(rs)d & rsbd & b / r & multi & 0.40 + cig237 & sc & sc & & & + cig309 & sa(r)ab & rs(r)a & r & & + cig314 & sab(rs)c & sab(rs)c & b / r & multi & + cig434 & i m & i m & b & & 0.80 + cig472 & sab(rs)c & sa(rs)c & r & multi & + cig518 & sa(s)c & sb(s)bc & b & multi & 0.20 + cig528 & sabc & sa(rs)cd & r & multi & + cig549 & sa(rs)c & sb(rs)c & b / r & multi & 0.31 + cig604 & ( r)sb(s)a & rsb(s)a pec & b(boxy)/r & & 0.22 + cig605 & sb(r)ab & sb(r)b & b / r & multi & 0.44 + cig691 & sb(rs)d & sb(rs)cd & b / r & multi & 0.53 + cig710 & sa(s)cd & sb(s)cd & b & multi & 0.20 + cig889 & sa & sba & b(peanut ) & & + cig906 & sb & sb & & & + cig910 & sbab & sbab & b & & + cig911 & sbb & sb(r)b & b / r & multi & 0.47 + cig935 & sab(rs)cd & sab(rs)c & b / r & multi & + cig976 & sab & sa(s)ab & & multi & + cig983 & sab(rs)c & sb(rs)cd & b / r & multi & 0.30 + cig1004 & sb(s)c & sb(s)c & b & multi & 0.60 + cig1009 & sa & s(r)ab & r & multi & + cig1023 & sb(r)b & rsb(r)b & b / r & & 0.40 + cig1 & 66.8 & [email protected] & [email protected] & [email protected] & + cig4 & 84.1 & [email protected] & [email protected] & [email protected] & + cig33 & 54.1 & [email protected] & [email protected] & [email protected] & + cig53 & 34.6 & [email protected] & [email protected] & [email protected] & + cig56 & 51.4 & [email protected] & [email protected] & [email protected] & + cig68 & 32.1 & [email protected] & [email protected] & [email protected] & + cig80 & 48.5 & [email protected] & [email protected] & [email protected] & + cig103 & 57.5 & [email protected] & [email protected] & [email protected] & + cig116 & 47.0 & [email protected] & [email protected] & [email protected] & + cig123 & 49.7 & [email protected] & [email protected] & [email protected] & + cig138 & 78.7 & [email protected] & [email protected] & [email protected] & + cig139 & 90.0 & [email protected] & [email protected] & [email protected] & + cig144 & 90.0 & [email protected] & [email protected] & [email protected] & + cig151 & 90.0 & [email protected] & [email protected] & [email protected] & + cig154 & 37.0 & [email protected] & [email protected] & [email protected] & + cig168 & 57.1 & [email protected] & [email protected] & [email protected] & + cig175 & 63.0 & [email protected] & [email protected] & [email protected] & + cig180 & 26.0 & [email protected] & [email protected] & [email protected] & + cig188 & 36.5 & [email protected] & [email protected] & [email protected] & + cig208 & 90.0 & [email protected] & [email protected] & [email protected] & + cig224 & 21.7 & [email protected] & [email protected] & [email protected] & + cig237 & 90.0 & [email protected] & [email protected] & [email protected] & + cig309 & 40.6 & [email protected] & [email protected] & [email protected] & + cig314 & 58.3 & [email protected] & [email protected] & [email protected] & + cig434 & 24.5 & [email protected] & [email protected] & [email protected] & + cig472 & 23.2 & [email protected] & [email protected] & [email protected] & + cig518 & 68.2 & [email protected] & [email protected] & [email protected] & + cig528 & 73.6 & [email protected] & [email protected] & [email protected] & + cig549 & 53.0 & [email protected] & [email protected] & [email protected] & + cig604 & 70.6 & [email protected] & [email protected] & [email protected] & + cig605 & 42.8 & [email protected] & [email protected] & [email protected] & + cig691 & 40.7 & [email protected] & [email protected] & [email protected] & + cig710 & 67.4 & [email protected] & [email protected] & [email protected] & + cig889 & 90.0 & [email protected] & [email protected] & [email protected] & + cig906 & 90.0 & [email protected] & [email protected] & [email protected] & + cig910 & 90.0 & [email protected] & [email protected] & [email protected] & + cig911 & 20.9 & [email protected] & [email protected] & [email protected] & + cig935 & 40.0 & [email protected] & [email protected] & [email protected] & + cig976 & 78.3 & [email protected] & [email protected] & [email protected] & + cig983 & 62.0 & [email protected] & [email protected] & [email protected] & + cig1004 & 36.4 & [email protected] & [email protected] & [email protected] & + cig1009 & 48.7 & [email protected] & [email protected] & [email protected] & + cig1023 & 49.7 & [email protected] & [email protected] & [email protected] & + cig1 & 1.11&14.10&13.41&12.81&12.22 & 1.41&14.01&13.39&12.65&12.20 & + cig4 & 1.14&13.24&12.31&11.65&10.75 & 1.44&12.89&11.97&11.33&10.35 & + cig33 & 1.05 & 13.82&13.19&12.67&12.03 & 1.35 & 13.67&13.06&12.54&11.87 & + cig53 & 1.10 & 14.03&13.34&12.83&12.16 & 1.35 & 13.68&13.02&12.52&11.83 & + cig56 & 1.11 & 14.07&13.37&12.80&12.11 & 1.41 & 13.98&13.31&12.73&12.03 & + cig68 & 1.10 & 12.87&12.03&11.48&10.75 & 1.40 & 12.55&11.72&11.16&10.38 & + cig80 & 1.12 & 12.48&11.57&10.96&10.13 & 1.42 & 11.87&11.01&10.41&9.57 & + cig103 & 1.10 & 14.07&13.12&12.44&11.55&1.40 & 13.40&12.48&11.80&10.91 & + cig116 & 1.08 & 13.91&13.04&12.48&11.64 & 1.38 & 13.72&12.83&12.27&11.48 & + cig123 & 1.11 & 14.22&13.31&12.67&11.85 & 1.41 & 13.84&13.01&12.37&11.61 & + cig138 & 1.11 & 15.02&13.80&12.90&11.77 & 1.41 & 14.62&13.40&12.53&11.35 & + cig139 & 1.13 & 13.34&12.80&12.35&11.76 & 1.43 & 12.93&12.38&11.93&11.45 & + cig144 & 1.12 & 15.53&14.18&13.26&12.08 & 1.42 & 15.31&13.996&13.06&11.79 & + cig151 & 1.10 & 15.24&14.32&13.65&12.85 & 1.40 & 15.00&14.16&13.49&12.81 & + cig154 & 1.05 & 14.71&14.09&13.59&12.99 & 1.35 & 14.63&14.05&13.52&12.97 & + cig168 & 1.20 & 13.272 & 12.571 & 11.960 & 11.131&1.50 & 12.623 & 11.980 & 11.386 & 10.536 & + cig175 & 1.20 & 13.418 & 12.861 & 12.361 & 11.668&1.50 & 13.116 & 12.657 & 12.195 & 11.346 & + cig180 & 1.06 & 13.50&12.61&12.03&11.27&1.36 & 13.11&12.24&11.67&10.91 & + cig188 & 1.20 & 13.803 & 13.309 & 12.898 & 12.176&1.50 & 13.137 & 12.68 & 12.255 & 11.475 & + cig208 & 1.20 & 14.62 & 14.007 & 13.593 & 12.935 & 1.50 & 14.275 & 13.851 & 13.478 & 12.772 & + cig224 & 1.20 & 12.73 & 12.212 & 11.775 & 11.085 & 1.50 & 12.025 & 11.501 & 11.127 & 10.098 & + cig237 & 1.20 & 14.825 & 13.849 & 13.17 & 12.279&1.50 & 14.341 & 13.522 & 12.936 & 11.928 & + cig309 & 1.20 & 11.765 & 10.831 & 10.164 & 9.35&1.50 & 11.377 & 10.445 & 9.778 & 8.927 & + cig314 & 1.20 & 12.586 & 12.0 & 11.527 & 10.909&1.50 & 12.244 & 11.711 & 11.245 & 10.626 & + cig434 & 1.20 & 14.036 & 13.787 & 13.417 & 13.024&1.50 & 13.292 & 13.145 & 12.809 & 12.394 & + cig472 & 1.20 & 12.4 & 11.822 & 11.282 & 10.617 & 1.50 & 12.098 & 11.567 & 11.042 & 10.365 & + cig518 & 1.20 & 12.418 & 11.677 & 11.07 & 10.332&1.50 & 11.962 & 11.257 & 10.665 & 9.918 & + cig528 & 1.20 & 13.571 & 12.79 & 12.233 & 11.548&1.50 & 13.336 & 12.645 & 12.135 & 11.463 & + cig549 & 1.20 & 12.002 & 11.235 & 10.676 & 9.978&1.50 & 11.621 & 10.91 & 10.381 & 9.686 & + cig604 & 1.20 & 12.731 & 11.844 & 11.196 & 10.426&1.50 & 12.386 & 11.551 & 10.917 & 10.14 & + cig605 & 1.20 & 13.207 & 12.337 & 11.94 & 11.169&1.50 & 12.778 & 11.833 & 11.555 & 10.446 & + cig691 & 1.20 & 13.273 & 12.609 & 12.095 & 11.474&1.50 & 12.404 & 11.819 & 11.363 & 10.766 & + cig710 & 1.20 & 12.212 & 11.62 & 11.078 & 10.399&1.50 & 11.733 & 11.208 & 10.682 & 10.044 & + cig889 & 1.10 & 14.52&13.53&12.90&12.11&1.40 & 14.50&13.53&12.88&12.08 & + cig906 & 0.96 & 15.26&14.33&13.66&12.80&1.26 & 14.95&14.09&13.44&12.61 & + cig910 & 0.96 & 14.99&14.08&13.44&12.65&1.26 & 14.82&13.89&13.28&12.52 & + cig911 & 1.10 & 13.88&13.16&12.53&11.83&1.40 & 13.86&13.12&12.49&11.76 & + cig935 & 1.08 & 13.561 & 12.866 & 12.375 & 11.716&1.39 & 13.424 & 12.731 & 12.245 & 11.558 & + cig976 & 1.06 & 14.102 & 13.29 & 12.747 & 11.992&1.36 & 13.962 & 13.157 & 12.623 & 11.852 & + cig983 & 1.10 & 14.061 & 13.265 & 12.626 & 11.921&1.40 & 13.825 & 13.051 & 12.336 & 11.681 & + cig1004 & 1.10 & 12.916 & 12.019 & 11.402 & 10.583&1.40 & 12.087 & 11.27 & 10.687 & 9.88 & + cig1009 & 0.88 & 14.276 & 13.421 & 12.853 & 12.135&1.18 & 13.989 & 13.15 & 12.572 & 11.831 & + cig1023 & 1.09 & 14.388 & 13.511 & 12.94 & 12.15&1.39 & 14.124 & 13.279 & 12.724 & 11.833 & + | a release of multicolor broad band ( @xmath0 ) photometry for a subsample of 44 isolated spirals drawn from the _ catalogue of isolated galaxies _ ( cig ) is presented .
total magnitudes and colors at various circular apertures , as well as some global structural / morphological parameters are estimated .
morphology is reevaluated through optical and sharp / filtered @xmath1 band images , ( @xmath2 ) color index maps , and archive near ir @xmath3 images from the two micron survey .
the @xmath4 structural parameters ( concentration , asymmetry , and clumpiness ) were calculated from the images in each one of the bands .
the fraction of galaxies with well identified optical / near ir bars ( sb ) is 63% , while a 17% more shows evidence of weak or suspected bars ( sab ) .
the sample average value of the maximum bar ellipticity is @xmath5 .
half of the galaxies in the sample shows rings .
we identify two candidates for isolated galaxies with disturbed morphology .
the structural @xmath4 parameters change with the observed band , and the tendencies they follow with the morphological type and global color are more evident in the redder bands . in any band , the major difference between our isolated spirals and a sample of interacting spirals is revealed in the @xmath6 plane .
a deep and uniformly observed sample of isolated galaxies is intended for various purposes including ( i ) comparative studies of environmental effects , ( ii ) confronting model predictions of galaxy evolution and ( iii ) evaluating the change of galaxy properties with redshift . |
heavy quark production in hard collisions of hadrons , leptons , and photons has been considered as a clean test of perturbative qcd ( for a recent review see , e.g. , ref @xcite ) .
this process provides not only many tests of perturbative qcd , but also some of the most important backgrounds to new physics processes , which have motivated an extense phenomenology at desy - hera , tevatron and lhc .
these studies are mainly motivated by the strong dependence of the cross section on the behavior of the gluon distribution , which determines the qcd dynamics at high energies .
in particular , the heavy quark photoproduction on nucleon and nuclei targets has been studied in detail in refs .
@xcite , considering the several available scenarios for the qcd dynamics at high energies .
the results of those analysis show that future electron - proton ( nucleus ) colliders at hera and rhic @xcite , probably could determine whether parton distributions saturate and a stringent constraint to the behavior of the gluon distribution in the full kinematical range could be posed . along these lines , in ref .
@xcite we have analyzed the possibility of using ultraperipheral heavy ion collisions ( upc s ) as a photonuclear collider and studied the heavy quark production assuming distinct formalisms for the qcd evolution .
recently , klein and nystrand @xcite have analyzed the quarkonium photoproduction in proton - proton collisions , considering the energetic protons as search of large electromagnetic fields .
in particular , these authors have used the photon spectrum from ref .
@xcite and a photon - proton cross section for quarkonium production obtained by fitting the h1 and zeus data . the main conclusion of that study is that the cross sections are large enough for this channel to be observed experimentally and that this reaction can be used to study the gluon distribution in protons at small values of the bjorken @xmath1 variable .
this achievement motivates the analysis of other processes that are sensitive to the gluon distribution and have larger cross sections . in this paper
, we study the photoproduction of heavy quarks in proton - proton collisions considering distinct theoretical scenarios , which have been analyzed in detail in our previous papers @xcite .
our main motivation comes from the fact that in this process the photon reaches energies higher than those currently accessible at desy - hera .
similar motivation is present in ref .
@xcite , where vector meson photoproduction is investigated . here
, we estimate , for the first time , the total cross section and the rapidity dependence of the photoproduction of heavy quarks in proton - proton collisions , considering the @xmath0-factorization approach taking into account distinct unintegrated gluon distributions .
moreover , based on the color dipole picture @xcite , we compute the total cross section for charm and bottom photoproduction within the iancu - itakura - munier ( iim ) model for the dipole cross section @xcite .
these results will be used as input for our further calculations in proton - proton collisions . for comparison
, we also present the predictions from the collinear factorization approach . in relativistic heavy ion colliders ,
the heavy nuclei give rise to strong electromagnetic fields , which can interact with each other . in a similar way
, these processes also occur when considering energetic protons in @xmath2 colliders .
namely , quasi - real photons scatters off protons at very high energies in the current hadron colliders .
in particular , the heavy quark photoproduction cross section in a proton - proton collision is given by , @xmath3 where @xmath4 is the photon energy in the center - of - mass frame ( c.m.s . ) , @xmath5 is the c.m.s .
photon - proton energy and @xmath6 denotes the proton - proton c.m.s.energy .
the photon spectrum is given by @xcite , @xmath7 \left ( \ln{\omega } - \frac{11}{6 } + \frac{3}{\omega } - \frac{3}{2 \,\omega^2 } + \frac{1}{3 \,\omega^3 } \right ) \
, , \label{eq : photon_spectrum}\end{aligned}\ ] ] with the notation @xmath8 $ ] and @xmath9 \approx ( \omega/ \gamma_l)^2 $ ] , where @xmath10 is the lorentz factor .
the expression above is derived considering the weizscker - williams method of virtual photons and using an elastic proton form factor ( for more detail see refs .
it is important to emphasize that the expression ( [ eq : photon_spectrum ] ) is based on a heuristic approximation , which leads to an overestimation of the cross section at high energies ( @xmath11 at @xmath12 tev ) in comparison with the more rigorous derivation of the photon spectrum for elastic scattering on protons derived in ref .
for a more detailed comparison among the different photon spectra see ref .
@xcite .
other process of interest for heavy quark production is the coherent interaction between two photons . as verified in ref .
@xcite for ultraperipheral heavy ion collisions , the qcd dynamics implies an enhancement of the cross section in comparison with quark - parton model calculations @xcite .
however , its cross section is a factor about @xmath13 smaller than for the one - photon process .
furthermore , differently from two - photon interactions , which leave both photon intact , photoproduction should only dissociate the proton target .
then one way to select photoproduction events is to eliminate events where both protons breakup and requiring a single rapidity gap . in what follows , we briefly present sound models for the heavy quark photoproduction at the photon level , @xmath14 , which enter as input in the calculations for proton - proton collisions in eq .
( [ eq : sigma_pp ] ) .
our results are compared with the desy - hera data @xcite and the prediction from the iim model for this process is presented for the first time . in the last section , the numerical results for the rapidity @xmath15 of the produced states and their total cross sections are shown .
a comparison on the order of magnitude of the cross sections for the distinct approaches is performed .
moreover , we present our main conclusions .
the photon - proton cross section can be calculated considering different theoretical scenarios @xcite . usually , one calculates it assuming the validity of the collinear factorization , where the cross sections involving incoming hadrons are given , at all orders , by the convolution of intrinsically non - perturbative , but universal , quantities - the parton densities , with perturbatively calculable hard matrix elements , which are process dependent . in this approach ,
all partons involved are assumed to be on mass shell , carrying only longitudinal momenta , and their transverse momenta are neglected in the qcd matrix elements .
the heavy quark cross section is given in terms of the convolution between the elementary cross section for the subprocess @xmath16 and the probability of finding a gluon inside the proton , namely the gluon distribution .
the photoproduction cross section at leading order is given by @xcite , @xmath17 \,\ , , \label{sigpho}\end{aligned}\ ] ] where @xmath18 is the invariant mass of the heavy quark pair , with @xmath19 and @xmath20 is the gluon density inside the proton at the factorization scale @xmath21 .
in addition , @xmath22 is the heavy quark mass , @xmath23 is its electric charge and @xmath24 . for the present purpose we will use @xmath25 , with @xmath26 gev and @xmath27 gev . in our further calculations on the collinear approach one takes the gluon distribution given by the grv98(lo ) parameterization @xcite .
it should be noticed that different choices for the factorization scale and quark mass produce distinct overall normalization to the total cross section at photon - nucleon interactions and that nlo corrections can be absorbed in these redefinitions of @xmath21 and @xmath28 . in the next section
we discuss in more detail the dependence of our results in the parton distributions used as input in our calculations .
on the other hand , in the large energy ( small-@xmath1 ) limit , the effects of the finite transverse momenta of the incoming partons become important , and the factorization must be generalized , implying that the cross sections are now @xmath0-factorized into an off - shell partonic cross section and a @xmath0-unintegrated parton density function @xmath29 , characterizing the @xmath0-factorization approach @xcite . the function @xmath30 is obtained as a solution of the evolution equations associated to the dynamics that governs the qcd at high energies ( see @xcite for a review ) .
the latter can recover the usual parton distributions in the double leading logarithmic limit ( dll ) by its integration over the transverse momentum of the @xmath31 exchanged gluon .
the gluon longitudinal momentum fraction is related to the c.m.s .
energy , @xmath32 , in the heavy quark photoproduction case as @xmath33 .
this assumption is a very good approximation , thought the scaling variable @xmath1 in fact depends on the kinematic variables of the incoming particles , namely on @xmath34 and parton momenta ( for details on these issues see @xcite and references therein ) .
the cross section for the heavy - quark photoproduction process is given by the convolution of the unintegrated gluon function with the off - shell matrix elements @xcite .
considering only the direct component of the photon we have that the total cross section reads as @xcite , @xmath35\,\left ( \frac{{\mbox{\boldmath $ p$}}_{1\perp}}{d_1 } + \frac{({\mbox{\boldmath $ k$}}_{\perp}-{\mbox{\boldmath $ p$}}_{1\perp})}{d_2 } \right)^2 + m_q^2 \,\left(\frac{1}{d_1 } + \frac{1}{d_2 } \right)^2 \right\}\ , , \label{sigmakt}\end{aligned}\ ] ] where @xmath36 and @xmath37 .
the transverse momenta of the heavy quark ( antiquark ) are denoted by @xmath38 and @xmath39 , respectively
. the heavy quark longitudinal momentum fraction is labeled by @xmath34 .
the hard scale @xmath40 in general is taken to be equal to the gluon virtuality , in close connection with the blm scheme @xcite . therefore , for our purpose we will use the prescription @xmath41 . as some evolution equations for the unintegrated gluon distribution , for instance the ccfm equation @xcite , enable also its dependence on the scale @xmath40 , we have explicitly written down it in eq .
( [ sigmakt ] ) for sake of generality .
in performing a numerical analysis within the @xmath0-factorization approach ( hereafter labeled semihard ) , we use two distinct parameterizations for the unintegrated gluon distribution ( for details see ref .
first , one considers the derivative of the collinear gluon distribution , quite successful in the proton case @xcite and investigated in the nuclear case in ref .
it simply reads as @xmath42 where @xmath43 is the gluon distribution , which was taken from the grv98(lo ) parton distribution @xcite .
a shortcoming of the function above is that it eventually produces negative values for the unintegrated gluon distribution at large @xmath1 .
this has consequences in the description of the behavior near threshold , being more important for bottom production than for charm .
this is due to the scaling variable to be proportional to the quark mass , @xmath44 , and in general the calculation slightly underestimates the cross section value at low energies .
a more accurate calculation consists on multiplying the collinear gluon function above by the sudakov - like form factor @xmath45 @xcite .
as the heavy quark production is characterized by a scale of order of the quark mass , the charm production , in particular , is considered a primary candidate for investigating the kinematic region of qcd on the boundary between perturbative and high density qcd ( for recent reviews see , e.g. , refs .
recently , we have demonstrated that the inclusive charm total cross section exhibits the property of geometric scaling @xcite , which is one of the main characteristics of the high density approaches .
this fact motivates to estimate these features in the photoproduction case . in order to do so ,
we consider the golec - biernat - wsthoff parameterization @xcite ( hereafter sat - mod ) , which gives the following unintegrated gluon distribution , @xmath46 where @xmath47 is the saturation scale , which defines the onset of the nonlinear qcd ( saturation ) effects .
the parameters were obtained from a fit of the inclusive structure function @xmath48 and total photoproduction cross section data at desy - hera data .
the fit using the expression of the color dipole picture , as explained below , and the saturation ansatz for the dipole cross section produced @xmath49 mb , @xmath50 and @xmath51 .
this fit procedure also includes the charm contribution .
in addition , we have included the large-@xmath1 threshold factor @xmath52 ( for the heavy quark case ) in eq .
( [ gluonsat ] ) to take into account the correct energy dependence at low energies near production threshold .
the saturation model is based on the color dipole picture of the photon - proton interaction @xcite . in the proton rest frame ,
the dis process can be seen as a succession in time of three factorisable subprocesses : i ) the photon fluctuates in a quark - antiquark pair with transverse separation @xmath53 long after the interaction , ii ) this color dipole interacts with the proton target , iii ) the quark pair annihilates in a virtual photon .
the interaction @xmath54 is further factorized in the simple formulation @xcite , @xmath55 where @xmath34 is the longitudinal momentum fraction of the quark .
the photon wavefunctions @xmath56 are determined from light cone perturbation theory and read as @xmath57\ , \varepsilon^2 \,k_1 ^
2(\varepsilon \,{\mbox{\boldmath $ r$ } } ) + \ , m_f^2 \ , \,k_0 ^ 2(\varepsilon\,{\mbox{\boldmath $ r$ } } ) \right\}\label{wtrans } \nonumber \\
|\psi_{l}^f|^2 & = & \ ! \frac{6\alpha_{\mathrm{em}}}{\pi^2 } \ , e_f^2 \ , \left\{q^2 \,z^2 ( 1-z)^2 \,k_0 ^ 2(\varepsilon\,{\mbox{\boldmath $ r$ } } ) \right\ } , \label{wlongs } \end{aligned}\ ] ] where the auxiliary variable @xmath58 depends on the quark mass , @xmath59 .
the @xmath60 are the mcdonald functions and the summation is performed over the quark flavors . the dipole - hadron cross section @xmath61 contains all information about the target and the strong interaction physics . in general , the saturation models @xcite interpolate between the small and large dipole configurations , providing color transparency behavior , @xmath62 , as @xmath63 and constant behavior at large dipole separations @xmath64 . along these lines ,
the phenomenological saturation model resembles the mains features of the glauber - mueller resummation .
namely , the dipole cross section takes the eikonal - like form , @xmath65\ , .
\label{gbwdip}\end{aligned}\ ] ] its phenomenological application has been successful in a wide class of processes with a photon probe .
an important aspect of the saturation models is that they resume a class of higher twist contributions which should be non - negligible towards the low @xmath66 region @xcite .
moreover , it is important to emphasize that the dipole approach was extended for heavy quark production in proton - proton ( nucleus ) collisions in refs .
@xcite , and its equivalence with the collinear approach as well as the comparison with @xmath67 data was presented in ref .
@xcite .
although the saturation model describes reasonably well the hera data , its functional form is only an approximation of the theoretical non - linear qcd approaches . on the other hand , an analytical expression for the dipole cross section
can be obtained within the bfkl formalism .
currently , intense theoretical studies has been performed towards an understanding of the bfkl approach in the border of the saturation region @xcite . in particular , the dipole cross section has been calculated in both lo and nlo bfkl approach in the geometric scaling region @xcite .
it reads as , @xmath68^{\gamma_{\mathrm{sat}}}\,\exp\left [ -\frac{\ln^2\,\left({\mbox{\boldmath $ r$}}^2 q_{\mathrm{sat}}^2\right)}{2\,\beta \,\bar{\alpha}_sy}\right]\ , , \label{sigmabfkl}\end{aligned}\ ] ] where @xmath69 ( @xmath70 is the proton radius ) is the overall normalization and the power @xmath71 is the ( bfkl ) saddle point in the vicinity of the saturation line @xmath72 ( the anomalous dimension is defined as @xmath73 ) . as usual in the bfkl formalism , @xmath74 , @xmath75 and @xmath76 .
the quadratic diffusion factor in the exponential gives rise to the scaling violations .
the dipole cross section in eq .
( [ sigmabfkl ] ) does not include an extrapolation from the geometric scaling region to the saturation region .
this has been recently implemented in ref .
@xcite , where the dipole amplitude @xmath77 was constructed to smoothly interpol between the limiting behaviors analytically under control : the solution of the bfkl equation for small dipole sizes , @xmath78 , and the levin - tuchin law @xcite for larger ones , @xmath79 .
a fit to the structure function @xmath80 was performed in the kinematical range of interest , showing that it is not very sensitive to the details of the interpolation .
the dipole cross section was parameterized as follows , @xmath81\ , , & \mbox{for $ { \mbox{\boldmath $ r$}}q_{\mathrm{sat}}(x ) > 2$}\ , , \end{array } \right .
\label{cgcfit}\end{aligned}\ ] ] where the expression for @xmath82 ( saturation region ) has the correct functional form , as obtained either by solving the balitsky - kovchegov ( bk ) equation @xcite , or from the theory of the color glass condensate ( cgc ) @xcite .
hereafter , we label the model above by cgc .
the coefficients @xmath83 and @xmath84 are determined from the continuity conditions of the dipole cross section at @xmath85 .
the coefficients @xmath86 and @xmath87 are fixed from their lo bfkl values . in our further calculations
it will be used the parameters @xmath88 fm , @xmath89 , @xmath90 and @xmath91 , which give the best fit result .
we have included also a large-@xmath1 factor as for the saturation model . in the color dipole picture ,
the heavy quark photoproduction cross section using the cgc model reads as , @xmath92 where the longitudinal piece does not contribute as @xmath93 .
the transverse contribution is computed using eq .
( [ wtrans ] ) and introducing the appropriated mass and charge of the charm or bottom quark .
the dipole approach is very useful in providing semi - analytic solutions for the cross section . in the heavy quark case
, it can be shown that the scattering process is dominated by small size dipoles with mean value @xmath94 . in this scenario ,
the photon wavefunction selects the color transparency region as the main contribution to the cross section .
concerning the longitudinal fraction @xmath34 of the dipoles , at the limit @xmath95 only the symmetric configurations @xmath96 contribute . in this case , which it is obeyed in the photoproduction regime , we can write in a semi - quantitative way , @xmath97 where @xmath98 is the small-@xmath99 limit of the dipole cross section . for heavy quark production
we have that the saturation model predicts @xmath100 , whereas for the cgc model it reads as @xmath101 modulo the diffusion term in eq .
( [ cgcfit ] ) . therefore , by introducing these values in eq .
( [ semianalytic ] ) one produces the following analytical results for both models , @xmath102 which implies in general a slightly smoother energy growth for cgc than for sat - mod since @xmath103 .
it should be noticed that this behavior is obtained only in the case of sufficiently small @xmath1 ( large rapidity y ) , where we could neglect the diffusion term for cgc and take only the anomalous dimension at the saturation vicinity . on the other hand , for not so small @xmath1 the diffusion factor
can not be disregarded and the effective anomalous dimension @xmath104 $ ] in eq .
( [ cgcfit ] ) should be taken .
this situation can occur in the bottom production at intermediate energies due to the large bottom mass , which implies the cgc having practically the same slope on energy as the saturation model .
this can be understood in realizing that the effective anomalous dimension for sat - mod is @xmath105 and at sufficiently large @xmath1 they are very close of each other .
having presented the main sound approaches for heavy quark photoproduction at the photon level , let us compare their numerical results with the experimental desy - hera data @xcite . they are shown in fig .
[ fig1 ] , where the following notation is considered : the results from the saturation model ( sat - mod ) are denoted by the solid lines and the usual collinear factorization calculation is labeled by the dot - dashed curves .
the @xmath0-factorization formalism ( semihard ) , using the unintegrated gluon function in eq .
( [ ugfdll ] ) produces the long - dashed curves , where the iancu - itakura - munier parameterization for the dipole cross section ( cgc ) gives the dashed curves . in all calculations we have
use the same quark masses @xmath26 gev and @xmath27 gev .
we quote refs .
@xcite for detailed investigations on the dependence of the cross section in choosing different collinear gluon parameterizations and different factorization scale for both the collinear and @xmath0-factorization approaches .
lets compare the distinct behaviors on energy and overall normalizations . both the collinear approach and the semihard formalism
give a consistent data description at high energies .
at the region near the threshold , the semihard approach slightly underestimate the cross section , being more pronounced in the bottom case .
this is understood in terms of the considered unintegrated gluon distribution @xmath106 , as discussed before .
the saturation model underestimates the cross section at high energies by a factor about 2 , producing a reasonable description of the region near threshold .
the cgc model follows similar trends , but the high energy values are closer to the experimental measurements than the saturation model .
the slope on energy for sat - mod and cgc can be easily understood from the semi - quantitative results in eq .
( [ sigtot_analytic ] ) . for the bottom case ,
we are not in sufficiently small @xmath1 and sat - mod and cgc give similar energy slope in agreement with our discussion above
. unfortunately , the current precision and statistics of the experimental measurements of the photoproduction cross section are either low to formulate definitive conclusions about the robustness of the different approaches presented here .
more precise measurements could be pose stringent constraints on the energy dependence and overall normalization . finally , it should be noticed that the present calculations concern only the direct photon contribution to the cross section , whereas the resolved component has been neglected . in some extent the results from the saturation models presented here
let some room for this contribution .
details on its calculation and size of its contribution can be found , for instance , in ref .
in what follows , we will compute the rapidity distribution and total cross sections for the photoproduction of open charm and bottom from proton - proton collisions at high energies .
the approaches shortly reviewed in the previous section serve as input for the numerical calculations using eq .
( [ eq : sigma_pp ] ) for the energies of the current and future @xmath67 and @xmath107 accelerators .
namely , one considers the shorter @xmath67 running at rhic upon energy of @xmath108 gev and the tevatron value @xmath109 tev for its @xmath107 running . for the planned lhc
@xmath67 running ( atlas and/or cms ) one takes the design energy @xmath110 tev .
the distribution on rapidity @xmath15 of the produced open heavy quark state can be directly computed from eq .
( [ eq : sigma_pp ] ) , by using its relation with the photon energy @xmath4 , i.e. @xmath111 . a reflection around @xmath112 takes into account the interchanging between the proton s photon emitter and the proton target .
explicitly , the rapidity distribution is written down as , @xmath113}{dy } = \omega \ , \frac{n_{\gamma } ( \omega ) } { d\omega } \,\sigma_{\gamma p \rightarrow q\overline{q}x}\,\left(\omega \right)\,.\end{aligned}\ ] ] lets start by analyzing the predictions from the collinear approach , obtained using distinct gluon pdf s . in fig . [ figcomparacao ] ,
the comparison among the resulting rapidity distributions for open charm photoproduction at lhc energy is presented .
it has been considered the grv ( 94 and 98 ) and mrst ( 2001 ) parton parameterizations . the results for bottom and/or charm at rhic / tevatron energies give somewhat very close curves and they will be not presented here .
the results presented in fig .
[ figcomparacao ] are strongly dependent on the gluon distribution , which motivates the study of this process in order to constrain it .
moreover , they can useful in understanding what is the underlying qcd dynamics at high energies .
we have that the deviations between the predictions from the grv98(lo ) and mrst2001(lo ) pdf s are approximately of 30 % . in order to demonstrate the strong dependence of the rapidity distribution on the gluon density
, we also present the result obtained from the obsolete grv94(lo ) pdf , which it has a steeper gluon distribution at small @xmath1 . in our further comparisons with other approaches
, we will use the grv98 pdf .
the resulting cross sections coming out of the distinct theoretical inputs considered in previous section are depicted in figs .
( [ fig2]-[fig4 ] ) at rhic , tevatron and lhc energies , respectively .
lets discuss in general lines the present results . at rhic energies , larger values of @xmath114
are probed in the process than at tevatron and lhc , mostly for bottom .
this fact explains the very similar result for all approaches in the bottom case in fig .
[ fig2 ] ( right plot ) . for the charm case ,
the situation starts to be different .
the semihard result presents a sharper tail at large rapidities as a consequence of its behavior on energy near threshold , as discussed before .
the overall normalization and its behavior at central rapidities follow the same trend coming from the cross section at photon level ( see fig . [ fig1 ] ) .
collinear and @xmath0-factorization formalisms often give closer results , whereas the saturation models ( sat - mod and cgc ) produce smaller cross sections . at tevatron energies ( fig .
[ fig3 ] ) , smaller @xmath1 are being probed in the reaction , and the cross section becomes more dependent on the high energy behavior of the cross section at photon level .
the overall normalization becomes increasingly sizeable for different approaches .
for the charm case , the result could already distinguish between the saturation approaches and the usual pqcd calculations . at lhc energies ( fig .
[ fig4 ] ) , the separation is more clear once smaller @xmath1 are probed .
high values are obtained from the collinear approach due to its steeper energy growth , followed by the semihard approach .
the latter still considers an unintegrated gluon function dependent on the collinear one and therefore has closer behavior on energy .
concerning the saturation approaches , sat - mod provides the lower limit , whereas cgc gives somewhat larger values .
this is once again a consequence of their high energy behavior , as verified in fig .
[ fig1 ] .
let us now compute the integrated cross section considering the distinct qcd approaches .
the results are presented in table [ tabhq ] , for the open charm and bottom pair production at rhic , tevatron and lhc , respectively .
the collinear factorization approach gives the largest rates among the models studied , followed by the semihard formalism , as a clear trend from the distribution on rapidity .
concerning the saturation models , the cgc results in a cross section of order 20 % larger than sat - mod .
the values are either large at tevatron and lhc , going from some units of @xmath40bs at rhic to hundreds of @xmath40bs at lhc .
therefore , these reactions can have high rates at the lhc kinematical regime . as stated before
, we do nt consider the resolved photon contribution in our calculations .
however , it is possible to present an estimate of this process considering the results from refs .
@xcite for ultraperipheral heavy ion collisions , where this contribution was studied using collinear factorization .
one of the main results is that these contributions are @xmath115 and @xmath116 of the total charm and bottom photoproduction cross sections at lhc energy , respectively . on the other hand , in ref .
@xcite this contribution was estimated to be of order of 20 - 30 @xmath117 of the direct photon cross section using @xmath0-factorization .
it is important to emphasize that the inclusion of resolved photon contribution brings the predictions for bottom production closer to the @xmath118 data , but a major inconsistency can not be claimed due to the large experimental errors and theoretical uncertainties . we postpone a detailed analysis on these issues for a future publication . the cross section computed here can be contrasted with the open charm hadroproduction in central @xmath2 collisions . recently ,
raufeisen and peng @xcite have computed it considering the nlo parton model and the color dipole formulation .
they found that the results are subject to uncertainties coming from different choices for quark mass and parameters of the models , but they are able to describe all data ( except recent star measurement ) inside the uncertainty band .
the photoproduction cross section is of order 0.1 % from the corresponding hadroproduction .
however , as we will see below , the production rates are either high and the experimental signal is significantly clear . at rhic , where the luminosity is assumed to be @xmath119 @xmath120s@xmath121 , the open charm rate ranges on @xmath122 events by year .
it should be noticed that rhic uses most part of the running in the heavy - ion mode , so we used @xmath123 s in the last estimation .
for the bottom case , the rates are @xmath124 events by year . at tevatron , assuming the the running time @xmath125 s and design luminosity @xmath126 @xmath120s@xmath121 , we have for charm @xmath127 and for bottom @xmath128 events / year .
the lhc produces the greatest rates ( @xmath129 @xmath120s@xmath121 ) , giving for charm @xmath130 and for bottom @xmath131 events / year .
notice the large rate for bottom at lhc .
finally , lets discuss the experimental separation of this reaction channel . as emphasized in ref .
@xcite , although the photoproduction cross section to be a small fraction of the hadronic cross section , the separation of this channel is feasible if we impose the presence of a rapidity gap in the final state .
it occurs due to the proton which is the photon emitter remains intact in the process . similarly to the @xmath132 case
, we expect that a cut in the transverse momentum of the pair could eliminate most part of the contribution associated to the hadroproduction of heavy quarks .
moreover , in comparison with the hadroproduction of heavy quarks , the event multiplicity for photoproduction interactions is lower , which implies that it may be used as a separation factor between these processes . as stated in ref .
@xcite , one way to select photoproduction events is to eliminate events where both protons breakup .
this should eliminate almost all of the hadroproduction events while retaining most of the photoproduction interactions . in ref .
@xcite the rejection factor @xmath133 , which is the probability of finding a rapidity gap with width @xmath15 in a @xmath67 collision has been calculated . for photoproduction , the authors have obtained @xmath134 at rhic ( lhc ) , requiring a single rapidity gap and @xmath135 ( for details see section vi from ref .
these estimates can be directly applied in our analysis .
as photoproduction always leads to a rapidity gap , this requiriment should reject relatively few signal events , leading to a good signal to noise ratio for selecting these events .
an important background which it should be analyzed is the diffractive heavy quark production in the single diffraction process @xcite .
we postpone this study for a future publication . in summary ,
we have computed the cross sections for photoproduction of open heavy quarks in @xmath67 and @xmath107 collisions .
this has been performed using well established qcd approaches , namely the collinear and semihard factorization formalisms as well as saturation models within the color dipole approach .
for the first time , quantitative predictions for these approaches are presented .
the obtained values are shown to be sizeable at the current accelerators energies ( rhic and tevatron ) and are increasingly larger at lhc .
the feasibility of detection of these reactions is encouraging , since their experimental signature should be suitably clear .
furthermore , they enable to constraint already in the current colliders the qcd dynamics since the main features from photon - proton collisions hold in proton - proton collisions .
the authors are grateful to professor bernd kniehl ( _ ii .
institut fr theoretische physics , universitt hamburg _ ) for calling our attention to the ref .
@xcite and for his comments related to the equivalent - photon distribution .
one of us ( m.v.t.m . ) thanks the support of the high energy physics phenomenology group at the institute of physics , gfpae if - ufrgs , porto alegre .
this work was partially financed by the brazilian funding agencies cnpq and fapergs .
j. breitweg _
[ zeus collaboration ] , eur .
j. c * 6 * , 67 ( 1999 ) ; + c. adloff
_ et al . _
[ h1 collaboration ] , nucl .
b * 545 * , 21 ( 1999 ) ; + j. breitweg _
[ zeus collaboration ] , eur .
j. c * 12 * , 35 ( 2000 ) ; + j.j .
[ european muon collaboration ] , phys .
b * 106 * , 419 ( 1981 ) ; + c. adloff _ et al . _
[ h1 collaboration ] , phys .
b * 467 * , 156 ( 1999 ) [ erratum - ibid .
b * 518 * , 331 ( 2001 ) ] ; + j. breitweg _
[ zeus collaboration ] , eur .
j. c * 18 * , 625 ( 2001 ) .
lipatov , v.a .
saleev and n.p .
zotov , mod .
a * 15 * , 1727 ( 2000 ) ; + y.m .
shabelski and a.g .
shuvaev , arxiv : hep - ph/0107106 ; + m.g .
ryskin , a.g .
shuvaev and y.m .
shabelski , phys . atom .
nucl . * 64 * , 1995 ( 2001 ) ; + s.p .
baranov _ et al .
_ , eur .
j. c * 24 * , 425 ( 2002 ) ; + h. jung , phys .
d * 65 * , 034015 ( 2002 ) .
l. d. mclerran and r. venugopalan , phys .
d * 49 * , 2233 ( 1994 ) ; e. iancu , a. leonidov , l. mclerran , nucl .
a * 692 * , 583 ( 2001 ) ; e. ferreiro , e. iancu , a. leonidov , l. mclerran , nucl .
a * 703 * , 489 ( 2002);j .
jalilian - marian , a. kovner , l. mclerran and h. weigert , phys .
d * 55 * , 5414 ( 1997 ) ; j. jalilian - marian , a. kovner and h. weigert , phys .
d * 59 * , 014014 ( 1999 ) , _ ibid .
_ * 59 * , 014015 ( 1999 ) , _ ibid . _
* 59 * 034007 ( 1999 ) ; a. kovner , j. guilherme milhano and h. weigert , phys
d * 62 * , 114005 ( 2000 ) ; h. weigert , nucl . phys . * a703 * , 823 ( 2002 ) .
* collinear * & * cgc * + & @xmath136 & 377 nb & 687 nb & 782 nb & 492 nb + & @xmath137 & 3.7 nb & 3.6 nb & 4.3 nb & 4.2 nb + & @xmath136 & 1.04 @xmath40b & 2.77 @xmath40b & 3.21 @xmath40b & 1.36 @xmath40b + & @xmath137 & 13.3 nb & 24.0 nb & 29.2 nb & 19.8 nb + & @xmath136 & 3.54 @xmath40b & 14.2 @xmath40b & 16.7 @xmath40b & 4.37 @xmath40b + & @xmath137 & 55.0 nb & 182 nb & 236 nb & 108 nb + | we calculate the photoproduction of heavy quarks in proton - proton collisions at rhic , tevatron and lhc energies , where the photon reaches energies larger than those ones accessible at desy - hera . the integrated cross section and the rapidity distributions for open charm and bottom production
are computed employing sound high energy qcd formalisms . for the linear pqcd
approaches we consider both the usual collinear factorization and the @xmath0-factorization formalisms , whereas for the nonlinear qcd ( saturation ) calculations one considers the golec - biernat - wsthoff and the iancu - itakura - munier parameterizations for the dipole cross section within the color dipole picture . |
the study of non - hubble motions and of the peculiar velocity field of galaxies on different scales forms an observational foundation for many tasks of present - day cosmology ( see the survey of strauss and willick , 1995 ) . for detailed analysis of peculiar velocities ample samples of galaxies with measured radial velocities @xmath0 and independent distance estimates @xmath1 are required .
the first homogeneous catalogue of this kind , markiii , was compiled as a result of integration and unification of data of observations of radial velocities and rotational velocities ( in the radio or optical ranges ) of galaxies from different samples , which were made by different groups of authors in 19821996 ( willick et al . , 1995 , 1997b ) .
it was based on mutual coordination of the tf relations ( tully fisher , 1977 ) , which were represented in the form `` luminosity line width '' for the samples of spiral galaxies of clusters and of the field .
the sample of spirals ( han and mould , 1992 ) was adopted as the reference one . the @xmath2 samples were later also included in markiii , and relative zero points for the arrays of e and s galaxies ( dekel et al . , 1999 ) were determined . without going into details of constructing markiii , which were described comprehensively by the authors ,
let us note that standardization of the selection criterion was an important part of the work since the initial observational samples incorporated galaxies from different optical catalogues with different limits in apparent magnitude or angular diameter and having different angles of inclination to the line of sight .
notice that strongly inclined , `` flat '' spirals were not originally included in markiii .
the total number of individual galaxies involved in the final version of markiii amounts to 3000 ; they were divided into 1200 groups .
thus , the catalogue markiii is important for it contains homogeneous data for early- and late - type galaxies , both cluster members and field galaxies and distributed practically throughout the sky .
this allows a map of matter density distribution to be constructed from the distribution of peculiar velocities , and the values of cosmological parameters to be obtained .
there is another approach to establishing a homogeneous sample of galaxies for studying the non - hubble motions of galaxies .
it was realized when compiling the catalogue of flat edge - on galaxies , rfgc ( karachentsev et al .
, 1999a ; the first version is fgc , karachentsev et al .
the main idea of this approach is a special all - sky search for late edge - on spiral galaxies and selection for the catalogue of objects satisfying the conditions @xmath3 and @xmath4 , where @xmath5 and @xmath6 are the major and minor axes .
the rfgc catalogue contains 4236 galaxies and covers the entire sky .
since the selection was performed using the surveys poss - i and eso / serc , which have different photometric depth , we reduced the diameters of the southern - sky galaxies to the system poss - i , which turned out to be close to the system @xmath7 .
the substantiation of selecting exactly flat galaxies and a detailed analysis of optical properties of the catalogue objects are available in the texts of fgc , rfgc and in the papers of the authors cited therein .
some extensive work has so far been done on measurements of radial velocities , hi 21 cm line widths , @xmath8 , or rotational curves @xmath9 for the galaxies of the rfgc catalogue .
besides , we have collected such data from literature .
part of the data have been used to derive the radial velocities and direction of bulk motion of flat galaxies ( karachentsev et al .
, 1995 , 1999b ) .
herein we report the radial velocities , hi line widths and peculiar velocities for 1327 galaxies of the rfgc catalogue .
observational data were divided into several samples . 1 . the observations of flat galaxies from fgc performed with the 305 m telescope at arecibo ( giovanelli et al . ,
the observations are confined within the zone @xmath10 accessible to the radio telescope .
there was no selection by the visible angular diameter , type , axes ratio and other characteristics .
we have not included in the summary the flat galaxies from the supplement to fgc , which do not satisfy the condition @xmath3 , and also the galaxies with uncertain values of @xmath8 , in accordance with the notes in the paper by giovanelli et al .
our list contains 490 flat galaxies from this paper .
2 . the observations of optical rotational curves made with the 6 m telescope of sao ras ( makarov et al . , 1997a , b ; 1999 ; 2000 ) .
the objects located in the zone @xmath11 , with the axes ratio @xmath12 and a large diameter @xmath13 were selected for the observations .
the maximum rotational velocities were converted to @xmath8 by a relation derived through comparison of optical and radio observations of 59 galaxies common with sample `` 1 '' ( makarov et al . , 1997a ) .
300 galaxies from these papers are included into our list .
3 . the data on radial velocities and hydrogen line widths in the fgc galaxies identified with the rc3 catalogue ( de vaucouleurs et al . , 1991 ) . in a few cases , where only @xmath14 are available in rc3 , they were converted to @xmath8 according to karachentsev et al . ( 1993 ) .
this sample comprises flat galaxies all over the sky , a total of 167 objects .
4 . the data on hi line widths ( 64 m radio telescope , parkes ) and on optical rotational curves @xmath9 ( 2.3 m telescope of siding spring ) for the flat galaxies identified with the lists by mathewson et al .
( 1992 ) , mathewson and ford ( 1996 ) .
the optical data were converted to the widths @xmath8 according to mathewson and ford ( 1996 ) .
the sb sd galaxies from the catalogue eso / uppsala ( lauberts , 1982 ) with angular dimensions @xmath15 , inclinations @xmath16 , and a galactic latitude @xmath17 have been included in the lists . as mathewson et al .
( 1992 ) report , the data obtained with the 64 m and 305 m telescopes are in good agreement .
our sample contains 177 flat galaxies from these papers .
the hi line observations of flat galaxies carried out by matthews and van driel ( 2000 ) using the radio telescopes in nancay ( @xmath18 ) and green bank ( @xmath19 ) .
they have selected the flat galaxies from fgc(e ) with angular dimensions @xmath20 , of scd types and later , mainly of low surface brightness ( sb = iii and iv according to rfgc ) .
we did not include in our list uncertain measurements from the data of matthews and van driel ( 2000 ) . in the case of common objects with samples `` 1 '' or `` 2 ''
we excluded the data of matthews and van driel ( 2000 ) on the basis of the comparisons of @xmath0 and @xmath8 .
the subsample `` 5 '' comprises 193 galaxies .
thus , each galaxy enters into our sample with a single set of @xmath0 and @xmath8 estimates .
we have included in the list a total of 1327 flat galaxies .
the heliocentric velocities were reduced to the system cmb , @xmath21 according to kogut et al .
( 1993 ) , and to the centroid of the local group , @xmath22 ( karachentsev and makarov , 1996 ) .
the observed widths were corrected for the cosmological broadening and turbulence ( tully and fouqu , 1985 ) .
the angular diameters were corrected for the extinction in the galaxy and intrinsic absorption ( karachentsev , 1991 ) .
the mean characteristics of the galaxies are collected in table1 .
the distribution of the catalogue blue diameters as a function of @xmath21 for 1327 flat galaxies is shown in fig.1 .
it can be seen that the diameters of galaxies are close to the limiting catalogue value , @xmath23 , and this holds practically for all ( but for the smallest ) distances .
thus the selection by angular diameters in our list is not strong , although individual samples have different depths ( table1 ) and are differently located on the lg@xmath24 diagram ( we do not present the figures here ) .
.mean sample characteristics of flat galaxies [ cols="^,^,^,^,^ " , ] the distribution of 1327 flat galaxies over the sky in the galactic coordinates is displayed in fig.2 .
one can see that they cover the sky fairly uniformly except the milky way , with some excess in the number of galaxies north of @xmath25 .
before description of determination of the distance to a rfgc galaxy independent upon the radial velocity , we will make some general remarks . for each of the galaxies included in our list
, we have a homogeneous set of catalogue parameters : 1 ) measured `` blue '' and `` red '' angular diameters @xmath26 in a system close to @xmath7 ; 2 ) hubble type ty ( 5 corresponds to type sc ) ; 3 ) surface brightness index sb ( brightness decreases from i to iv ) etc .
we used the tf relation in the form `` linear diameter
width '' and therefore , in distinction to the markiii compilers , did not need to coordinate apparent magnitudes observed in different photometric bands and with different apertures . the data on radial velocities and widths were taken from different lists , and in the case of `` duplicates '' we took account of the data of only one source .
each galaxy entered into our list with the equal weight .
the second principal distinction of our list from markiii is that the distances to galaxies were determined not by coordination of the tf relations for individual samples , but from the generalized tf relation common to all galaxies .
karachentsev et al .
( 2000 ) describe in detail the procedure of determining the distances to galaxies from the generalized tully - fisher relation . the distance @xmath27 ( expressed in km / s and corresponding to the radial velocity of a galaxy in the case of purely hubble expansion )
was represented as a linear combination of functions , dependent on the galaxy characteristics , with coefficients @xmath28 . using the fisher criterion the statistical significance was determined for each regressor .
the regressors , which entered into the approximation @xmath29 with a confidence level of less than 99.9% , were rejected .
we use in this paper the distance approximation @xmath30 it is implied here the corrected diameters and widths ; @xmath31 , and @xmath32 . as compared to our paper of 2000 , the summand , which included
the axes ratio , has been discarded in this approximation as statistically insignificant , and a new statistically significant regressor @xmath33 is added .
considering the model of bulk motion of galaxies to be the most simple dipole approximation , we represented the measured radial velocity of each galaxy @xmath21 in the relic radiation rest frame as the sum of three components : r , the projection of the ordered ( bulk ) motion , @xmath34 and a random small - scale component @xmath35 . using the least - square method , we derived the @xmath28 coefficients of the generalized tully - fisher relation and the values of @xmath36 ( in km / s ) with which the sum of squares of the deviations @xmath37 is minimized .
for all the galaxies of the list peculiar velocities were computed @xmath38 having excluded 56 galaxies with peculiar velocities over @xmath39 and having introduced a restriction @xmath40km / s to make the sample more homogeneous in depth ( see fig.1 ) , we derived a generalized tf relation with coefficients : @xmath41 @xmath42 @xmath43 @xmath44 this relation was again applied to the computation of peculiar velocities of all 1327 galaxies .
they are presented in table2 .
the content of the columns is as follows : + ( 1 ) , ( 2 ) the number of the galaxy in the rfgc and fgc catalogues , respectively ; + ( 3 ) the right ascension and declination for the epoch 2000.0 ; + ( 4 ) , ( 5 ) the corrected `` blue '' major and minor diameters , in arcmin ; + ( 6 ) the corrected line width @xmath45 in km / s ; + ( 7 ) the observed heliocentric radial velocity , in km / s ; + ( 8) the radial velocity reduced to the centroid of the local group , in km / s ; + ( 9 ) the radial velocity in the system of @xmath46 cosmic microwave radiation , in km / s ; + ( 10 ) the distance ( in km / s ) measured from the basic regression on the assumption that the model of motion of galaxies is dipole ; + ( 11 ) the dipole component of the radial velocity , in km / s ; + ( 12 ) the peculiar velocity , in km / s ; + ( 13 ) the number of the sample from which the original data @xmath0 and @xmath8 were taken .
fig.3 shows @xmath47 against @xmath1 .
the relation between the predicted distance @xmath1 and the measured radial velocity @xmath21 is shown in fig.4 for 1271 flat galaxies .
the scatter of points corresponds to @xmath48km / s .
after the exclusion of 56 `` outliers '' the positive and negative radial velocities on the `` peculiar velocity distance '' relationship are located symmetrically about zero and do not exhibit noticeable variation of slope with distance .
the problem of refinement of distances independent of their radial velocities ( in the case of spiral galaxies this is the improvement of the tf relation ) remains the most important , any samples being used . in many papers of the past few years , necessity is shown of taking into account the contribution made by both the surface brightness and the galaxy type to tf relation .
basing on detailed photometric investigation of spiral galaxies in clusters , willick ( 1999 ) would solve this problem in a more refined way .
he derived not strong but statistically significant dependence between tf relation and surface brightness , and also between tf and galaxy light concentration index `` c '' . in our case the value of @xmath49 , which characterizes the variation of the disk
bulge relationship along the hubble sequence , may be a rough analog of the index `` c '' .
it seems promising to examine models of bulk motion of galaxies not only in the most simple dipole approximation , but also with involvement of the quadrupole ( willick et al . , 1997a ) , as well as the quadrupole and octupole ( parnovsky et al . , 2000
about 4% of the galaxies from our list have considerable ( over 3@xmath50 ) peculiar velocities .
the existence of galaxies , moving at velocities @xmath51(40005000)km / s with respect to the whole bulk , appears to be fairly intriguing .
we assumed that the cases of such deviations are accounted for by the errors in measuring the basic observable quantities : angular diameters , radial velocities , widths .
these errors may be due to morphology peculiarities , to the existence of a close neighbour , etc .
after the revision of these galaxies on the poss - i and eso prints we found no discrepancies between values of newly measured diameters and the catalogue ones .
about half of the galaxies``outliers '' have condensations near the centre or along the disk ( rfgc 59 , 1396 , 1441 , 2637 , 4080 ... ) , bent ( rfgc 1033 , 1272 , 1719 , 3396 , 4054 ... ) , or strongly diffuse ( rfgc 1234 , 1719 , 3200 ... ) disks . some galaxies with high peculiar velocities
are located in the region of clusters and have fairly close neighbours ( rfgc 1542 , 1568 , 1897 ... ) .
new observations will possibly elucidate the matter . in conclusion
the following will be noted : the markiii catalogue and the list presented here are currently the most complete , homogeneous and covering all the sky samples for studying the field of peculiar velocities of galaxies .
having a few common objects , they complement each other both in types of galaxies and in depth of the survey . beginning with the creation of the first version of fgc
, we have kept searching for new estimates from literature and making observations of radial velocities and widths of flat galaxies . in 1995 , from the data for about 800 fgc galaxies , we estimated the velocity modulus and the apex of the bulk motion of flat galaxies , using the forward tf relation and the dipole approximation : @xmath52km / s , @xmath53 , @xmath54 ( karachentsev et al . , 1995 ) .
at the second stage ( karachentsev et al . , 2000 ) a generalized multiparametric tf relation has been derived from which in a dipole approximation over @xmath55 flat galaxies the values @xmath56km / s , @xmath57 , @xmath58 have been obtained and an averaged field of peculiar velocities has been constructed .
our data have turned out to be in good agreement with the results obtained by dekel et al .
( 1999 ) , @xmath59km / s , @xmath60 , @xmath61 .
we have used the present list to investigate the bulk motion on the basis of the generalized tf regression and involvement , in addition to the dipole component , of the quadrupole and octupole components ( parnovsky et al . , 2000 ) . the bulk motion direction in this case too is close to that obtained previously .
dekel a. , eldar a. , kolatt t. , yahil a. , willick j.a . ,
faber s.m . , courteau s. , burstein d. , 1999 , , * 522 * , 1 giovanelli r. , avera a. , karachentsev i.d . , 1997 , , * 114 * , 122 han m .- s . , mould j.r . , 1992 , , * 396 * , 453 karachentsev i.d . , 1991 , ,
* 17 * , 97 karachentsev i.d . , karachentseva v.e . , parnovsky s.l .
, 1993 , astron .
nachr . , * 314 * , 3 ( fgc ) karachentsev i.d .
, karachentseva v.e . , kudrya yu.n . ,
parnovsky s.l . , 1995 ,
nachr . , * 316 * , 369 karachentsev i.d . ,
karachentseva v.e . ,
kudrya yu.n . ,
sharina m.e .
, parnovsky s.l . , 1999 , , *
47 * , 5 ( rfgc ) karachentsev i.d .
, karachentseva v.e . , kudrya yu.n . ,
parnovsky s.l . , 2000 , ,
* 77 * , 175 karachentsev i.d . , makarov d.i . , 1996 , , * 111 * , 794 kogut a. , lineweaver c. , smoot g.f .
et al . , 1993 , , * 419 * , 1 lauberts a. , 1982 , the eso / uppsala survey of the eso(b ) atlas , munich ( eso ) makarov d.i . ,
karachentsev i.d .
, tyurina n.v . ,
kajsin s.s . , 1997a , , * 23 * , 509 makarov d.i . , karachentsev i.d
, burenkov a.n .
, tyurina n.v . , korotkova g.g . , 1997b , , * 23 * , 736 makarov d.i .
, burenkov a.n . ,
tyurina n.v . , 1999 , ,
* 25 * , 706 makarov d.i . , burenkov a.n .
, tyurina n.v .
, 2000 , , in preparation mathewson d.s .
, ford v.l . , buchhorn m. , 1992 , , * 81 * , 413 mathewson d.s . , ford v.l .
, 1996 , , * 107 * , 97 matthews l.d . , van driel w. , 2000 , , * 143 * , 421 parnovsky s.l .
, karachentsev i.d .
, karachentseva v.e . , kudrya yu.n . , 2000 , in preparation strauss m.a . ,
willick j.a . , 1995 ,
, * 261 * , 271 tully r.b . ,
fisher j.r . , 1977 , , * 54 * , 661 tully r.b . , fouqu p. , 1985
, , * 58 * , 67 de vaucouleurs g. , de vaucouleurs a. , corwin h.c . ,
buta r.j . , paturel g. , fouqu p. , 1991
, third reference catalogue of bright galaxies , new york , springer - verlag , v.1 - 3 willick j.a .
, courteau s. , faber s.m . , burstein d. , dekel a. , 1995 , , * 446 * , 12 willick j.a . ,
strauss m.a .
, dekel a. , kolatt t. , 1997a , , * 486 * , 629 willick j.a . , courteau s. , faber s.m . , burstein d. , dekel a. , strauss m.a .
, 1997b , , * 109 * , 333 willick j.a . , 1999 , , * 516 * , 47 | a list of radial velocities , hi line widths and peculiar velocities of 1327 galaxies from the rfgc catalogue has been compiled using actual observations and literature data .
the list can be used for studying bulk motions of galaxies , construction of the field of peculiar velocities and other tasks .
september 27 , 2000 |
hole - doped manganites r@xmath4a@xmath5mno@xmath1 ( r = la , nd , pr , ... and a = sr , ca , ... ) , exhibit colossal magnetoresistive@xcite ( cmr ) properties , associated with the mixed manganese valence mn@xmath6 ( @xmath7)/mn@xmath8(@xmath9 ) resulting from the substitution of trivalent r ions by @xmath10 divalent a ions .
the @xmath10 @xmath11 0.5 doping is particularly interesting , since the magnetic interaction is affected by i ) the ordering of the mn@xmath6 and mn@xmath8 charges@xcite , commonly referred as charge - ordering ( co ) , ii ) the ordering of the @xmath12 electron orbitals on the mn@xmath6 sites and iii ) the coupling between orbital degree of freedom and the lattice .
for example , nd@xmath0ca@xmath0mno@xmath1 ( ncmo ) undergoes a co transition@xcite at @xmath2=245k with partial orbital ordering ( oo ) and magnetic correlations of short range . at lower temperatures , the oo increases and a long range ce - type@xcite antiferromagnetic ( afm ) state
is established at @xmath3=145k .
an insulator - to - metal transition occurs around this temperature in intermediate ( @xmath115 t or higher ) magnetic fields@xcite .
the ce - type phase involves both fm and afm interactions .
it consists of ferromagnetic ( fm ) chains in the ( a , b ) plane , afm coupled between each other within this plane , and along the @xmath13-axis .
neutron@xcite and brillouin@xcite scattering experiments detect additional fm correlations in the low temperature ce - type afm phase .
also resistivity noise measurements reveal two - level fluctuations@xcite related to the phase separation scenario@xcite , including mixed - phase states of different magnetic and electrical properties . in this context , the observed magnetic field induced metal - insulator transition@xcite reflects the percolative growth of conducting fm clusters embedded in an afm matrix@xcite .
+ in the present article , we investigate the stability of the low temperature afm phase of a nd@xmath0ca@xmath0mno@xmath1 single crystal using ac and dc magnetization measurements .
the zero field cooled ( zfc ) and field cooled ( fc ) magnetizations are recorded vs. temperature and time after specific cooling protocols .
similar measurements are performed on a gd@xmath0ca@xmath0mno@xmath1 ( gcmo ) single crystal for comparison .
gcmo also shows charge - ordering at @xmath2 = 260k , but no long range afm , and it remains insulating at all temperatures , even in large magnetic fields@xcite . in
the case of ncmo , the magnetization is cooling rate dependent below @xmath3 , and a weak spontaneous moment appears in the afm state .
the corresponding excess magnetization appearing along the zig - zag chains of the ce - type structure is related to the presence of domain walls in the ( a , b ) plane breaking the orbital coherency .
single crystals of ncmo@xcite and gcmo were grown using a floating zone furnace ( nec , japan ) . temperature and time dependent
zero field cooled and field cooled magnetization measurements were performed using a quantum design mpms5 superconducting quantum interference device ( squid ) magnetometer . in the zfc and fc case ,
the magnetization @xmath14 was collected on re - heating in a small magnetic field @xmath15=20 oe after slow ( 3k / min ) and fast ( 60k / min ) cooling from room temperature down to 5k .
@xmath14 vs. @xmath15 measurements were performed at @xmath16=35k after similar cooling protocols .
additional ac - susceptibility @xmath17 measurements ( @xmath18=125hz , @xmath19=20 oe ) were recorded using the same cooling protocol on a lakeshore 7225 susceptometer for comparison .
the volume susceptibility , @xmath20 in dc and @xmath17 in ac , is in the following plotted in dimensionless ( si ) units .
the correspondence between @xmath20 in ( si ) units and @xmath14 in bohr magnetons per formula unit ( @xmath21/f.u ) is indicated in the text and figures .
figure [ fig1 ] shows the cooling rate dependence of the zfc and fc magnetization for ncmo .
( a ) shows the temperature dependence of the zfc ( markers ) and fc ( simple line ) magnetization , measured on re - heating after fast ( continuous lines ) and slow ( dotted ) cooling . in the case of fast cooling or quench to low temperatures
, an excess magnetization @xmath22 appears below @xmath3 , both in the zfc and fc curves .
difference plots of the fc and zfc curves ( same symbols as in the main frame ) are added in the insert , showing that the excess magnetization @xmath22 [ = @xmath23 @xmath24 @xmath25 appears slightly below @xmath3 , around @xmath16=130k .
this excess magnetization relaxes with time , as illustrated in fig .
[ fig1](b ) which shows the temperature dependence of the zfc magnetization @xmath26(@xmath16 ) : as in fig .
[ fig1](a ) , @xmath26 is recorded on reheating from the lowest temperature @xmath27=5k up to a temperature @xmath28=80k below @xmath3 after a fast cooling .
the @xmath26(@xmath29 ) curve is marked using circles .
the sample is cooled back ( 3k / min ) to @xmath27 and @xmath14 is recorded on re - heating up to @xmath30=300k , well above @xmath31 ; the @xmath26(@xmath32 ) curve is marked using crosses .
the sample is cooled down ( 3k / min ) to @xmath27 , and @xmath14 is once again recorded up to @xmath30 ; the @xmath26(@xmath33 ) curve is plotted using a simple line .
the @xmath26(@xmath16 ) curves obtained in fig .
[ fig1](a ) are added as dotted lines for comparison . as expected , both the first ( from @xmath27 to @xmath28 after a fast cooling ) and the last ( from @xmath27 to @xmath30 after a slow cooling ) measurements coincide with the earlier results shown in fig . [ fig1](a ) .
the second measurement , recorded from @xmath27 to @xmath30 after a temperature cycle to @xmath28 , yields instead a lower @xmath22 , showing that the excess magnetization is relaxing with time .
it should be stressed that only a small dc - magnetic field , acting essentially as a non perturbing probe of the system , is used to record the magnetization ; i.e. the observed effects are not driven by the magnetic field employed in the experiments@xcite . to evidence this ,
we show in insert of fig .
[ fig1](b ) the temperature dependence of the in - phase component of the ac - susceptibility recorded under the same conditions as the zfc and fc magnetization , after fast ( simple line ) and slow ( dotted line ) cooling ; a similar cooling rate dependence is observed in the ac - susceptibility .
+ in our weak probing field , @xmath22 amounts to 2.9@xmath3410@xmath35 @xmath21/f.u at @xmath16=35k ( c.f . fig .
[ fig1](a ) ) .
@xmath14 vs. @xmath15 measurements up to higher fields ( 4000 ka / m ) recorded after fast ( continuous line ) and slow ( dotted line ) cooling to @xmath16=35k are shown in figure [ fig2 ] . in both cases , a closely linear field dependence of the magnetization
is observed , reflecting the afm order .
the difference plot of the two curves ( see insert ) reveals a weak spontaneous moment of 4.4@xmath3410@xmath36 @xmath21/f.u . , superposed on a small excess susceptibility .
the small moment reflects the presence of uncompensated spins in the afm state .
the origin of these local defects of the magnetic structure will be discussed below .
+ in the above described measurements , @xmath14 was recorded using a magnetic field directed along the @xmath13-axis of ncmo .
figure [ fig3](a ) shows the corresponding results on the zfc magnetization measured in the ( a , b ) plane
. a cooling rate dependence is again observed , but the excess magnetization has a smaller magnitude , as seen in the main frame and also in the insert where the difference plots of the zfc curves for the two different orientations of @xmath15 are shown .
most of the excess magnetization thus lies along the @xmath13 direction .
the single crystals show no sign of twinning , but of course some magnetic anisotropy in the ( a , b ) plane would affect the magnitude of @xmath22 .
+ in the case of gcmo , a cooling rate dependence is not observed , and the magnetizations curves recorded along @xmath13 after slow and fast cooling virtually coincide , as shown in fig . [ fig3](b ) . on the other hand ,
the zfc and fc curves deviate from each other below @xmath37100k , indicating irreversibility below this temperature , and thus the development of magnetic correlation . the insert of fig .
[ fig3](b ) shows the normalized difference between the zfc and fc curves for both gcmo and ncmo , which is a measure of the irreversibility . in the case of ncmo ,
the irreversibility arises sharply at @xmath3 , while it appears more gradually in gcmo .
the magnetic correlation developing below @xmath37 100k in gcmo seems to remain of short range , as in the very similar y@xmath0ca@xmath0mno@xmath1@xcite manganite .
the excess magnetization , and the observed cooling rate dependence of the magnetization at low temperatures in ncmo are instead related to the establishment of the long range afm state , possibly via the large increase in orthorhombic distortion@xcite and @xmath13-axis contraction@xcite occurring between @xmath2 and @xmath3 upon cooling .
the spontaneous moment could then be related to the presence of defects in the low temperature antiferromagnetic arrangement , or an antiferromagnetic domain state . in this case ,
uncompensated spins at domain walls would give rise to an excess magnetization .
the curie - like increase of the magnetization at very low temperatures ( below @xmath16 @xmath11 25k ) both in ncmo and gcmo is attributed to paramagnetic nd and gd ions respectively.@xcite + to further elucidate the dynamic nature of the magnetization of ncmo , we have performed fc - relaxation experiments , in which the fc magnetization @xmath38 is recorded versus time ( @xmath39 ) during @xmath40=10000s at @xmath41=35k after different thermal protocols ( * a * , * b * , * c * and * d * ) . the obtained @xmath38(@xmath16 )
curves are plotted in fig .
[ fig4](a ) , and the corresponding relaxation curves @xmath38(@xmath39 ) are shown in fig .
[ fig4](b ) ; all curves are labelled on the figure according to the thermal protocol ( * a*-*d * ) employed . *
a * : the sample is rapidly cooled ( fast cooling rate ) to the measurement temperature @xmath41 @xmath42 @xmath3 in @xmath15=20 oe , and after achieving temperature stability ( @xmath1120s ) , the magnetization is recorded vs. time during @xmath40 . *
b * : the sample is rapidly cooled to the lowest temperature , and the fc - relaxation collected after re - heating to @xmath41 .
the temperature dependence of the magnetization is recorded during the re - heating to @xmath41 and above .
* c * : the sample is rapidly cooled to @xmath41 , from where the cooling proceeds with a slower rate and the magnetization is recorded on cooling and re - heating to @xmath41 , where the fc relaxation is collected during @xmath40 . as in * b * , the magnetization is further recorded during the re - heating to room temperature . *
d * is similar to * c * , but using a slower cooling rate when initially cooling to @xmath41 . as seen in fig .
[ fig4](b ) , the relaxation curves obtained for experiments * a * and * b * are nearly identical , showing again that the cooling rate at all temperatures below @xmath3 is the key parameter of our effect . as observed earlier in fig .
[ fig1](b ) , the relaxation diminishes when the effective cooling slows down , from experiments * a*,*b * to experiment * d*. the relaxation at @xmath41 reflects the evolution of the domain configuration of the afm state .
+ neutron powder diffraction studies on a similar charge and orbital ordered ce - type afm manganite@xcite indicate magnetic disorder in the mn@xmath6 sublattice , associated to domain boundaries breaking the long range orbital ordering .
recent x - ray scattering results@xcite also indicate a partial orbital ordering of the low temperature phase , leading to an orbital domain state .
the here observed cooling rate dependence of @xmath14 below @xmath3 could thus be related to the intrinsic inhomogeneities of the ce - type structure , and the nucleation or rearrangement of orbital domains and domain walls to accommodate the large contraction of the structure occurring upon cooling .
the cooling rate determines the time allowed to the system to accomodate the structural modifications governed by the temperature . in a similar way
, one can also perturb the ce - type state by introducing impurities in the structure , for example by replacing some of the mn cations by cr or ru@xcite .
the orbital ordering is again affected , and fm - like correlations induced . + the cooling rate dependence of the fc magnetization depicted in fig .
[ fig4](a ) is unusual and deserves some additional comments . in experiment
b * , the magnetization curve always lies above the curve obtained for the slow cooling case , closely following the curve obtained for a fast cooling . in experiments
* a * and * c * instead , @xmath43 at @xmath41=35k amounts to 0.0185 [ si ] immediatly after a rapid cooling from room temperature and the magnetization curve thus surprisingly remains below the fc curve obtained for a slow cooling . during a fast cooling to @xmath41 , the system can not accomodate the excess moments appearing with the domain walls , and its magnetization remains lower than in the slow cooling case . in experiment * b * , the sample is rapidly cooled down to the lowest temperature , and reaches a high magnetization value .
albeit having a different initial magnetization level , the relaxation curves corresponding to experiments * a * and * b * ( shown in fig.[fig4](b ) ) appear nearly identical , which indicates that a similar magnetic configuration is probed in both experiments .
the magnetization of a single crystal of the charge ordered manganite nd@xmath0ca@xmath0mno@xmath1 is cooling rate dependent below @xmath3 .
the results reveal the presence of a weak spontaneous moment , related to uncompensated spins in the ce - type afm structure . the moment and its associated excess magnetization
are connected to domain walls separating fully orbital ordered parts of the zig - zag chains of the ce - type structure .
the cooling rate is then used to probe the orbital state , and study how it accommodates the large structural transformation occurring upon cooling .
+ the presence of magnetic inhomogeneities in the low temperature ce - type structure of nd@xmath0ca@xmath0mno@xmath1 is likely to weaken the afm state , so that the application of a large magnetic field could induce the observed insulator - metal transition near @xmath3 . in the case of gd@xmath0ca@xmath0mno@xmath1
, no long range antiferomagnetism is established at any temperature , and neither a cooling rate dependence nor an insulator - metal transition is observed .
j. p. hill , c. s. nelson , m. v. zimmermann , y .-
kim , d. gibbs , d. casa , b. keimer , y. murakami , c. venkataraman , t. gog , y. tomioka , y. tokura , v. kiryukhin , t. y. koo , and s .- w .
cheong , _ unpublished , cond - mat/0105064_. = 20 oe .
( a ) the magnetization is recorded along the c - axis on re - heating after fast ( continuous line ) and slow ( dotted line ) cooling down to 5k .
the corresponding value of @xmath38(@xmath16=35k ) in @xmath21/f.u is added for comparison .
the insert shows the difference plots of the fc and zfc magnetization curves ( same symbols as in the main frame ) .
the maximum value of @xmath22 in @xmath21/f.u is also indicated .
( b ) idem adding a cooling and re - heating to a temperature @xmath28 below @xmath3 while recording @xmath26(@xmath16 ) .
@xmath27 , @xmath28 and @xmath30 correspond to @xmath16=5 , 80 and 300k respectively ; see main text .
the insert shows the cooling rate dependence of the ac - susceptibility of ncmo , recorded vs. temperature using a small ac - field , after similar fast and slow cooling .
@xmath19=20 oe , @xmath18=125hz .
the simple line corresponds to fast cooling , and the dotted line to slow cooling . ] vs. @xmath15 up to high magnetic fields recorded after fast ( continuous line ) and slow ( dotted line ) cooling of ncmo ; @xmath16=35k .
the insert shows the corresponding @xmath22 [ = @xmath23 @xmath24 @xmath25 difference plot . ]
is measured along the c - axis and both zfc ( markers ) and fc ( simple line ) are measured with fast ( continuous line ) and slower ( dotted line ) cooling rates .
the insert shows the temperature dependence of the irreversibility observed in the magnetization curves of ncmo and gcmo . ] for ncmo .
intermediate temperature stops are made during the cooling and heating at @xmath44=35k ( see main text ) ( a ) shows the results plotted vs. temperature , as well as the reference @xmath38(@xmath16 ) curves from fig .
1(a ) in dotted lines . curve * a * is plotted separately for clarity ; @xmath38 was only recorded vs. time in this experiment .
( b ) shows the the relaxation of @xmath38 during @xmath40 at @xmath41=35k . ] | the low temperature phase of single crystals of nd@xmath0ca@xmath0mno@xmath1 and gd@xmath0ca@xmath0mno@xmath1 manganites is investigated by squid magnetometry .
nd@xmath0ca@xmath0mno@xmath1 undergoes a charge - ordering transition at @xmath2=245k , and a long range ce - type antiferromagnetic state is established at @xmath3=145k .
the dc - magnetization shows a cooling rate dependence below @xmath3 , associated with a weak spontaneous moment .
the associated excess magnetization is related to uncompensated spins in the ce - type antiferromagnetic structure , and to the presence in this state of fully orbital ordered regions separated by orbital domain walls .
the observed cooling rate dependence is interpreted to be a consequence of the rearrangement of the orbital domain state induced by the large structural changes occurring upon cooling . |
sky transparency is a fundamental parameter governing the operation of any optical observatory .
best telescope performance comes when skies are photometric - cloud - free - and in general , efficiency at a given site depends directly on the fraction of time that skies are clear enough to observe . as more telescopes employ adaptive optics ( ao ) facilities employing lasers , even thin
cloud can have a significant impact on telescope operations .
attenuation of the beam from cirrus will not only reduce laser return power and affect ao wavefront - sensor performance , but reflected rayleigh - backscatter light may be of concern to other nearby optical observatories .
several sources of sky clarity measurements are available for mauna kea .
the longest historical records date back to the 1970s , consisting of observers nightly notes on sky conditions in the logbooks of the university of hawaii 88-inch , united kingdom infrared telescope ( ukirt ) , and the canada - france - hawaii telescope ( cfht ) .
quantitative measurements have been made in `` campaign mode '' @xcite .
but only in the last few years has there existed dedicated instruments for measuring sky clarity .
two of these are the continuous camera ( concam ) operated by the _ night sky live _ network since 2005 @xcite and the thirty meter telescope ( tmt ) all sky camera ( asca ) installed in 2006 .
another is the cfht skyprobe .
the concam and tmt asca are all - sky imagers , and as such are best used to provide a panoramic view of cloud cover above the summit .
skyprobe is specifically designed to measure sky transparency as an input to telescope data reduction @xcite .
skyprobe is fixed to the side of the cfht .
while the dome is open , once every minute it images bright stars in a 35 square - degree field overlapping the telescope field of view , performs automated photometry , and returns a measurement of attenuation corrected for zenith angle .
it derives an absolute zeropoint based on the hipparcos tycho catalog , which provides a solid photometric reference ; uncertainty is on the order of 0.01 mag .
an excellent feature of this design is that this measurement is in the direction of the telescope , and as such can be directly compared to a concurrent zeropoint measurement of another instrument , for example the megacam imager .
the consistency of the skyprobe data allows the ensemble of measurements taken during a night , or over more extended time periods , to be used to estimate general properties of the atmosphere .
nearby weather measurements from the cfht meteorological tower can be used to check for consistency .
skyprobe does not take data at all times though , as it is shut off when the dome is closed .
one utility of skyprobe , helpful for the observer in pruning data of poor quality , is the detection of clouds .
thick clouds , say 0.5 magnitudes in extinction or more , are easily recognized by sharp fluctuations in attenuation .
less obvious are thin clouds , especially as their attenuations approach that of the atmosphere itself . to elucidate this , it is useful to separate attenuation into air and cloud components as @xmath0 where cloud thickness @xmath1 is analogous to airmass @xmath2 . like @xmath2 , it parameterizes the bulk quantity of absorber without making any assumptions about its distribution along the beam .
constants @xmath3 and @xmath4 indicate attenuations for @xmath5 and @xmath6 . separated in this way
, it can be seen that if conditions are usually photometric , @xmath7 where @xmath8 indicates the mode of @xmath9 , and @xmath10 where @xmath11 is the mean .
so clouds are detectable if their attenuations are greater than @xmath8 or when @xmath1 varies on timescales shorter than @xmath2 - or both . under photometric skies
, astronomers often fix their photometry to a zeropoint assuming an airmass @xmath5 at zenith for the observatory , although that is not strictly true .
the density of air fluctuates with barometric pressure and temperature , and the concentrations of water vapour and other atmospheric constituents such as aerosols ( see , e.g. * ? ? ? * and references therein ) . to first order , however , airmass will increase linearly with barometric pressure @xmath12 and relative humdity @xmath13 , @xmath14 a complication is that @xmath13 measured on the ground does not necessarily correlate with the bulk of the atmosphere .
one extreme case is fog . but barometric pressure is generally stable during a night .
so a plausible empirical model - requiring input only from skyprobe - is that @xmath8 , calculated nightly , tracks the observed atmospheric attenuation .
and an even simpler model avoids the details of fluctuations in airmass entirely .
this assumes that highs in barometric pressure ( or any other atmospheric parameter ) will be balanced by corresponding lows , giving a gaussian distribution with a long term average of @xmath15 where @xmath16 is the standard deviation in nightly @xmath17 , expressed as a percent .
when clouds are present , they may go unnoticed if they are thin , persistent and of uniform thickness .
one possible means of detecting thin cloud is motivated by the fractal properties of cirrus ( and stratus ) clouds , in that their density and lifetime are correlated over many orders of magnitude , governed by a tight power law @xcite .
if so , the number of skyprobe samples of attenuation @xmath18 - the distribution of attenuations - should follow @xmath19 where @xmath20 is a constant and the thickness of clouds is given by @xmath21 . for uniformly timed samples
, the number of samples of a given attenuation should then be directly proportional to the duration of a cloud , @xmath22 , and hence @xmath23 where @xmath24 is the duration of clear skies during an observation lasting @xmath25 , and @xmath26 is the fraction of time that is clear .
this implicitly assumes that samples are taken whatever the cloudiness , that is , without bias towards pointing the telescope at clearer parts of the sky , for example .
that seems reasonable for conditions so good that clouds are hard to detect .
we will simply assume for now that samples are unbiased , and see if that still holds as clouds become thicker .
because the average thickness of a cloud is defined to be @xmath27 and since @xmath28 for maximal thickness @xmath29 , the thickness of a cloud would be given by @xmath30 this suggests that an observer , after subtracting attenuation due to air , and knowing only @xmath20 , could at least estimate the possible variation in cloud thickness and attenuation during a night , even if individual clouds themselves are hard to detect .
in section [ data ] we present the skyprobe data and compare them to contemporaneous megacam photometric zeropoint measurements .
then , in section [ analysis ] we show that our simple models of @xmath2 and @xmath1 are a good match to these data , reproducing both the effects of atmospheric attenuation and up to 3 magnitudes of extinction by clouds .
a summary of the results follows in section [ summary ] .
skyprobe @xmath31-band data were obtained from the period 1 january 2003 through 31 october 2005 .
each record runs from 6 pm to 6 am hawaii standard time , with very few nights lost to the instrument being down .
the first and last hour of data were deleted to avoid high sky brightness , which can lead to incorrect attenuation measurements due to slight nonlinearity in the skyprobe detector .
weather data from the cfht meteorological tower were obtained for the same period .
the fraction of time when either skyprobe did not take data , or the dome shutter was closed ( including due to clouds ) was 31.2% .
it should be noted that the years 2003 through 2005 include some of the worst winter weather on record at cfht .
even so , the mean attenuation from skyprobe is @xmath32 magnitudes per airmass , which is comparable , and even slightly better than the value of @xmath33 reported by @xcite based on ukirt data spanning 1980 to 1986 .
this result is also in good agreement with observer estimates of the sky - transparency for megacam during the period of overlap with the skyprobe data .
table [ table_statistics ] is a breakdown of the average nights per year lost to each instrument due to weather and telescope downtime , along with the remainder with attenuation less than 0.2 magnitudes .
nightly modal skyprobe attenuations and megacam zeropoints are plotted in figure [ figure_data ] , along with air pressure , temperature , and relative humidity for the cfht weather tower for the same period .
note the strong seasonal variation in air pressure and temperature , with winters clearly evident .
there is no obvious seasonal dependence for relative humidity , apart from increased occurrences of saturation in winter .
modal attenuations from skyprobe vary slowly from night to night , for the most part , with smooth oscillations over timescales of a month or more .
the value of @xmath34 inferred is @xmath35 , taken to be the global mode of attenuations and the standard deviation of nightly modes over the same period . for comparison , equation 4
is overplotted on the skyprobe data as a thick curve , boxcar smoothed to better show trends over a month .
this agrees reasonably well with skyprobe apart from some deviations , notably an underestimate for about three months in summer 2005 ( beginning with night 800 ) .
however , there are several sources of atmospheric attenuation unaccounted for in equation 4 .
for example , rayleigh scattering and absorption by ozone have a minimum of about 0.03 magnitudes at 0.55 @xmath36 m for the altitude and latitude of mauna kea , according to the standard calculation of @xcite .
another significant source of atmospheric attenuation is aerosols .
nephelometer measurements of aerosol absorption at nearby mauna loa observatory were downloaded from the national oceanic and atmospheric administration public archives . averaged monthly absorptions ( in the 0.5 @xmath36 m channel ) are plotted as a thin curve in figure [ figure_data ] for the same period .
these have been normalized to a mean attenuation of 0.065 magnitudes , to help show trends relative to equation 4 .
variation in aerosol concentration is seen , and one period of elevated levels does correspond to the `` enhancement '' in skyprobe attenuations in spring / summer 2005 already mentioned .
another rise during the spring of 2003 ( beginning approximately on night 50 ) might be evident in skyprobe data ; a third possibly in spring 2004 ( night 450 ) .
seasonal variation in the concentration of aerosols has been seen at other observatories , notably in central europe @xcite . there
the variation is thought to be the effect of higher summer temperatures raising smoke and soot into the upper atmosphere , combined with greater absolute humidity .
it is not clear that the same mechanisms would be at work on mauna kea , although the seasonal pattern seems similar .
one final `` contaminant '' of skyprobe atmospheric attenuation estimates is evidently thin cloud itself .
a few nights even have modal attenuations of up to 0.45 magnitudes .
further color information might be helpful in discriminating between thin cloud and particulates , and a blue filter ( to match the hipparchos catalog ) has recently been installed on skyprobe @xcite .
the bottom panel of figure [ figure_data ] shows megacam zeropoints ( subtracted from the minimum zeropoint to indicate attenuation ) over the same period .
overplotted are the nightly mode ( for @xmath37 mag ) and median of skyprobe , the latter highlighting cloudy periods , smoothed to help show trends over a month .
note that megacam zeropoints are already corrected by the observer for atmospheric attenuation , based partly on comparison with skyprobe .
it is possible that at times thin cloud may make this task more difficult , and the result more noisy , although the `` striping '' of zeropoints is due to a known processing problem unrelated to skyprobe : residual uncorrected chip - to - chip offsets in megacam .
the value of @xmath8 inferred from megacam is @xmath38 , in agreement with skyprobe ( @xmath35 ) .
figure [ figure_raw ] is a histogram of the skyprobe data , plotted as attenuation in magnitudes .
there is a strong peak at 0.065 magnitudes , indicating the global mode of atmospheric attenuation . to help see this , the distribution of nightly modal attenuations is overplotted as a dashed line . the second bump " near 0.6 magnitudes is an instrumental artifact .
it is caused by untracked exposures at the beginning of the night . at this time
the telescope is sometimes parked at zenith and the resulting star trails are not processed properly by the automated software .
a total of 145 nights identified with the resulting bogus zeropoint have been isolated and are not included in the analysis , although it is possible that some untracked exposures remain .
away from the peak associated with the zeropoint , a power - law behavior dominates .
a linear least - squares fit to data with between 0.4 and 1 magnitude of extinction ( solid line ; extended above and below the fitted region ) gives a power - law slope of -1.84 .
comparing this with equation 6 gives @xmath39 .
note how well this curve fits - with constant @xmath20 - even for attenuations approaching 3 magnitudes .
we are not interested in this end of the distribution of course , as there is so little power there , but it is still worth mentioning .
a bias against observing high @xmath9 values - closing the dome for clouds , or systematically pointing the telescope to clearer parts of the sky - would presumably steepen the slope of this curve , perhaps even resulting in a knee at an attenuation where clouds become obvious , say , at 0.5 mag . a bend in the opposite sense
would indicate observations taken _ preferentially _ in cloudy conditions .
a more plausible scenario , consistent with the cfht nightly log is that the dome is generally opened for nights that begin as mostly clear ( with perhaps some cirrus ) , and observations continue as long as interruptions by thick cloud are brief .
the dome is not opened during poor nights , as this leads to inefficient observing , and instead these nights are allocated to engineering tasks , if possible .
the fraction of power under the curve ( lightly shaded region ) compared to that in the total distribution is 18.6% .
this is the total fraction of dome - open time that was cloudy .
this indicates that for 81.4% of the skyprobe samples , attenuation was due solely to air .
skyprobe was essentially always operational during the period discussed here and took data whenever the dome shutter was open . taking this to be 68.8% of the time ,
conditions were photometric @xmath40% of the time .
the true value must be something more than this because downtime due to non - weather reasons ( when the sky may not be cloudy ) have been included , lumped in with times when the dome was closed ( darkly shaded region ) .
for example , because skyprobe is attached to the telescope , datataking is subject to the telescope slewing to a new target .
a more detailed histogram is shown in figure [ figure_histogram ] , now plotted only for attenuations less than 0.8 magnitudes ; the middle panel is that same , except with the nightly modal attenuation subtracted ; and the bottom panel is a cumulative histogram of separated air and cloud .
the thick black curve is the result of combining the distribution of nightly modal attenuations ( thick dashed curve ) with that for the cloud model , indicated by the shaded region .
this provides a good fit , although it seems that the tail of the nightly modal distribution includes some nights when thin cloud was prevalent . to see this , the cloud model combined with the simple gaussian model of atmospheric attenuation
is indicated by the thin black curve .
the dotted line indicates gaussian noise of 0.01 magnitudes about a fixed atmospheric attenuation of 0.065 mag , that is , @xmath41 and no cloud .
it is the addition of this noise which sufficiently broadens the distribution of atmospheric attenuation that it leads to the nonphysical result of attenuations approaching zero , even becoming negative .
this effectively restricts the accuracy of subtraction of the modal attenuation - only a spike could be perfectly subtracted - which is the cause of the scatter in the cloud measurements below about 0.2 mag , and ultimately makes the estimate of the clear fraction uncertain to within perhaps 2% . to see if any more information can be gleaned from the distribution of skyprobe attenuations , for example , if the clear fraction of time varies seasonally , this calcuation was repeated after binning the data by calendar month .
the results are presented in figure [ figure_monthly ] .
the solid line is the calculation of cloud - free fraction of time , with the lightly - shaded region indicating cloud ; dark shading indicates the dome is closed .
note that cfht is very efficient in maximimizing the amount of time that the dome is open when skies are clear ; this fraction of time varies moderately over the year , with perhaps a peak in late summer approaching 60% .
this and the values of other parameters appearing in equations 1 through 6 are given in table [ table_parameters ] .
so far , we have only fit a power law to the skyprobe data as per equation 6 without need for an explicit assumption about the duration of cloud - free skies other than to bin the data by month .
it is interesting to consider for what values of @xmath42 equation 10 is a good fit to the data .
a computer simulation was developed to help explore this .
seven artificial skyprobe datasets , assuming @xmath43 1 , 2 , 6 , 12 , 24 , 72 and 168 hours , were generated as follows : an initial cloud duration was selected at random from between zero and @xmath42 . if a duration of zero was selected , then a period of @xmath42 had no attenuation .
otherwise , @xmath1 was calculated via equation 10 .
a power - law slope of @xmath44 was used . in one - minute timesteps until the end of this period all datapoints each had this value .
then a new random duration was selected and the process repeated for a number of samples similar to the observations .
poisson noise of 0.01 magnitudes was added to simulate instrumental uncertainty .
data lost due to the dome being shut and to overheads was simulated by masking out 31.2% of nights with the highest median attenuation , just as cfht closes for cloudy nights . as can be expected , since the mean of @xmath1 has been defined to be unity , each of these simulations has the same mean attenuation , and correspondingly the same mean duration of clouds , that is @xmath45 magnitudes and @xmath46 hours .
the results are shown in figure [ figure_images ] along with the skyprobe data , displayed as grey - scale images .
the nightly modal attenuation has been subtracted from the data to provide an estimate of attenuation due only to cloud .
white indicates high attenuation ; the same stretch is used throughout .
each pixel represents the average of three skyprobe samples .
each row is one night of data , with a full year along a column .
instrumental noise is included , so clear skies appear as a uniform grey ; black indicates the dome is shut .
small values of @xmath42 , less than 24 hours , seem to reproduce the short , thick clouds seen in the data .
larger values of @xmath42 provide longer
clouded out " periods , some lasting several days , also evident in the data . but
larger values of @xmath42 produce too few short , thick clouds .
it may be that a single value of @xmath42 is insufficient to reproduce both long - term periods of cloud and the appropriate variation in attenuation during a night .
this can be seen in figure [ figure_statistics ] , which is a plot of the standard deviation of attenuation as a function of various time intervals .
this has been calculated by dividing up each night s skyprobe data into blocks of a given interval , and calculating the standard deviation over each .
this analysis is restricted to attenuations of 0.5 magnitudes or less , so conditions excluding obvious clouds . shown
are values for 1 , 2 , 5 , 10 , 15 , 30 , and 60 minutes , and then every hour up to 10 hours .
note the strong power - law behaviour , as expected from the cloud model .
thin black curves indicate the results for choices of @xmath42 shown in figure [ figure_images ] plus that of @xmath47 hour ; the thick black curve is the average of these .
although the average of the models does not exactly reproduce the data , it is a reasonably good fit , especially for observational durations longer than an hour .
that the average of the models underestimates variation in attenuation on timescales less than an hour suggests that @xmath42 might include shorter timescales than this .
and so it seems that a refined distribution of @xmath42 - possibly spread over all timescales - could produce a better fit .
but determining this distribution is beyond the scope of this paper .
one path to it may be to investigate similarities with the fried parameter @xmath48 associated with optical turbulence , where strong self - similar behaviour is also evident .
a final result , evident in figure [ figure_statistics ] , is that thin clouds set a threshold for photometric accuracy .
even if atmospheric attenuation is subtracted perfectly , if clouds less than 0.5 mag in attenuation are not accounted for , observations lasting an hour or more will typically incur a 0.03 magnitude absolute error in photometry .
this `` cloud - induced ''
error is comparable to the standard deviation in all megacam zeropoints ( 0.031 mag ) .
as one might expect , the error component due to thin cloud could be reduced by confining observations to better sky conditions .
it is not shown in figure [ figure_statistics ] , but further restricting the skyprobe analysis to attenuations less than 0.2 magnitudes reduces the slope of the cloud - induced error , yielding a 0.01 mag standard deviation after 1 hour .
of course , the available observing time under these better conditions is reduced too , down to 60% based on table [ table_statistics ] ; only 56% of the time will have no clouds at all .
this serves as a reminder to observers to obtain frequent photometric calibration ; further validation of skyprobe as an instrument dedicated to measuring sky transparency .
we have presented an analysis of cfht skyprobe data covering almost three calendar years beginning in 2003 .
the results are in agreement with megacam data during the same period .
the skyprobe data show that the v - band attenuation had a modal value of 0.065 magnitudes corresponding to 56% of time with attenuation due to an atmosphere without clouds , and that 59% of the time v - band attenuation was 0.20 magnitudes or less ( including both air and thin cloud ) .
these estimates are conservative , as they include small telescope overheads .
some modest seasonal variation , @xmath49% , is seen in the fraction of time that skies are clear , but ultimately the accuracy in that estimate is perhaps @xmath50% , limited by the photometric uncertainty of skyprobe . as these results are for open dome shutter time on cfht , and account for overheads , they can be used as inputs to designing ao with lasers , for example . a power - law model of cloud attenuation and duration has been presented which is a good fit to the skyprobe data . it is consistent both with the mean attenuation due to thin clouds , and the appearance of brief thick clouds at times causing the clouding out " of nights .
it is hoped that this simple model will lead to further investigation of the self - similar nature of clouds on mauna kea and elsewhere , which could possibly lead to improvements in photometric accuracy at the telescopes .
we thank derrick salmon and konstantinos vogiatzis for their assistance in obtaining data from the cfht weather tower and chris pritchet for his megacam zeropoints , and thoughtful comments .
we would like to thank glen herriot and jean - pierre veran for helpful conversations , and the thirty meter telescope site - testing advisory group for useful comments on an earlier internal report on these results .
this work includes archival data obtained from the national oceanic and atmospheric administration , climate monitoring and diagnostics laboratory , aerosols group .
chambers , k.c .
2005 , asp conf .
ser . , 338 , 134 cuillandre , j .- c . ,
magnier , e.a . ,
isani , s. , sabin , d. , knight , w. , kras , s. , & lai , k. 2002 , , 4844 , 501 cuillandre , j .- c .
magnier , e. , sabin , d. & mahoney , b. , 2008 , , 7016 , 25 hayes , d.s . , & latham , d.w .
1975 , , 197 , 593 ivanova , k. , ausloos , m. , clothiaux , e.e . , shirer , h.n , & ackerman , t.p . , 2000 , tenth arm science team mtng .
san antonio , nm , 1 ivanova , k. , shirer , h.n . ,
clothiaux , e.e . ,
kitova , n. , mikhalev , m.a . ,
ackerman , t.p .
, & ausloos , m. , 2002 , phys .
a. , 308 , 518 krisciunas , k. , sinton , w. , tholen , d. , tokunaga , a. , golisch , w. , griep , g. , kaminski , c. , impey , c. & christian , c. 1987 , , 99 , 887 mccord , t.b . & clark , r.n .
1979 , , 91 , 571 pakstiene , e. 2001 , baltic astronomy , 10 , 651 reimann , h .- g . , ossenkopf , v. , & beyersdorfer , s. 1992 , astron . and astrophys . , 265 , 360 shamir , l. & nemiroff r.j .
2005 , , 117 , 972 | nighttime sky transparency statistics on mauna kea are reported based on data from the canada - france - hawaii telescope skyprobe monitor .
we focus on the period beginning with the start of megacam wide - field optical imager operations in 2003 , and continuing for almost three years .
skies were clear enough to observe on 76% of those nights ; attenuations were less than 0.2 magnitudes up to 60% of the time .
an empirical model of cloud attenuation and duration is presented allowing us to further characterize the photometric conditions .
this is a good fit to the skyprobe data , and indicates that mauna kea skies are truly photometric ( without cloud ) an average of 56% of the time , with moderate seasonal variation .
continuous monitoring of transparency during the night is necessary to overcome fluctuations in attenuation due to thin cloud . |
last decade witnessed a brutal transformation in neutrino physics .
it has been experimentally observed that neutrinos have nonzero masses , implying that leptons blend .
this fact was demonstrated by the experimental evidence that neutrinos can change from one state , or `` flavour '' , to another .
all the information we have accumulated about neutrinos , is quite recent .
less that twenty years old .
neutrino physics as a solid science is in its teenage years and therefore as any adolescence , in a wild and very exciting ( and excited ) state .
however , before jumping into the late `` news '' about neutrinos , lets understand how and why neutrinos were conceived .
the 20s saw the death of numerous sacred cows , and physics was no exemption .
one of physic s most holly principles , energy conservation , apparently showed up not to hold inside the subatomic world . for some radioactive nuclei , it appeared that a non - negligible fraction of its energy simply vanished , leaving no trace of its presence . in 1920 , in a ( by now famous ) letter to a meeting @xcite , pauli quasi apologetically wrote,``dear radioactive ladies and gentlemen , ... as a desperate remedy to save the principle of energy conservation in beta decay , ...
i propose the idea of a neutral particle of spin half '' .
pauli hypothesised that the missing energy was taken off by another particle , whose properties were such that made it invisible and impossible to detect : it had no electric charge , no mass and only very rarely interacted with matter . along these lines
, the neutrino was naturally introduced to the universe of particle physics .
before long , fermi postulated the four - fermi hamiltonian in order to describe beta decay utilising the neutrino , electron , neutron and proton .
another field was born : weak interactions took the stage to never leave it .
closing the loop , twenty years after pauli s letter , cowan and reines got the experimental signature of anti - neutrinos emitted by a nuclear power plant .
as more particles who participated in weak interactions were found in the years following neutrino discovery , weak interactions got credibility as an authentic new force of nature and the neutrino got to be a key element of it . further experimental tests through the span of the following 30 years demonstrated that there were not one but three sort , or `` flavours '' of neutrinos ( electron neutrinos ( @xmath0 ) , muon neutrinos ( @xmath1 ) and tau neutrinos ( @xmath2 ) ) and that , to the extent we could test , had no mass ( and no charge ) whatsoever .
the neutrino adventure could have easily finish there , however new analyses in neutrinos coming from the sun shown us that the neutrino saga was just beginning .... in the canonical standard model , neutrinos are completely massless and as a consequence are flavour eigenstates , @xmath3 precisely because they are massless , they travel at the speed of light and accordingly their flavour does not change from generation up to detection . it is evident then , that as flavour is concerned , zero mass neutrinos are not an attractive object to study , specially when contrasted with quarks .
however , if neutrinos were massive , and these masses where not degenerate ( degenerate masses flavour - wise is identical to the zero mass case ) would mean that neutrino mass eigenstates exist @xmath4 , each with a mass @xmath5 .
the impact of leptonic mixing becomes apparent by looking at the leptonic decays , @xmath6 of the charged vector boson @xmath7 .
where , @xmath8 , or @xmath9 , and @xmath10 refers to the electron , @xmath11 the muon , or @xmath12 the tau .
we call particle @xmath13 as the charged lepton of flavour @xmath14 .
mixing basically implies that when the charged boson @xmath15 decays to a given kind of charged lepton @xmath16 , the neutrino that goes along is not generally the same mass eigenstate @xmath17 . _
any _ of the different @xmath17 can appear . the amplitude for the decay of a vector boson @xmath15 to a particular mix @xmath18 is given by @xmath19 .
the neutrino that is radiated in this decay alongside the given charged lepton @xmath16 is then @xmath20 this specific mixture of mass eigenstates yields the neutrino of flavour @xmath14 .
the different @xmath21 can be gathered in a unitary matrix ( in the same way they were collected in the ckm matrix in the quark sector ) that receives the name of the leptonic mixing matrix or @xmath22 @xcite .
the unitarity of @xmath23 ensures that each time a neutrino of flavour @xmath14 through its interaction produces a charged lepton , the produced charged lepton will always be @xmath13 , the charged lepton of flavour @xmath24 .
that is , a @xmath25 produces exclusively an @xmath26 , a @xmath27 exclusively a @xmath28 , and in a similar way @xmath29 a @xmath9 . the expression ( [ eq1 ] ) , portraying each neutrino of a given flavour as a linear combination of the three mass eigenstates , can be easily inverted to depict every mass eigenstate @xmath17 as an analogous linear combination of the three flavours : @xmath30 the amount of @xmath14-flavour ( or the @xmath14-fraction ) of @xmath17 is obviously @xmath31 .
when a @xmath17 interacts and creates a charged lepton , this @xmath14-content ( or fraction ) expresses the probability that the created charged lepton be of flavour @xmath14 .
the phenomenon of neutrino morphing , flavour transition or in short oscillation , can be understood in the following form . a neutrino is created or emitted by a source along with a charged lepton @xmath16 of flavour @xmath14 . in this way , at the emission point , the neutrino does have a definite flavour .
it is a @xmath32 .
after that point , after production , the neutrino covers some length ( propagates thorough a distance ) @xmath33 until it is absorbed . at this point ,
when it has already covered the distance to the target , the neutrino interacts and these interactions create another charged lepton @xmath34 of flavour @xmath35 , which we can detect . in this way , at the target , we can know that the neutrino is again a neutrino of definite flavour , a @xmath36 . of course there is a chance that @xmath37 ( for instance , if @xmath13 is a @xmath38 however @xmath34 is a @xmath9 ) , then , all along his journey from the source to the identification point , the neutrino has morphed or transformed from a @xmath32 into a @xmath36 .
this transition from one flavour to the other , @xmath39 , is a canonical case of the widely known quantum - mechanical effect present in a variety of two state systems and not a particular property of neutrinos .
since , as shown clearly by eq .
( [ eq1 ] ) , a @xmath32 is truly a coherent superposition of the three mass eigenstates @xmath17 , the neutrino that travels since it is born until it is detected , can be any of the three @xmath17 s .
because of that , we should include the contributions of each of the @xmath17 in a coherent way . as a consequence , the transition amplitude , amp(@xmath39 ) receives a contribution of each @xmath17 and turns out to be the product of three pieces .
the first factor is the amplitude for the neutrino created at the generation point along with a charged lepton @xmath16 to be , particularly , a @xmath17 . and
as we have said already , it is given by @xmath40 .
the second component of our product is the amplitude for the @xmath17 made by the source to cover the distance up to the detector .
we will name this element prop(@xmath17 ) for the time being and will postpone the calculation of its value until later . the last ( third ) piece is the amplitude for the charged lepton born out of the interaction of the neutrino @xmath17 with the target to be , particularly , a @xmath34 .
being the hamiltonian that describes the interactions between neutrinos , charged leptons and charged bosons @xmath7 bosons hermitian ( otherwise probability wo nt be conserved ) , it follows that if amp(@xmath41 , then amp @xmath42 . in this way
, the third and last component of the product the @xmath17 contribution is given by @xmath43 , and @xmath44 it still remains to be established the value of prop(@xmath17 ) . to determine it , we d better study the @xmath17 in its rest frame .
we will label the time in that system @xmath45 .
if @xmath17 does have a rest mass @xmath5 , then in this frame of reference its state vector satisfies the good old schrdinger equation @xmath46 whose solution is given clearly by @xmath47 then , the amplitude for a given mass eigenstate @xmath17 to travel freely during a time @xmath45 , is simply the amplitude @xmath48 for observing the initial state @xmath17 , @xmath49 after some time as the evoluted state @xmath50 , _ ie . _
@xmath51 $ ] .
thus prop(@xmath17 ) is only this amplitude where we have used that the time taken by @xmath17 to cover the distance from the source to the detector is just @xmath45 , the proper time . nevertheless ,
if we want prop(@xmath17 ) to be of any use to us , we must write it first in terms of variables we can measure , this means to express it , in variables in the lab frame .
the natural choice is obviously the distance , @xmath33 , that the neutrino covers between the source and the detector as seen in the lab frame , and the time , @xmath52 , that slips away during the journey , again in the lab frame .
the distance @xmath33 is set by the experimentalists through the selection of the place of settlement of the source and that of the detector and is unique to each experimental setting .
likewise , the value of @xmath52 is selected by the experimentalists through their election for the time at which the neutrino is made and that when it dies ( or gets detected ) . therefore , @xmath33 and @xmath52 are determined ( hopefully carefully enough ) by the experiment design , and are the same for all the @xmath17 in the beam .
the different @xmath17 do travel through an identical distance @xmath33 , in an identical time @xmath52 .
we still have two additional lab frame variables to determine , the energy @xmath53 and three momentum @xmath54 of the neutrino mass eigenstate @xmath17 .
by using the lorentz invariance of the four component internal product ( scalar product ) , we can obtain the expression for the @xmath55 appearing in the @xmath56 propagator prop(@xmath17 ) in terms of the ( easy to measure ) lab frame variable we have been looking for , which is given by @xmath57 at this point however one may argue that , in real life , neutrino sources are basically constant in time , and that the time @xmath52 that slips away since the neutrino is produced till it dies in the detector is actually not measured .
this argument is absolutely right . in reality ,
an experiment smears over the time @xmath52 used by the neutrino to complete its route .
however , lets consider that two constituents of the neutrino beam , the first one with energy @xmath58 and the second one with energy @xmath59 ( both measured in the lab frame ) , add up coherently to the neutrino signal produced in the detector .
now , let us call @xmath52 to the the time used by the neutrino to cover the distance separating the production and detection points .
then by the time the constituent whose energy is @xmath60 arrives to the detector , it has raised a phase factor @xmath61 $ ] .
therefore , we will have an interference between the @xmath58 and @xmath59 beam participants that will include a phase factor @xmath62 $ ] .
when smeared over the non - observed travel time @xmath52 , this factor goes away , _ except when @xmath63_. therefore , only those constituents of the neutrino beam that share the same energy contribute coherently to the neutrino oscillation signal @xcite . specifically , only the different mass eigenstates constituents of the beams that have the same energy weight in .
the rest gets averaged out .
courtesy to is dispersion relation , a mass eigenstate @xmath17 , with mass @xmath5 , and energy @xmath64 , has a three momentum @xmath54 whose absolute value is given by @xmath65 where , we have utilised that as the masses of the neutrinos are miserably small , @xmath66 for a typical energy @xmath64 attainable at any experiment ( the lowest energy neutrinos have mev energies and sub - ev masses ) . from eqs .
( [ eq6 ] ) and ( [ eq7 ] ) , it is easy to see that at a given energy @xmath64 the phase @xmath67 appearing in prop(@xmath17 ) takes the value @xmath68 as the phase @xmath69 appears in all the interfering terms it will eventually disappear when calculating the transition amplitude .
after all is a common phase factor ( its absolute value is one ) .
thus , we can get rid of it already now and use @xmath70 ~~ . \label{eq9}\ ] ] plugging this into eq .
( [ eq3 ] ) , we can obtain that the amplitude for a neutrino born as a @xmath32 to be detected as a @xmath36 after covering a distance @xmath33 with energy @xmath64 yields @xmath71 the expression above is valid for an arbitrary number of neutrino flavours and an identical number of mass eigenstates , as far as they travel through vacuum .
the probability p(@xmath39 ) for @xmath39 can be found by squaring it , giving @xmath72 with @xmath73 in order to get eq .
( [ eq11 ] ) we have used that the mixing matrix @xmath23 is unitary .
the oscillation probability p(@xmath39 ) we have just obtained corresponds to that of a _ neutrino _ , and not to an _ antineutrino _ , as we have used that the oscillating neutrino was produced along with a charged _ antilepton _
@xmath74 , and gives birth to a charged _ lepton _ @xmath75 once it reaches the detector .
the corresponding probability p(@xmath76 ) for an antineutrino oscillation can be obtained from p(@xmath39 ) taking advantage of the fact that the two transitions @xmath76 and @xmath77 are cpt conjugated processes .
thus , assuming that neutrino interactions respect cpt@xcite , @xmath78 then , from eq .
( [ eq11 ] ) we obtain that @xmath79 therefore , if cpt is a good symmetry ( as far as neutrino interactions are concerned ) , eq .
( [ eq11 ] ) tells us that @xmath80{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\alpha $ } \longrightarrow \shortstack{{\tiny ( \rule[.4ex]{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\beta $ } ) & = & \delta_{\alpha\beta } - 4\sum_{i
> j } \re ( u^*_{\alpha i } u_{\beta i } u_{\alpha j}u^*_{\beta j } ) \sin^2 \left ( \delta m^2_{ij}\frac{l}{4e } \right ) \nonumber \\ & & + ( - ) \ ; 2\sum_{i > j } \im ( u^*_{\alpha i } u_{\beta i } u_{\alpha j}u^*_{\beta j } ) \sin\ , \left ( \delta m^2_{ij}\frac{l}{2e } \right ) ~~ . \label{eq15}\end{aligned}\ ] ] these expressions make it clear that if the mixing matrix @xmath23 is complex , p(@xmath81 ) and p(@xmath39 ) will not be identical , in general .
as @xmath82 and @xmath39 are cp conjugated processes , p@xmath83 would provide evidence of cp violation in neutrino oscillations ( if nature has chosen its mixing parameters so that the mixing matrix is indeed complex ) .
until now , cp violation has been observed only in the quark sector , so its measurement in neutrino physics would be quite exciting .
so far , we have been working in natural units . a fact that becomes transparent by looking at the dispersion relation eq .
( [ eq8 ] ) .
if we restore now the @xmath84 s and @xmath85 factors ( we have happily set to one ) into the oscillation probability we find that @xmath86 having done that , it is easy and instructive to explore the semi - classical limit , @xmath87 . in this limit
the oscillation length goes to zero ( the oscillation phase goes to infinity ) and the oscillations are averaged to 1/2 . the interference pattern is lost .
a similar situation appears if we let the mass difference @xmath88 become large .
this is exactly what happens in the quark sector ( and the reason why we never study quark oscillations despite knowing that mass eigenstates do not coincide with flavour eigenstates ) . in terms of real life units ( which are not `` natural '' units ) , the oscillation phase is given by @xmath89 then , since @xmath90 $ ] can be experimentally observed ( _ ie .
_ not smeared out ) only if its argument is in a ballpark around one , an experimental set - up with a baseline @xmath33 ( km ) and an energy @xmath64 ( gev ) is sensitive to neutrino mass squared differences @xmath91 of order @xmath92^{-1}$ ] .
for example , an experiment with a baseline of @xmath93 km , roughly the size of earth s diameter , and @xmath94 gev would explore mass differences @xmath95 down to @xmath96 ev@xmath97 .
this fact makes it clear that neutrino long - baseline experiments can test even miserably small neutrino mass differences .
it does so by exploiting the quantum mechanical interference between amplitudes whose relative phases are given precisely by these super tiny neutrino mass differences , which can be transformed into sizeable effects by choosing @xmath98 appropriately . but
let s keep analysing the oscillation probability and see whether we can learn more about neutrino oscillations by studying its expression .
it is clear from @xmath99{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\alpha $ } \longrightarrow \shortstack{{\tiny ( \rule[.4ex]{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\beta $ } ) $ ] that if neutrinos have zero mass , in such a way that all @xmath100 , then , @xmath101{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\alpha $ } \longrightarrow \shortstack{{\tiny ( \rule[.4ex]{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\beta $ } ) = \delta_{\alpha\beta}$ ] .
therefore , the experimental observation that neutrinos can morph from one flavour to a different one indicates that neutrinos are not only massive but also that their masses are not degenerate . actually , it was precisely this evidence the one that proved beyond any reasonable doubt that neutrinos are massive .
however , every neutrino oscillation seen so far has involved at some point neutrinos that travel through matter .
but the expression we derived is valid only for flavour change in vacuum , and does not take into account any interaction between the neutrinos and the matter traversed between their source and their detector .
thus , the question remains whether it may be that some unknown flavour changing interactions between neutrinos and matter are indeed responsible of the observed flavour transitions , and not neutrino masses .
regarding this question , a couple of things should be said .
first , although it is true that the standard model of elementary particle physics contains only massless neutrinos , it provides an amazingly well corroborated description of weak interactions , and therefore of all the ways a neutrino interacts .
such a description does not include flavour change .
second , for some of the processes experimentally observed where neutrinos do change flavour , matter effects are expected to be miserably small , and on those cases the evidence points towards a dependence on @xmath33 and @xmath64 in the flavour transition probability through the combination @xmath98 , as anticipated by the oscillation hypothesis .
modulo a constant , @xmath98 is precisely the proper time that goes by in the rest frame of the neutrino as it covers a distance @xmath33 possessing an energy @xmath64 .
therefore , these flavour transitions behave as if they were a true progression of the neutrino itself over time , and not a result of an interaction with matter .
now , lets explore the case where the leptonic mixing were trivial .
this would imply that in the charged boson decay @xmath102 , which as we established has an amplitude @xmath40 , the emerging charged antilepton @xmath16 of flavour @xmath14 comes along always with the _ same _ neutrino mass eigenstate @xmath17 .
that is , if @xmath103 , then due to unitarity , @xmath104 becomes zero for all @xmath105 .
therefore , from eq .
( [ eq15 ] ) it is clear that , @xmath99{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\alpha $ } \longrightarrow \shortstack{{\tiny ( \rule[.4ex]{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\beta $ } ) = \delta_{\alpha\beta}$ ] .
thus , the observation that neutrinos morph indicates non trivial a mixing matrix .
then , we are left with basically two ways to detect neutrino flavour change .
the first one is to observe , in a beam of neutrinos which are all created with the same flavour , say @xmath14 , some amount of neutrinos of a new flavour @xmath35 that is different from the flavour @xmath14 we started with .
this goes under the name of appearance experiments .
the second way is to start with a beam of identical @xmath32s , whose flux is either measured or known , and observe that after travelling some distance this flux is depleted .
such experiments are called disappearance experiments . as eq .
( [ eq15 ] ) shows , the transition probability in vacuum does not only depend on @xmath98 but also oscillates with it .
it is because of this fact that neutrino flavour transitions are named `` neutrino oscillations '' .
now notice also that neutrino transition probabilities do not depend on the individual neutrino masses ( or masses squared ) but on the squared - mass _ differences_. thus , oscillation experiments can only measure the neutrino mass squared spectrum . not its absolute scale .
experiments can test the pattern but can not determine the distance above zero the whole spectra lies .
it is clear that neutrino transitions can not modify the total flux in a neutrino beam , but simply alter its distribution between the different flavours . actually , from eq .
( [ eq15 ] ) and the unitarity of the @xmath23 matrix , it is obvious that @xmath106{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\alpha $ } \longrightarrow \shortstack{{\tiny ( \rule[.4ex]{1em}{.1 mm } ) } \\ [ -.7ex ] $ \nu_\beta $ } ) = 1 ~~ , \label{eq17}\ ] ] where the sum runs over all flavours @xmath35 , including the original one @xmath14 .
( [ eq17 ] ) makes it transparent that the probability that a neutrino morphs its flavour , added to the probability that it keeps the flavour it had at birth , is one .
ergo , flavour transitions do not modify the total flux .
nevertheless , some of the flavours @xmath37 into which a neutrino can oscillate into may be _ sterile _ flavours ; that is , flavours that do not take part in weak interactions and therefore escape detection . if any of the original ( active ) neutrino flux turns into sterile , then an experiment able to measure the total _ active _ neutrino flux that is , the flux associated to those neutrinos that couple to the weak gauge bosons : @xmath107 , and @xmath29 will observe it to be not exactly the original one , but smaller than it . in the experiments
performed up today , no flux was ever missed . in the literature ,
description of neutrino oscillations normally assume that the different mass eigenstates @xmath17 that contribute coherently to a beam share the same _ momentum _ , rather than the same _ energy _ as we have argued they must have . while the supposition of equal momentum is technically wrong
, it is an inoffensive mistake , since , as can easily be shown , it conveys to the same oscillation probabilities as the ones we have obtained .
a relevant and interesting case of the ( not that simple ) formula for p@xmath108 is the case where only two flavours participate in the oscillation .
the only - two - neutrino scenario is a rather rigorous description of a vast number of experiments .
in fact only recently ( and in few experiments ) a more sophisticated ( three neutrino description ) was needed to fit observations .
lets assume then , that only two mass eigenstates , which we will name @xmath109 and @xmath110 , and two reciprocal flavour states , which we will name @xmath27 and @xmath29 , are relevant , in such a way that only one squared - mass difference , @xmath111 arises .
even more , neglecting phase factors that can be proven to have no impact on oscillation probabilities , the mixing matrix @xmath23 can be written as @xmath112 the unitary mixing matrix @xmath23 of eq .
( [ eq18 ] ) is just a 2@xmath1132 rotation matrix , and as such , parameterized by a single rotation angle @xmath114 which is named ( in neutrino physics ) as the mixing angle .
plugging the @xmath23 of eq .
( [ eq18 ] ) and the unique @xmath88 into the general formula of the transition probability p@xmath108 , eq .
( [ eq15 ] ) , we can readily see that , for @xmath37 , when only two neutrinos are relevant , @xmath115 moreover , the survival probability , _ ie . _
the probability that the neutrino remains with the same flavour its was created with is , as expected , one minus the probability that it changes flavour .
when we create a beam of neutrinos on earth through an accelerator and send it up to thousand kilometres away to a meet detector , the beam does not move through vacuum , but through matter , earth matter .
the beam of neutrinos then scatters from the particles it meets along the way .
such a coherent forward scattering can have a large effect on the transition probabilities .
we will assume for the time being that neutrino interactions with matter are flavour conserving , as described by the standard model , and comment on the possibility of flavour changing interactions later .
then as there are only two types of weak interactions ( mediated by charged and neutral currents ) the would be accordingly only two possibilities for this coherent forward scattering from matter particles to take place .
charged current mediated weak interactions will occur only if and only if the incoming neutrino is an electron neutrino .
as only the @xmath25 can exchange charged boson @xmath7 with an earth electron . thus neutrino - electron coherent forward scattering via @xmath7 exchange opens up an extra source of interaction energy @xmath116 suffered exclusively by electron neutrinos . obviously , this additional energy being from weak interactions origin has to be proportional to @xmath117 , the fermi coupling constant .
in addition , the interaction energy coming from @xmath118 scattering grows with the number of targets , @xmath119 , the number of electrons per unit volume ( given by the density of the earth ) .
putting everything together it is not difficult to see that @xmath120 clearly , this interaction energy affects also antineutrinos ( in a opposite way though ) .
it changes sign if we switch the @xmath25 by @xmath121 .
the interactions mediated by neutral currents correspond to the case where a neutrino in matter interacts with a matter electron , proton , or neutron by exchanging a neutral @xmath122 boson . according to the standard model weak interactions
are flavour blind .
every flavour of neutrino enjoys them , and the amplitude for this @xmath122 exchange is always the same
. it also teaches us that , at zero momentum transfer , electrons and protons couple to the @xmath122 boson with equal strength .
the interaction has though , opposite sign .
therefore , counting on the fact that the matter through which our neutrino moves is electrically neutral ( it contains equal number of electrons and protons ) , the contribution of both , electrons and protons to coherent forward neutrino scattering through @xmath122 exchange will add up to zero .
consequently only interactions with neutrons will survive so that , the effect of the @xmath122 exchange contribution to the interaction potential energy @xmath123 reduces exclusively to that with neutrons and will be proportional to @xmath124 , the number density of neutrons .
it goes without saying that it will be equal to all flavours .
this time , we find that @xmath125 as was the case before , for @xmath116 , this contribution will flip sign if we replace the neutrinos by anti - neutrinos .
but if , as we said , the standard model interactions do not change neutrino flavour , neutrino flavour transitions or neutrino oscillations point undoubtedly to neutrino mass and mixing even when neutrinos are propagating through matter . unless non - standard - model flavour changing interactions play a role .
neutrino propagation in matter is easy to understand when analysed through the time dependent schrdinger equation in the lab frame @xmath126 where , @xmath127 is a ( three component ) neutrino vector state , in which each neutrino flavour corresponds to one component . in the same way , the hamiltonian @xmath128 is a ( tree @xmath113 three ) matrix in flavour space . to make our lives easy , lets analyse the case where only two neutrino flavours are relevant , say @xmath25 and @xmath27 . then @xmath129 with @xmath130 the amplitude of the neutrino to be a @xmath17 at time @xmath52 .
this time the hamiltonian , @xmath128 , is a 2@xmath1132 matrix in neutrino flavour space , @xmath131 space . it will prove to be clarifying to work out the two flavour case in vacuum first , and add matter effects afterwards . using eq .
( [ eq1 ] ) to express @xmath132 as a linear combination of mass eigenstates , we can see that the @xmath133 matrix element of the hamiltonian in vacuum , @xmath134 , can be written as @xmath135 where we are supposing that the neutrinos belong to a beam where all its mass components ( the mass eigenstates ) share the same definite momentum @xmath136 .
as we have already mentioned , despite this supposition being technically wrong , it leads anyway to the right transition amplitude . in the second line of eq .
( [ eq24 ] ) , we have used that the neutrinos @xmath137 with momentum @xmath136 , the mass eigenstates , are the asymptotic states of the hamiltonian , @xmath134 for which constitute an orthonormal basis , satisfy @xmath138 and have the standard dispersion relation , @xmath139 .
as we have already mentioned , neutrino oscillations are the archetype quantum interference phenomenon , where only the _ relative _ phases of the interfering states play a role .
therefore , only the _ relative _ energies of these states , which set their relative phases , are relevant . as a consequence ,
if it proves to be convenient ( and it will ) , we can feel free to happily remove from the hamiltonian @xmath128 any contribution proportional to the identity matrix @xmath140 .
as we have said , this subtraction will leave unaffected the differences between the eigenvalues of @xmath128 , and therefore will leave unaffected the prediction of @xmath128 for flavour transitions .
it goes without saying that as in this case only two neutrinos are relevant , there are only two mass eigenstates , @xmath109 and @xmath110 , and only one mass splitting @xmath141 , and therefore there should be , as before a unitary @xmath23 matrix given by eq .
( [ eq18 ] ) which rotates from one basis to the other . inserting it into eq .
( [ eq24 ] ) , and assuming that our neutrinos have low masses as compared to their momenta , @xmath142 , and removing from @xmath134 a term proportional to the the identity matrix ( a removal we know is going to be harmless ) , we get @xmath143 to write this expression , the highly relativistic approximation , which says that @xmath144 is used . where @xmath64 is the average energy of the neutrino mass eigenstates in our neutrino beam of ultra high momentum @xmath136 .
it is not difficult to corroborate that the hamiltonian @xmath134 of eq .
( [ eq25 ] ) for the two neutrino scenario would give an identical oscillation probability , eq .
( [ eq19 ] ) , as the one we have already obtained in a different way .
an easy way to do it is to analyse the transition probability for the process @xmath145 . from eq .
( [ eq18 ] ) it is clear that in terms of the mixing angle , the electron neutrino state composition is @xmath146 while that of the muon neutrino is given by @xmath147 in the same way , we can also write the eigenvalues of the vacuum hamiltonian @xmath134 , , in terms of the mass squared differences as @xmath148 the mass eigenbasis of this hamiltonian , @xmath149 and @xmath150 , can also be written in terms of flavour eigenbasis @xmath151 and @xmath152 by means of eqs .
( [ eq26 ] ) and ( [ eq27 ] ) .
therefore , the schrdinger equation of eq .
( [ eq22 ] ) , with the identification of @xmath128 in this case with @xmath134 tells us that if at time @xmath153 we begin from a @xmath151 , then once some time @xmath52 elapses this @xmath151 will progress into the state given by @xmath154 thus , the probability p@xmath155 that this evoluted neutrino be detected as a different flavour @xmath27 , from eqs .
( [ eq27 ] ) and ( [ eq29 ] ) , is given by , @xmath156 where we have substituted the time @xmath52 travelled by our highly relativistic state by the distance @xmath33 it has covered . the flavour transition or oscillation probability of eq .
( [ eq30 ] ) , as expected , is exactly the same we have found before , eq .
( [ eq19 ] ) .
we can now move on to analyse neutrino propagation in matter . in this case , the 2@xmath1132 hamiltonian representing the propagation in vacuum @xmath134 receives the two additional contributions we have discussed before , and becomes @xmath157 , which is given by @xmath158 in the new hamiltonian , the first additional contribution corresponds to the interaction potential due to the charged bosons exchange , eq .
( [ eq20 ] ) . as this interaction
is suffered only by @xmath25 , this contribution is different from zero only in the @xmath157(1,1 ) element or the @xmath159 element .
the second additional contribution , the last term of eq .
( [ eq31 ] ) comes from the @xmath122 boson exchange , eq .
( [ eq21 ] ) .
since this interaction is flavour blind , it affects every neutrino flavour in the same way , its contribution to @xmath157 is proportional to the identity matrix , and can be safely neglected .
thus @xmath160 where ( for reasons that are going to become clear later ) we have divided the @xmath7-exchange contribution into two pieces , one proportional to the identity ( that we will disregarded in the next step ) and , a piece that it is not proportional to the identity , that we will keep .
disregarding the first piece as promised , we have from eqs .
( [ eq25 ] ) and ( [ eq32 ] ) @xmath161 where we have defined @xmath162 clearly , @xmath163 parameterizes the relative size of the matter effects as compared to the vacuum contribution given by the neutrino squared - mass splitting and signals the situations when they become important .
now , if we introduce ( a physically meaningful ) short - hand notation @xmath164 and @xmath165 then the hamiltonian in a medium @xmath157 turns out to be @xmath166 and can be diagonalised by inspection , as a result of our choice , the hamiltonian in a medium , @xmath157 , becomes formally indistinguishable to the vacuum one , @xmath134 , eq .
( [ eq25 ] ) .
the difference being that in this case what used to be the vacuum parameters @xmath88 and @xmath114 are presently given by the matter ones , @xmath167 and @xmath168 , respectively . obviously , the mass eigenstates and eigenvalues ( which determine the mixing angle ) of @xmath157 are not identical to the ones in vacuum .
the mass squared difference of the matter eigenstates is not the same as the vacuum @xmath88 , and the same happens with the mixing angle .
the eigenstates in matter , _ ie . _
the files of the unitary matrix that rotates from the flavour basis to the mass basis , are different from the vacuum eigenvalues that form the vacuum mixing matrix , and therefore @xmath169 is not @xmath114 .
but , the matter hamiltonian @xmath157 does indeed contain all about the propagation of neutrinos in matter , in the same way @xmath134 contains all about the propagation in vacuum . according to eq .
( [ eq37 ] ) , @xmath157 has the same functional dependence on the matter parameters @xmath167 and @xmath168 as the vacuum hamiltonian @xmath134 , eq .
( [ eq25 ] ) , on the vacuum ones , @xmath88 and @xmath114 .
therefore , @xmath167 can be identified with an effective mass squared difference in matter , and accordingly @xmath168 can be unidentified with an effective mixing angle in matter . in a typical experimental set - up where the neutrino beam is generated by an accelerator and sent away to a detector that is , say , several hundred , or even thousand kilometres away
, it traverses through earth matter , but only superficially , it does not get deep into the earth . then , during this voyage the matter density encountered by such a beam can be taken to be approximately constant .
but if the density of the earth s matter is constant , the same happens with the electron density @xmath119 , and the @xmath163 parameter in which it is incorporated , which after all is determined by it .
and it is also true about the hamiltonian @xmath157 .
they all become approximately constant , and therefore quite identical to the vacuum hamiltonian @xmath134 , except for the particular values of their parameters . by comparing eqs .
( [ eq37 ] ) and ( [ eq25 ] ) , we can immediately conclude that exactly in the same way @xmath134 gives rise to vacuum oscillations with probability p@xmath155 of eq .
( [ eq30 ] ) , @xmath157 must give rise to matter oscillations with probability @xmath170 namely , the transition and survival probabilities in matter are the same as those in vacuum , except that the vacuum parameters @xmath88 and @xmath114 are now replaced by their matter counterparts , @xmath167 and @xmath168 . in theory ,
judging simply by its potential , matter effects can have very drastic repercussions in the oscillation probabilities . the exact impact ( if any ) can be estimated only after the details of the experimental set - up of the experiment in question are given . as a rule of thumb , to guess the importance of matter effects
, we should keep in mind that for neutrinos propagating through the earth s mantle ( not deeper than 200 km below the surface ) and if the kinematic phase associated to the solar mass difference is still negligible , @xmath171 so that only for beam energies of several gev matter effects do matter . and how much do they matter ? they matter a lot ! from eq .
( [ eq36 ] ) for the matter mixing angle , @xmath168 , we can appreciate that even when the vacuum mixing angle @xmath114 is incredible small , say , @xmath172 , if we get to have @xmath173 , for energies of a few tens of gev , then @xmath174 can be brutally enhanced as compared to its vacuum value and can even reach maximal mixing , _ ie .
_ @xmath175 .
this wild enhancement of a small mixing angle in vacuum up to a sizeable ( even maximal ) one in matter is the `` resonant '' enhancement , the largest possible version of the mikheyev - smirnov - wolfenstein effect @xcite . in the beginning of solar neutrino experiments
, people entertained the idea that this brutal enhancement was actually taking place while neutrinos crossed the sun .
nonetheless , as we will see soon the mixing angle associated with solar neutrinos is quite sizeable ( @xmath176 ) already in vacuum @xcite . then , although matter effects on the sun are important and they do enhance the solar mixing angle , unfortunately they are not as drastic as we once dreamt .
nevertheless , for long - baselines they will play ( they are already playing ! ) a key role in the determination of the ordering of the neutrino spectrum .
almost twenty years have elapsed since we were presented solid and convincing evidence of neutrino masses and mixings , and since then , the evidence has only grown .
superkamiokande ( sk ) was the first experiment to present compelling evidence of @xmath27 disappearance in their atmospheric neutrino fluxes , see @xcite . in fig .
[ f1 ] the zenith angle ( the angle subtended with the horizontal ) dependence of the multi - gev @xmath27 sample is shown together with the disappearance as a function of @xmath98 plot .
these data fit amazingly well the naive two component neutrino hypothesis with @xmath177 roughly speaking sk corresponds to an @xmath98 for oscillations of 500 km / gev and almost maximal mixing ( the mass eigenstates are nearly even admixtures of muon and tau neutrinos ) .
no signal of an involvement of the third flavour , @xmath25 is found so the assumption is that atmospheric neutrino disappearance is basically @xmath178 .
notice however , that the first nova results seem to point toward a mixing angle which is not maximal ( excluding maximal mixing at the 2 sigma level ) .
after atmospheric neutrino oscillations were established , a new series of neutrino experiments were built , sending ( man - made ) beams of @xmath27 neutrinos to detectors located at large distances : the k2k ( t2k ) experiment @xcite , sends neutrinos from the kek accelerator complex to the old sk mine , with a baseline of 120 ( 235 ) km while the minos ( nova ) experiment @xcite , sends its beam from fermilab , near chicago , to the soudan mine ( ash river ) in minnesota , a baseline of 735 ( 810 ) km .
all these experiments have seen evidence for @xmath27 disappearance consistent with the one found by sk .
their results are summarised in fig .
[ f2 ] . vs @xmath179 plane for minos data as well as for t2k data and two of the sk analyses .
minos s best fit point is at @xmath180 and @xmath181 .
notice that new nova data seem to exclude maximal mixing at the 2 sigma level , width=340 ] the kamland reactor experiment , an antineutrino disappearance experiment , receiving neutrinos from sixteen different reactors , at distances ranging from hundred to thousand kilometres , with an average baseline of 180 km and neutrinos of a few ev , @xcite , has seen evidence of neutrino oscillations .
such evidence was collected not only at a different @xmath98 than the atmospheric and accelerator experiments but also consists on oscillations involving electron neutrinos , @xmath25 , the ones which were not involved before .
these oscillations have also been seen for neutrinos coming from the sun ( the sun produces only electron neutrinos ) .
however , in order to compare the two experiments we should assume that neutrinos ( solar ) and antineutrinos ( reactor ) behave in the same way , _ ie .
_ assume cpt conservation .
the best fit values in the two neutrino scenario for the kamland experiment are @xmath182 in this case , the @xmath98 involved is 15 km / mev which is more than an order of magnitude larger than the atmospheric scale and the mixing angle , although large , is clearly not maximal .
[ f3 ] shows the disappearance probability for the @xmath183 for kamland as well as several older reactor experiments with shorter baselines.the second panel depicts the flavour content of the @xmath184boron solar neutrino flux ( with gev energies ) measured by sno , @xcite , and sk , @xcite .
the reactor outcome can be explained in terms of two flavour oscillations in vacuum , given that the fit to the disappearance probability , is appropriately averaged over @xmath64 and @xmath33 .. observed by reactor experiments as a function of distance from the reactor .
the flavour content of the @xmath185boron solar neutrinos for the various reactions for sno and sk .
cc : @xmath186 , nc:@xmath187 and es : @xmath188 , title="fig:",width=226 ] observed by reactor experiments as a function of distance from the reactor .
the flavour content of the @xmath185boron solar neutrinos for the various reactions for sno and sk .
cc : @xmath186 , nc:@xmath187 and es : @xmath188 , title="fig:",width=226 ] the analysis of neutrinos originating from the sun is marginally more complex that the one we did before because it should incorporate the matter effects that the neutrinos endure since they are born ( at the centre of the sun ) until they abandon it , which are imperative at least for the @xmath184boron neutrinos .
the pp and @xmath189be neutrinos are less energetic and therefore are not significantly altered by the presence of matter and leave the sun as though it were ethereal .
@xmath184boron neutrinos on the other hand , leave the sun unequivocally influenced by the presence of matter and this is evidenced by the fact that they leave the sun as @xmath110 , the second mass eigenstate and therefore do not experience oscillations .
this distinction among neutrinos coming from different reaction chains is , as mentioned , due mainly to their disparities at birth .
while pp ( @xmath189be ) neutrinos are created with an average energy of 0.2 mev ( 0.9 mev ) , @xmath184b are born with 10 mev and as we have seen the impact of matter effects grows with the energy of the neutrino .
however , we ought to emphasise that we do not really see solar neutrino oscillations . to trace the oscillation pattern , to be able to test
is distinctive shape , we need a kinematic phase of order one otherwise the oscillations either do not develop or get averaged to 1/2 . in the case of neutrinos coming from the sun the kinematic phase is @xmath190 consequently , solar neutrinos behave as `` effectively incoherent '' mass eigenstates once they leave the sun , and remain so once they reach the earth . consequently the @xmath25 disappearance or survival probability is given by @xmath191 where @xmath192 is the @xmath109 content or fraction of @xmath27 and @xmath193 is the @xmath110 content of @xmath27 and therefore both fractions satisfy @xmath194 nevertheless , as we have already mentioned , solar neutrinos originating from the pp and @xmath189be chains are not affected by the solar matter and oscillate as in vacuum and thus , in their case @xmath195 and @xmath196 . in the @xmath184b a neutrino case , however , the impact of solar matter is sizeable and the corresponding fractions are substantially altered , see fig .
[ f4 ] . in the core but
once they exit the sun thinking about them in the mass eigenstate basis is useful .
the fraction of @xmath109 and @xmath110 is energy dependent above 1 mev and has a dramatic effect on the @xmath184boron solar neutrinos , as first observed by davis.,title="fig:",width=264 ] in the core but once they exit the sun thinking about them in the mass eigenstate basis is useful .
the fraction of @xmath109 and @xmath110 is energy dependent above 1 mev and has a dramatic effect on the @xmath184boron solar neutrinos , as first observed by davis.,title="fig:",width=264 ] in a two neutrino scenario , the day - time cc / nc measured by sno , which is roughly identical to the day - time average @xmath25 survival probability , @xmath197 , reads @xmath198 where @xmath192 and @xmath199 are the @xmath109 and @xmath110 contents of the muon neutrino , respectively , averaged over the @xmath184b neutrino energy spectrum appropriately weighted with the charged current current cross section .
therefore , the @xmath109 fraction ( or how much @xmath193 differs from 100% ) is given by @xmath200 where the central values of the last sno analysis , @xcite , were used .
as there are strong correlations between the uncertainties of the cc / nc ratio and @xmath201 it is not obvious how to estimate the uncertainty on @xmath192 from their analysis .
note , that if the fraction of @xmath110 were 100% , then @xmath202 . utilising
the analytic analysis of the mikheyev - smirnov - wolfenstein ( msw ) effect , gave in @xcite , one can obtain the mass eigenstate fractions in a medium , which are given by @xmath203 with @xmath204 being the mixing angle as given at the @xmath25 production point and @xmath205 is the probability of the neutrino to hop from one mass eigenstate to the second one during the mikheyev - smirnov resonance crossing .
the average @xmath206 is over the electron density of the @xmath184b @xmath25 production region in the centre of the sun as given by the solar standard model and the energy spectrum of @xmath184b neutrinos has been appropriately weighted with sno s charged current cross section .
all in all , the @xmath184b energy weighted average content of @xmath110 s measured by sno is @xmath207 therefore , it is obvious that the @xmath184b solar neutrinos are the purest mass eigenstate neutrino beam known so far and sk super famous picture of the sun taken ( from underground ) with neutrinos is made with approximately 90% of @xmath110 , _ ie . _
almost a pure beam of mass eigenstates . on march 8 , 2012 a newly built reactor neutrino experiment , the daya bay experiment , located in china ,
announced the measurement of the third mixing angle @xcite , the only one which was still missing and found it to be @xmath208 following this announcement , several experiments confirmed the finding and during the last years the last mixing angle to be measured became the best ( most precisely ) measured one .
the fact that this angle , although smaller that the other two , is still sizeable opens the door to a new generation of neutrino experiments aiming to answer the open questions in the field .
now that we have comprehended the physics behind neutrinos oscillations and have leaned the experimental evidence about the parameters driving these oscillations , we can move ahead and construct the neutrino standard model : * it comprises three light ( @xmath5 @xmath210 1 ev ) neutrinos , _ ie . _
it involves just two mass differences + @xmath211 and @xmath212 .
* so far we have not seen any solid experimental indication ( or need ) for additional neutrinos .
as we have measured long time ago the invisible width of the @xmath122 boson and found it to be 3 , within errors , if additional neutrinos are going to be incorporated into the model , they can not couple to the @xmath122 boson , _ ie .
_ they can not enjoy weak interactions , so we call them sterile . however , as sterile neutrinos have not been seen ( although they may have been hinted ) , and are not needed to explain any solid experimental evidence , our neutrino standard model will contain just the three active flavours : @xmath26 , @xmath38 and @xmath9 .
+ ) .if cpt is conserved , the flavour content must be the same for neutrinos and anti - neutrinos .
notice that oscillation experiments can not tell us how far above zero the entire spectrum lies.,width=377 ] * the unitary mixing matrix which rotates from the flavour to the mass basis , called the pmns matrix , comprises three mixing angles ( the so called solar mixing angle:@xmath213 , the atmospheric mixing angle @xmath214 , and the last to be measured , the reactor mixing angle@xmath215 ) , one dirac phase ( @xmath216 ) and potentially two majorana phases ( @xmath24 , @xmath217 ) and is given by @xmath218 @xmath219 where @xmath220 and @xmath221 .
courtesy of the hierarchy in mass differences ( and to a less extent to the smallness of the reactor mixing angle ) we are permitted to recognise the ( 23 ) label in the three neutrino scenario as the atmospheric @xmath222 we obtained in the two neutrino scenario , in a similar fashion the ( 12 ) label can be assimilated to the solar @xmath223 .
the ( 13 ) sector drives the @xmath25 flavour oscillations at the atmospheric scale , and the depletion in reactor neutrino fluxes see @xcite . according to the experiments done so far , the three sigma ranges for the neutrino mixing angles are @xmath224 while the corresponding ones for the mass splittings are @xmath225 and @xmath226 these mixing angles and mass splittings are summarised in fig .
[ f5 ] . * as oscillation experiments only explore the two mass differences , two ordering are possible , as shown in fig .
they are called normal and inverted hierarchy and roughly identify whether the mass eigenstate with the smaller electron neutrino content is the lightest or the heaviest . * the absolute mass scale of the neutrinos , or the mass of the lightest neutrino is not know yet , but cosmological bounds already say that the heaviest one must be lighter than about .5 ev . *
as transition or survival probabilities depend on the combination @xmath227 no trace of the majorana phases could appear on oscillation phenomena , however they will have observable effects in those processes where the majorana character of the neutrino is essential for the process to happen , like neutrino - less double beta decay .
the absolute mass scale of the neutrino , _ ie . _ the mass of the lightest / heaviest neutrino , can not be obtained from oscillation experiments , however this does not mean we have no access to it .
direct experiments like tritium beta decay , or neutrinoless double beta decay and indirect ones , like cosmological observations , have potential to feed us the information on the absolute scale of neutrino mass , we so desperately need .
the katrin tritium beta decay experiment , @xcite , has sensitivity down to 200 mev for the `` mass '' of @xmath25 defined as @xmath228 decay , in cosmology and in tritium @xmath217 decay versus the mass of the lightest neutrino .
below the dashed lines , only the normal hierarchy is allowed .
notice that while double @xmath35 decay experiments bound the neutrino mass only in the majorana case , planck bounds apply for either case , title="fig:",width=264 ] decay , in cosmology and in tritium @xmath217 decay versus the mass of the lightest neutrino .
below the dashed lines , only the normal hierarchy is allowed .
notice that while double @xmath35 decay experiments bound the neutrino mass only in the majorana case , planck bounds apply for either case , title="fig:",width=264 ] decay , in cosmology and in tritium @xmath217 decay versus the mass of the lightest neutrino .
below the dashed lines , only the normal hierarchy is allowed .
notice that while double @xmath35 decay experiments bound the neutrino mass only in the majorana case , planck bounds apply for either case , title="fig:",width=264 ] neutrino - less double beta decay experiments , see @xcite for a review , do not measure the absolute mass of the neutrino directly but a particular combination of neutrino masses and mixings , @xmath229 where it is understood that neutrinos are taken to be majorana particles , _ ie .
_ truly neutral particles ( having all their quantum numbers to be zero ) .
the new generation of experiments seeks to reach below 10 mev for @xmath230 in double beta decay .
cosmological probes ( cmb and large scale structure experiments ) measure the sum of the neutrino masses @xmath231 and may have a say on the mass ordering ( direct or inverted spectrum ) as well as test other neutrino properties like neutrino asymmetries@xcite . if @xmath232 ev , the energy balance of the universe saturates the bound coming from its critical density .
the current limit , @xcite , is a few % of this number , @xmath233 ev . these bounds are model dependent but they do all give numbers of the same order of magnitude . however , given the systematic uncertainties characteristic of cosmology , a solid limit of less that 100 mev seems way too aggressive .
[ f6 ] shows the allowed parameter space for the neutrino masses ( as a function of the absolute scale ) for both the normal and inverted hierarchy .
a fermion mass is nothing but a coupling between a left handed state and a right handed one .
thus , if we examine a massive fermion at rest , then one can regard this state as a linear combination of two massless particles , one right handed and one left handed .
if the particle we are examining is electrically charged , like an electron or a muon , both particles , the left handed as well as the right handed must have the same charge ( we want the mass term to be electrically neutral ) .
this is a dirac mass term .
however , for a neutral particle , like a sterile neutrino , a new possibility opens up , the left handed particle can be coupled to the right handed anti - particle , ( a term which would have a net charge , if the fields are not absolutely and totally neutral ) this is a majorana mass term . thus a truly and absolutely neutral particle ( who will inevitably be its own antiparticle ) does have two ways of getting a mass term , a la dirac or a la majorana , and if there are no reasons to forbid one of them , will have them both , as shown in fig .
[ f7 ] .
[ f7 ] in the case of a neutrino , the left chiral field couples to @xmath234 implying that a majorana mass term is forbidden by gauge symmetry . however , the right chiral field carries no quantum numbers , is totally and absolutely neutral .
then , the majorana mass term is unprotected by any symmetry and it is expected to be very large , of the order of the largest scale in the theory . on the other hand , dirac mass terms are expected to be of the order of the electroweak scale times a yukawa coupling , giving a mass of the order of magnitude of the charged lepton or quark masses . putting all the pieces together , the mass matrix for the neutrinos results as in fig .
[ f8 ] . is the dirac mass terms while 0 and @xmath235 are majorana masses for the charged and uncharged ( under @xmath236 ) chiral components , width=264 ] to get the mass eigenstates we need to diagonalise the neutrino mass matrix . by doing so ,
one is left with two majorana neutrinos , one super - heavy majorana neutrino with mass @xmath237 and one super - light majorana neutrino with mass @xmath238 , _ ie .
_ one mass goes up while the other sinks , this is what we call the seesaw mechanism , @xcite . the light neutrino(s )
is(are ) the one(s ) observed in current experiments ( its mass differences ) while the heavy neutrino(s ) are not accessible to current experiments and could be responsible for explaining the baryon asymmetry of the universe through the generation of a lepton asymmetry at very high energy scales since its decays can in principle be cp violating ( they depend on the two majorana phases on the pnms matrix which are invisible for oscillations ) .
the super heavy majorana neutrinos being their masses so large can play a role at very high energies and can be related to inflation @xcite .
if neutrinos are majorana particles lepton number is no longer a good quantum number and a plethora of new processes forbidden by lepton number conservation can take place , it is not only neutrino - less double beta decay .
for example , a muon neutrino can produce a positively charged muon .
however , this process and any processes of this kind , would be suppressed by @xmath239 which is tiny , @xmath240 , and therefore , although they are technically allowed , are experimentally unobservable
. to most stringent limit nowadays comes from kamland - zen @xcite , and constraints the half - life of neutrino - less double beta decay to be @xmath241 years at 90% c.l .
forthcoming experiments such as gerda - phaseii , majorana , supernemo , cuore , and nexo will improve this sensitivity by one order of magnitude .
recently low energy sew saw models @xcite have experienced a revival and are actively being explored @xcite . in such models
the heavy states , of only few tens of tev can be searched for at the lhc .
the heavy right handed states in these models will be produced at lhc either through yukawa couplings of through gauge coupling to right handed gauge bosons .
some models contain also additional scalar that can be looked for .
the experimental observations of neutrino oscillations , meaning that neutrinos have mass and mix , answered questions that had endured since the establishment of the standard model .
as those veils have disappeared , new questions open up and challenge our understanding * what is the true nature of the neutrinos ? are they majorana particles or dirac ones ?
are neutrinos totally neutral ? * is there any new scale associated to neutrinos masses ?
can it be accessible at colliders ? * is the spectrum normal or inverted ? is the lightest neutrino the one with the least electron content on it , or is it the heaviest one ?
* is cp violated ( is @xmath242 ) ?
if so , is this phase related at any rate with the baryon asymmetry of the universe ?
what about the other two phases ?
* which is the absolute mass scale of the neutrinos ? * are there new interactions ?
are neutrinos related to the open questions in cosmology , like dark matter and/or dark energy ?
do ( presumably heavy ) neutrinos play a role in inflation ? * can neutrinos violate cpt @xcite ?
what about lorentz invariance ?
* if we ever measure a different spectrum for neutrinos and antineutrinos ( after matter effects are properly taken into account ) , how can we distinguish whether it is due to a true ( genuine ) ctp violation or to a non - standard neutrino interaction ? * are these intriguing signals in short baseline reactor neutrino experiments ( the missing fluxes ) a real effect ?
do they imply the existence of sterile neutrinos ?
we would like to answer these questions . for doing it , we are doing right now , and we plan to do new experiments .
these experiments will , for sure bring some answers and clearly open new , pressing questions .
only one thing is clear .
our journey into the neutrino world is just beginning .
i would like to thanks the students and the organisers of the european school on hep for giving me the opportunity to present these lectures in such a wonderful atmosphere ( and for pampering me beyond my expectations , which were not low ) .
i did enjoyed each day of the school enormously .
support from the mec and feder ( ec ) grants sev-2014 - 0398 and fpa2014 - 54459 and the generalitat valenciana under grant prometeoii/2013/017 is acknowledged .
the author s work has received funding from the european union horizon 2020 research and innovation programme under the marie sklodowska - curie grant elusives itn agreement no 674896 and invisiblesplus rise , agreement no 690575 .
an analysis of cpt violation in the neutrino sector can be found in : g. barenboim and j. d. lykken , phys .
b * 554 * ( 2003 ) 73 [ arxiv : hep - ph/0210411 ] ; g. barenboim , j. f. beacom , l. borissov and b. kayser , phys . lett .
b * 537 * ( 2002 ) 227 [ arxiv : hep - ph/0203261 ] .
s. p. mikheev and a. y. smirnov ,
neutrino oscillations in a variable - density medium and @xmath243 bursts due sov .
jetp * 64 * , 4 ( 1986 ) [ zh .
91 * , 7 ( 1986 ) ] [ arxiv:0706.0454 [ hep - ph ] ] .
y. ashie _ et al . _
[ super - kamiokande collaboration ] , a measurement of atmospheric neutrino oscillation parameters by phys .
d * 71 * , 112005 ( 2005 ) [ arxiv : hep - ex/0501064 ] ; y. takeuchi [ super - kamiokande collaboration ] , arxiv:1112.3425 [ hep - ex ] .
m. h. ahn _
[ k2k collaboration ] , phys .
d * 74 * ( 2006 ) 072003 [ hep - ex/0606032 ] ; c. mariani [ k2k collaboration ] , aip conf . proc .
* 981 * ( 2008 ) 247 .
doi:10.1063/1.2898949 m. scott [ t2k collaboration ] , arxiv:1606.01217 [ hep - ex ] .
a. b. sousa [ minos and minos+ collaborations ] , aip conf .
* 1666 * ( 2015 ) 110004 doi:10.1063/1.4915576 [ arxiv:1502.07715 [ hep - ex ] ] ; l. h. whitehead [ minos collaboration ] , nucl .
b * 908 * ( 2016 ) 130 doi:10.1016/j.nuclphysb.2016.03.004 [ arxiv:1601.05233 [ hep - ex ] ] .
p. adamson _ et al .
_ [ nova collaboration ] , phys . rev .
d * 93 * ( 2016 ) no.5 , 051104 doi:10.1103/physrevd.93.051104 [ arxiv:1601.05037 [ hep - ex ] ] .
i. shimizu [ kamland collaboration ] , nucl .
* 188 * ( 2009 ) 84 .
doi:10.1016/j.nuclphysbps.2009.02.020 m. p.
decowski [ kamland collaboration ] , nucl .
b * 908 * ( 2016 ) 52 .
doi:10.1016/j.nuclphysb.2016.04.014 b. aharmim
_ et al . _
[ sno collaboration ] , phys .
c * 88 * ( 2013 ) 025501 doi:10.1103/physrevc.88.025501 [ arxiv:1109.0763 [ nucl - ex ] ] .
a. hime , nucl .
phys . proc .
* 221 * ( 2011 ) 110 .
doi:10.1016/j.nuclphysbps.2011.03.104 s. j. parke and t. p. walker , phys .
lett . * 57 * , 2322 ( 1986 ) [ erratum - ibid . *
57 * , 3124 ( 1986 ) ] .
an _ et al . _
[ daya bay collaboration ] , arxiv:1610.04802 [ hep - ex ] ; s. b. kim [ reno collaboration ] , nucl .
b * 908 * ( 2016 ) 94 ; y. abe _ et al .
_ [ double chooz collaboration ] , jhep * 1410 * ( 2014 ) 086 erratum : [ jhep * 1502 * ( 2015 ) 074 ] [ arxiv:1406.7763 [ hep - ex ] ] .
m. lattanzi [ planck collaboration ] , j. phys .
* 718 * ( 2016 ) no.3 , 032008 .
doi:10.1088/1742 - 6596/718/3/032008 m. gell - mann , p. ramond and r. slansky , in _
supergravity _ , edited by p.van nieuwenhuizen and d. freedman , ( north - holland,1979 ) , p.315 .
g. barenboim , jhep * 0903 * ( 2009 ) 102 [ arxiv:0811.2998 [ hep - ph ] ] ; g. barenboim , phys .
d * 82 * ( 2010 ) 093014 [ arxiv:1009.2504 [ hep - ph ] ] .
y. gando [ kamland - zen collaboration ] , nucl .
* 273 - 275 * ( 2016 ) 1842 .
doi:10.1016/j.nuclphysbps.2015.09.297 f. borzumati and y. nomura , phys .
d * 64 * ( 2001 ) 053005 doi:10.1103/physrevd.64.053005 [ hep - ph/0007018 ] . c. g. cely , a. ibarra , e. molinaro and s. t. petcov , phys .
b * 718 * ( 2013 ) 957 doi:10.1016/j.physletb.2012.11.026 [ arxiv:1208.3654 [ hep - ph ] ] .
g. barenboim , l. borissov , j. d. lykken and a. y. smirnov , jhep * 0210 * ( 2002 ) 001 [ hep - ph/0108199 ] .
g. barenboim and j. d. lykken , phys .
d * 80 * ( 2009 ) 113008 [ arxiv:0908.2993 [ hep - ph ] ] . | the standard model has been effective way beyond expectations in foreseeing the result of almost all the experimental tests done up so far . in it
, neutrinos are massless .
nonetheless , in recent years we have collected solid proofs indicating little but non zero masses for the neutrinos ( when contrasted with those of the charged leptons ) .
these masses permit neutrinos to change their flavor and oscillate , indeed a unique treat . in these lectures ,
i discuss the properties and the amazing potential of neutrinos in and beyond the standard model . |
* synthesis of planar [ cu(pyz)@xmath12(pyo)@xmath12](pf@xmath14)@xmath12*. in the minimum amount of water , cucl@xmath582h@xmath2o ( 0.502 g , 2.95 mmol ) and agpf@xmath4 ( 1.489 g , 5.89 mmol ) were mixed together to yield a white precipitate of agcl and a pale blue solution containing the `` cu(pf@xmath4)@xmath2 '' product . to recover the blue solution , the agcl solid
was removed by careful vacuum filtration . in a separate solution , pyrazine and pyridine-@xmath5-oxide in a 3:1 ratio
were dissolved in a 1:1 mixture of h@xmath2o and ethanol to give a colourless solution . to this latter solution
was added the `` cu(pf@xmath4)@xmath2 '' solution . upon slow mixing of the chemical reagents , a small amount of pale blue powder formed , but was removed by vacuum filtration to afford a blue solution .
slow evaporation of this solution yielded a large mass of deep green square plates .
* synthesis of the chain - like compound [ cu(pyz)(pyo)@xmath12(h@xmath12o)@xmath12](pf@xmath14)@xmath12*. using a reaction scheme similar to that above , the ratio of of pyrazine to pyridine-@xmath5-oxide was decreased from 3:1 to 2:1 .
mixing of the chemical reagents immediately afforded a bright turquoise blue solution .
the solution was covered with a piece of perforated al - foil to allow slow solvent evaporation . upon standing at room temperature for several weeks
, small green blocks were deposited on the bottom of the beaker and subsequently isolated via vacuum filtration .
.room temperature lattice parameters determined from x - ray crystallography of [ cu(pyz)@xmath2(pyo)@xmath2](pf@xmath4)@xmath2 ( q2d ) and [ cu(pyz)(pyo)@xmath2(h@xmath2o)@xmath2](pf@xmath4)@xmath2 ( q1d ) . *
note that this value is the perpendicular distance between cu ions .
[ cols="<,^,>",options="header " , ] * quantum monte carlo simulations . *
the numerical results were obtained using the stochastic series expansion ( sse ) method to simulate the hamiltonian of equation 1 on finite - sized lattices .
sse is a finite temperature qmc method based on the evaluation of the diagonal matrix elements of the density matrix in a suitable basis .
the method is explained in detail in ref .
@xcite and references therein .
the simulations were performed on the nersc computing facility in berkeley , california
. * temperature - dependent susceptibility .
* measurements were performed using a quantum design physical properties measurement system equipped with a vibrating sample magnetometer option .
polycrystalline samples were placed into gelatin capsules and affixed to the end of a carbon fibre rod .
data were taken on warming from 2 k in a fixed magnetic field of 0.1 t. the datasets for the two compounds are quite distinct ( see fig .
[ chi ] ) , but both exhibit the broad hump in @xmath59 at @xmath60 expected for highly anisotropic antiferromagnets and caused by the onset of short - range correlations
. fits from 300 k down to temperatures just below the hump were made to theoretical models based on the low - dimensional heisenberg hamiltonian of equation 1 .
the layered compound is found to be in excellent agreement with the @xmath0 q2d model of woodward _ et al . _
@xcite with @xmath61 and @xmath62 k , while the best fit for the chain - like material is for the quasi - one - dimensional model of johnston _ et al . _
@xcite with @xmath63 and @xmath64 k. these values are in good agreement with those derived from the pulsed field magnetization .
we obtained similar results by also fitting to the earlier models of bonner & fisher @xcite ( q1d ) and lines @xcite ( q2d ) .
* electron - spin resonance .
* measurements of the @xmath31-factor anisotropy were performed in a superconducting magnet at the national high magnetic field laboratory in los alamos using a cavity perturbation technique @xcite .
the te102 mode of a rectangular resonator at 71.5 ghz , and a millimetre - wave vector network analyser manufactured by abmm were used .
the cavity was mounted on a goniometer and placed inside standard flow @xmath65he and single - shot @xmath66he cryostats . a temperature dependence ( down to 0.5 k ) of the spectrum along the principal axes
was performed to ensure the reported angular dependance of the resonance is not influenced by the long range order , but characteristic of the paramagnetic state .
for both compounds the data shown satisfy this condition . | gaining control of the building blocks of magnetic materials and thereby achieving particular characteristics will make possible the design and growth of bespoke magnetic devices .
while progress in the synthesis of molecular materials , and especially coordination polymers , represents a significant step towards this goal , the ability to tune the magnetic interactions within a particular framework remains in its infancy .
here we demonstrate a chemical method which achieves dimensionality selection via preferential inhibition of the magnetic exchange in an @xmath0 antiferromagnet along one crystal direction , switching the system from being quasi - two- to quasi - one - dimensional while effectively maintaining the nearest - neighbour coupling strength .
coordination polymers are self - organising materials consisting of arrays of metal ions linked via molecular ligands , with non - coordinated counterions supplying charge neutrality .
the choice of initial components permits a high level of control over the final product , enabling many different polymeric architectures to be obtained @xcite .
these materials provide a route to successful crystal engineering , and a number of functionalities are being actively studied , including gas storage @xcite , optoelectronic @xcite , ferroelectric @xcite and magnetic properties @xcite .
although it is now possible to generate an assortment of disparate magnetic lattices using this method @xcite , true control of magnetic exchange interactions implies an ability to adjust selected parameters while keeping others constant . to this end ,
a series of coordination polymers based on cu(ii ) ions bridged by pyrazine ( c@xmath1h@xmath1n@xmath2 ) molecules have proven to be highly versatile . in these systems
it has been shown that it is possible to alter significantly the primary exchange energies via adjustment of the ligands @xcite and the counterions @xcite , or fine - tune the exchange by a few percent via isotopic substitution @xcite , all the while maintaining the same basic metalpyrazine network . in this paper
we demonstrate the power of this strategy by chemically engineering a reduction in the dimensionality of a magnetic system .
after first designing a material based on coordinated planes of cu(ii ) , we adapt the recipe such that the ligand bridges are broken along a specific crystal direction , resulting in a chain - like compound . because the ligand mediating the magnetic interactions in both cases is unchanged , the nearest - neighbour exchange energies of the two materials are found to be equal to each other to within 5% .
the difference in numbers of nearest - neighbours , however , means that the strength of the combined exchange interactions acting on each magnetic ion in the quasi - two - dimensional material is twice that of its quasi - one - dimensional cousin
. figs .
[ fig1](a ) and ( b ) show the crystal structure of orthorhombic [ cu(pyz)@xmath2(pyo)@xmath2](pf@xmath4)@xmath2 ( where pyz = pyrazine and pyo = pyridine-@xmath5-oxide , c@xmath6h@xmath6no ) determined using single - crystal x - ray diffraction @xcite .
@xmath0 cu ions are linked by pyz molecules into nearly square planar arrays , with perpendicular non - bridging pyo ligands keeping the planes well - separated . because of the separation , as well as the staggered arrangement of adjacent planes shown in fig .
[ fig1](b ) , magnetic exchange energies are likely to be very small along the @xmath7-direction . in contrast , cupyzcu bridges are known to be good mediators of antiferromagnetic superexchange @xcite and so the magnetic properties of this material are expected to be quasi - two - dimensional .
this is confirmed by the magnetic measurements described below .
sample synthesis involves mixing together of the molecular components in a solution of water and ethanol .
intermolecular self - organisation means that only a small amount of intervention is subsequently required . to achieve the desired planar structure the pyz and pyo molecules
were added in a 3:1 ratio , previous experience suggesting that to account for the potential for pyo to substitute for pyz , the ligands must initially be in a proportion different to that found in the final product . in order to create a similar sample , but one based on cupyz chains rather than planes , we reduce the pyz : pyo ratio to 2:1 and proceed with the synthesis in a similar way .
the resulting material has the composition [ cu(pyz)(pyo)@xmath2(h@xmath2o)@xmath2](pf@xmath4)@xmath2 and the structure is shown in figs .
[ fig1](c ) and ( d ) . here
the pyz ligands link cu ions along the @xmath8-axis only , the other ligands being non - bridging pyo and water molecules .
the alteration in composition has the effect of changing the symmetry of the crystal from orthorhombic to monoclinic and reducing the number of formula units in the unit cell , but , most importantly for the magnetic exchange , the cupyzcu linkages along the @xmath9-axis are removed without altering the @xmath8-axis cu cu separation by more than a fraction of a percent ( @xmath10 for the planar material and @xmath11 for the chain compound ) .
sr ) spectra measured on planar [ cu(pyz)@xmath12(pyo)@xmath12](pf@xmath14)@xmath12 .
data at different temperatures are offset for clarity .
inset : evolution of the precession frequency @xmath15 with temperature .
long - range magnetic order is observed below 1.71 k. ( b ) @xmath16sr spectra measured on chain - like [ cu(pyz)(pyo)@xmath12(h@xmath12o)@xmath12](pf@xmath14)@xmath12 .
insets : the evolution of the amplitude @xmath17 ( left ) and relaxation rate @xmath18 ( right ) with temperature indicates the onset of long - range magnetic order below about 0.27 k. red lines are fits to functions described in the text and @xcite.,width=302 ] recent heat capacity measurements on the planar material in zero magnetic field see no evidence of a magnetic transition down to the lowest temperatures measured @xcite . however , thermodynamic probes are known to be less sensitive to transitions driven by interplanar couplings @xcite while local probes such as muon - spin relaxation ( @xmath16sr ) are much more effective at determining the antiferromagnetic transition temperature , @xmath19 @xcite . our @xmath20sr data on this compound , shown in fig .
[ muon](a ) , exhibits a clear precession signal which develops below @xmath21 k @xcite , demonstrating long - range magnetic order throughout the bulk of the sample .
in contrast , the @xmath16sr data for the chain - like compound exhibit no resolvable oscillations , see fig .
[ muon](b ) , probably due to a smaller ordered moment , but can be fitted to the expression @xmath22 , where @xmath17 represents the non - relaxing part of the signal .
both @xmath17 and @xmath18 rise markedly below @xmath23 k due to crossover from a regime in which the relaxation is dominated by dynamic magnetic fluctuations to one dominated by quasistatic magnetic order , see fig .
[ muon](b ) insets . from these fits
we estimate @xmath24 k for this material @xcite .
( pyo)@xmath12](pf@xmath14)@xmath12 at @xmath25 k , and chain - like [ cu(pyz)(pyo)@xmath12(h@xmath12o)@xmath12](pf@xmath14)@xmath12 at @xmath26 k. ( b ) the results of quantum monte carlo ( qmc ) simulations of the low temperature magnetization for 3d ( @xmath27 ) , 2d ( @xmath28 ) , and 1d ( @xmath29 ) antiferromagnets .
the lines are the pulsed - field data scaled by the saturation field ( @xmath30 ) .
( c ) the relation between the exchange anisotropy and the ratio of critical temperature and primary exchange energy in q1d and q2d antiferromagnets deduced from qmc simulations @xcite .
the circles indicate the materials reported here .
( d ) anisotropy of the @xmath31-factor in the planar and ( e ) chain - like compounds measured using esr at 10 k and 1.5k , respectively .
red lines are fits to @xmath32.,title="fig:",width=321 ] ( pyo)@xmath12](pf@xmath14)@xmath12 at @xmath25 k , and chain - like [ cu(pyz)(pyo)@xmath12(h@xmath12o)@xmath12](pf@xmath14)@xmath12 at @xmath26 k. ( b ) the results of quantum monte carlo ( qmc ) simulations of the low temperature magnetization for 3d ( @xmath27 ) , 2d ( @xmath28 ) , and 1d ( @xmath29 ) antiferromagnets .
the lines are the pulsed - field data scaled by the saturation field ( @xmath30 ) .
( c ) the relation between the exchange anisotropy and the ratio of critical temperature and primary exchange energy in q1d and q2d antiferromagnets deduced from qmc simulations @xcite .
the circles indicate the materials reported here . ( d ) anisotropy of the @xmath31-factor in the planar and ( e ) chain - like compounds measured using esr at 10 k and 1.5k , respectively .
red lines are fits to @xmath32.,title="fig:",width=321 ] the type of magnetic order displayed by the two compounds can be deduced from their low - temperature magnetization ( see fig . [ pulsed](a ) ) .
the form of our pulsed - field magnetization data up to saturation is in keeping with that expected for low - dimensional antiferromagnets .
the slightly concave curve exhibited by the planar material is typical of quasi - two - dimensional ( q2d ) antiferromagnetic interactions @xcite , while the more extreme curvature shown by data from the chain - like sample is indicative of a quasi - one - dimensional ( q1d ) magnetic lattice , where , for an ideal system , @xmath33 is known to diverge at the saturation field @xcite . to support these observations we compare the data with the results of low - temperature quantum monte carlo ( qmc ) simulations based on the hamiltonian @xmath34 here , for a q1d ( or q2d ) system , @xmath35 is the strength of the exchange coupling within the magnetic chains ( planes ) , @xmath36 is the coupling between chains ( planes ) , and the first and second summations refer to summing over unique pairs of nearest neighbours parallel and perpendicular to the chain ( plane ) , respectively .
comparisons of theory and data are shown in fig .
3(b ) .
theoretical magnetisation curves were calculated in finite steps of the @xmath37 parameter for both q1d and q2d magnetic lattices @xcite . for both materials curves corresponding to
the quoted values of @xmath37 gave the best matches with experimental data ; the q2d curve with @xmath38 for the planar sample , and the q1d curve with @xmath29 for the chain - like material .
the predicted curve for a three - dimensional antiferromagnet is also shown for contrast .
the deviation of the data from the simulations close to the saturation field is likely due to the finite temperatures at which the experiments were performed .
the saturation field ( @xmath30 ) can be extracted from the pulsed - field data and is found to be @xmath39 t for the planar material and @xmath40 t for the chain - like material . at saturation , the hamiltonian above implies that @xmath41 , where , for each spin , @xmath42 is the number of exchange bonds of interaction strength @xmath35 ( @xmath43 for q2d , 2 for q1d ) , and @xmath44 is the number of exchange bonds of interaction strength @xmath36 ( @xmath45 for q2d , 4 for q1d ) @xcite . from this , by assuming that @xmath46 and using the appropriate values for the @xmath31-factor determined from electron - spin resonance , we estimate the primary exchange couplings , @xmath35 , to be @xmath47 k for the planar compound and @xmath48 k for the chain - like compound .
another estimate of the magnetic dimensionality comes from the temperature dependence of the low - field magnetic susceptibility .
fits of such data for both materials results in estimates of @xmath35 that are in accord with those derived above from pulsed - field magnetization data @xcite .
the relative sizes of @xmath19 and @xmath35 are indicative of the anisotropy of the exchange interactions in a low - dimensional magnetic system . using quantum monte carlo ( qmc ) calculations yasuda _ et al . _
@xcite developed empirical relations between these values for @xmath49 heisenberg antiferromagnets , the results of which are shown in fig . [
pulsed](c ) . for compounds considered here the exchange anisotropies
are found to be @xmath50 in the planar material and @xmath51 in the chain - like material .
these values are in keeping with the comparison between data and simulation shown in fig .
[ pulsed](b ) .
the magnetic lattice in a low - dimensional system is not always obvious from an inspection of the crystal structure @xcite .
verification of the equivalence of the magnetic and structural planes and chains in our materials comes from the angle - dependence of the @xmath31-factor as determined by electron - spin resonance .
to first approximation , the cu(ii ) ions in the planar material have a local octahedral symmetry with a tetragonal distortion along the @xmath7-axis .
in such situations , the unpaired spin in a @xmath52 configuration is expected to occupy the @xmath53 orbital @xcite , and the @xmath31-factor parallel to the distorted @xmath54-axis takes a larger value than those in the @xmath55-plane ( see discussion in ref . ) .
accordingly , the data in fig . [ pulsed](d ) show that the @xmath31-factor is lowest when the excitation field is applied in the @xmath56-plane , implying that the direction of highest electronic orbital overlap is the cupyz directions , with the strong q2d exchange interactions being mediated via the molecular orbitals of the pyrazine . for the chain - like material ,
the distortion is more complicated due to the lower crystal symmetry . nevertheless , using two rotations , the smallest values of the @xmath31-factor
are found to be along the cupyz and cupyo bonds , as shown in fig .
[ pulsed](e ) , implying that again the magnetic orbital is @xmath53 .
this allows for the possibility of good exchange coupling along those bonds , but as the pyo molecules are non - bridging ligands , the q1d magnetism must be mediated along the cupyz chains .
taken together , these experimental observations paint a complete picture of the two closely - related magnetic systems . from the point of view of magnetic superexchange
, the chain direction in the q1d material looks very similar to the two cupyz directions in the q2d material and so , despite the compositional differences , the primary nearest - neighbour exchange energies remain largely unaltered . at the same time the critical field in the q2d material
is approximately double that for the q1d compound because it has twice the number of nearest neighbours .
the q2d compound is strongly anisotropic , comparable to the most 2d materials yet identified @xcite .
as suggested above , this is likely due to the disconnect and staggering that occurs between successive planes .
the extreme anisotropy explains why the zero - field heat capacity was not sensitive to the antiferromagnetic transition observed using @xmath16sr . in the q1d material
the anisotropy is less pronounced ( even though the @xmath57 ratio is smaller ) because the chains are not staggered , there are twice as many next - nearest neighbours , and the shortest distance between chains ( along the @xmath9-axis ) is approximately half the interlayer separation in the q2d material .
these materials showcase the ability to take deliberate control over the magnetic properties of polymeric systems .
the self - organisation of the coordination polymers enables them to spontaneously form crystalline lattices whose structure can be anticipated with a high level of predictability .
it is this predictive power , together with the ability to choose the starting ingredients and the knowledge of the exchange efficiency of various ligands accrued over the past few decades , that permits the preselection of exchange anisotropy . in the cupyz systems we have demonstrated the ability to make this preselection without significantly perturbing the magnitude of the primary interaction strengths , while previous studies of similar materials have highlighted the capacity to tune the exchange couplings without changing the overall dimensionality .
thus these compounds represent a promising approach to magnetic crystal engineering , and in particular raise the possibility of generating systems that exhibit higher ordering temperatures and other cooperative phenomena .
this project was supported by the epsrc ( uk ) , the nsf and doe ( us ) , and the european commission .
we thank peter baker for technical assistance .
pag would like to thank keola wierschem for useful discussions . 00
current address : department of physics , durham university , south road , durham , dh1 3le , uk .
s. r. batten , s. m. neville and d. r. turner , _ coordination polymers : design , analysis and application _ ( rsc , cambridge , 2009 ) .
o. k. farha _
et al . _ , nature chemistry * 2 * , 944 , ( 2010 ) .
s. bureekaew _
et al .
_ , nature materials * 8 * , 831 , ( 2009 ) .
n. l. rosi _
et al . _ ,
science * 300 * , 1127 ( 2003 ) .
a. m. kuchison , m. o. wolf and b. o. patrick , inorg . chem . * 49 * , 8802 ( 2010 )
. m. j. katz and d. b. leznoff , j. am .
chem .
soc . * 131 * , 18435 ( 2009 ) .
o. sengupta and p. s. mukherjee , inorg . chem . *
49 * , 8583 ( 2010 ) .
t. okubo , r. kawajiri , t. mitani and t. shimoda , j. am .
chem .
soc . , * 127 * , 1759 ( 2005 ) .
s. ohkoshi , k. imoto , y. tsunobuchi , s. takano and h. tokoro , nature chemistry * 3 * , 564 ( 2011 ) .
g. halder , k. w. chapman , j. a. schlueter and j. l. manson , angew .
chemie int
. ed . * 50 * , 419 ( 2011 ) .
j. h. her _
et al .
_ , inorg
. chem . * 49 * , 1524 ( 2010 ) . r. s. fishman , w. w. shum and j. s. miller , phys .
rev .
b * 81 * , 172407 ( 2010 ) . j. l. musfeldt _
et al . _ , phys .
rev .
lett . * 103 * , 157401 ( 2009 ) . j. l. manson _
et al . _ , j. am . chem . soc . * 131 * , 6733 ( 2009 ) . j. s. miller , chem .
soc . rev . *
40 * , 3266 , ( 2011 ) .
s. j. blundell , f. l. pratt , j. phys .
cond . mat .
* 16 * , r771 ( 2004 ) .
r. t. butcher , c. p. landee , m. m. turnbull and f. xiao , inorg .
chim .
acta * 361 * , 3654 ( 2008 ) .
p. a. goddard _
et al . _ , new j. phys . * 10 * , 083025 ( 2008 ) .
f. m. woodward _
et al . _ ,
inorg . chem . *
46 * , 4256 ( 2007 ) .
p. a. goddard _
et al . _ , phys .
rev .
b * 78 * , 052408 ( 2008 ) . see the supplemental material section at the end of the paper for detailed information on methods and crystal structures .
j. l. manson _
et al . _ ,
chem .
commun .
* 2006 * , 4894 ( 2006 ) .
j. darriet , m. s. haddad , e. n. duesler and d. n. hendrickson , inorg . chem . * 18 * , 2679 ( 1979 ) .
y. kohama _
et al . _ ,
phys .
rev .
b * 84 * , 184402 ( 2011 ) .
p. sengupta , a. w. sandvik and r. r. p. singh , phys .
rev .
b * 68 * , 094423 ( 2003 )
. a. j. steele _
et al . _ , phys .
rev .
b * 84 * , 064412 ( 2011 ) .
j. c. bonner , m. e. fisher , phys . rev . * 135 * a640 ( 1964 ) . c. yasuda
_ et al . _ ,
phys .
rev .
lett . * 94 * , 217201 ( 2005 ) .
j. l. manson _
et al . _ ,
chem .
mater . * 20 * , 7408 ( 2008 ) .
a. abragam and b. bleaney , _ electron paramagnetic resonance of transition ions . _ ( oup oxford , 1970 ) . |
it is widely believed that radio - loud quasars and radio galaxies differ from each other only in terms of the angle between their radio axes and the line of sight ( scheuer 1987 ; barthel 1989 ) .
quasars are observed fairly close to the line of sight ( @xmath1 ) and therefore frequently exhibit the effects of beaming , such as superluminal motion and luminous flat - spectrum radio cores , whereas radio galaxies are observed with their axes close to the plane of the sky , and do not show these effects .
however , quasars also possess a luminous non - stellar optical continuum and broad emission lines , which are absent in radio galaxies .
it has therefore been proposed that there is material around the nucleus which lies preferentially in the plane perpendicular to the radio axis and obscures the central regions from view in radio galaxies .
because of the geometry of this material , it is often referred to as the `` torus '' , but other geometries ( e.g. a warped disk ) are possible . although the broad lines are hidden from direct view by this material , the narrow line region ( nlr ) is much larger in size and should be less strongly affected .
it is therefore possible that narrow line luminosity could be an isotropic measure of the strength of the non - stellar continuum that is otherwise invisible in radio galaxies .
jackson & browne ( 1990 ) tested this idea by comparing the @xmath05007 luminosities of quasars and radio galaxies with similar redshifts and extended radio luminosities .
they reported that quasars are 510 times more luminous in this line than radio galaxies , and attributed the difference to higher extinction in radio galaxies .
he s , barthel & fosbury ( 1993 ) performed a similar test using the @xmath03727 doublet and found that there was no measurable difference between the line luminosities of the two classes .
since is at a shorter wavelength than and would be more greatly affected by a foreground screen of reddening , they suggested that most of the emission is produced close to the nucleus and is still obscured by the torus in radio galaxies , whereas is produced farther out and is unaffected .
unfortunately , jackson & browne s analysis was biased because , although the radio galaxies were selected from the 3cr - based complete sample of laing , riley & longair ( 1983 ; also called 3crr ) , the quasars were drawn from a number of incomplete surveys and were therefore subject to uncertain selection effects , most notably the tendency to preferentially include optically bright objects .
since line and continuum luminosities are very well correlated in quasars , their result would be biased towards finding systematically higher line luminosities in the quasars .
in addition , jackson & browne s radio galaxy sample included some objects with very weak emission lines ( class b optical spectra ; hine & longair 1979 ) that it is now believed may not belong to the unified scheme .
an unbiased analysis significantly reduces the magnitude of jackson & browne s result , but the quasars are still about twice as luminous in than the radio galaxies .
this result is also seen in the higher / ratios in broad - line objects than in narrow - line objects ( e.g. saunders et al .
1989 ; tadhunter et al .
1993 ) . at higher redshift , however , the luminosities of the two classes of object are comparable ( jackson & rawlings 1997 ) . rawlings & saunders ( 1991 ) note that the tendency for quasars to have a higher emission line luminosity than radio galaxies could be the result of a classification bias .
this could explain the discrepancy in luminosities between the two classes , but appears to run counter to the similarity in the line emission . in this _
, we apply the simple receding torus model ( lawrence 1991 ; hill , goodrich & depoy 1996 ) to provide a quantitative explanation of the difference in luminosities between quasars and radio galaxies ( and its redshift dependence ) , and explain why a similar effect is not seen in .
the free parameters involved in our explanation are constrained by observed quantities independent of the emission line luminosities .
we conclude that the luminosity of the @xmath05007 line is an excellent indicator of the strength of the underlying non - stellar continuum .
the receding torus model , as proposed by lawrence ( 1991 ) and illustrated in figure [ fig : rtm ] , is fairly simple .
dust evaporates at a temperature of about 1500k , and can not survive closer to the nucleus than the radius where the temperature of the nuclear radiation field is hotter than this .
all quasars display a near - infrared bump longward of 1@xmath2 m that is believed to be caused by thermal emission from dust , and the constancy of wavelength at which this bump appears indicates that dust is always present at the hottest possible temperatures . the inner radius of the torus is therefore determined by the radius at which dust evaporates , which scales as @xmath3 .
if the half - height , @xmath4 , of the torus is independent of the source luminosity , then the half opening angle @xmath5 will be larger in the more luminous objects . hill et al .
( 1996 ) find that this model fits the observed nuclear extinctions in radio galaxies very well .
we are more likely to be viewing the luminous objects within their cones , and are consequently more likely to classify them as quasars .
qualitatively then , this simple model could explain the tendency for quasars to have a higher luminosity than radio galaxies .
we therefore undertake a quantitative investigation using monte carlo simulations .
we consider an ensemble of objects with the same redshift and radio power , which could reasonably be assumed to have similar torus properties ( e.g.torus height ) .
their intrinsic optical ultraviolet luminosities will have a well - defined mean , @xmath6 , determined by the strong correlation between optical and radio luminosity , but will show scatter due to the intrinsic dispersion in this relationship .
we assume this dispersion to be gaussian in @xmath7 , with standard deviation @xmath8 .
we define @xmath9 to be the half opening angle for an object with the mean luminosity , @xmath6 .
an object with luminosity @xmath10 with therefore have an half opening angle , @xmath11 , \ ] ] and the probability that such an object will be observed as a radio galaxy ( i.e. our line of sight is outside the opening angle of the cone ) is @xmath12 obviously , @xmath13 . for our ensemble of objects
, we can therefore determine the ratio of the mean optical uv luminosities of quasars and radio galaxies ( we evaluate the mean in log space ) , for a given @xmath8 and @xmath9 .
however , there exist observational data which allow us to fix these free parameters .
first , serjeant et al . ( 1998 ) measure the scatter about the mean optical radio luminosity correlation for steep - spectrum quasars to be 1.5magnitudes ( 0.6dex ) .
this will be slightly smaller than @xmath8 , since there should exist radio galaxies whose intrinsic optical luminosities are on average fainter than those of the quasars
. however , serjeant et al.s sample was dominated by distant ( @xmath14 ) quasars whose high luminosities would imply a large opening angle for the torus . at large opening angles ,
most objects are observed as quasars , and therefore the dispersion in optical magnitude in a quasar sample is close to the intrinsic dispersion of the parent agn population ( figure [ fig : disp ] ) .
we thus adopt @xmath15 .
the second constraint is the observed quasar ( or radio galaxy ) fraction ; we must constrain @xmath9 such that the fraction in our simulation matches that in a radio - selected sample . following the reasoning of laing et al . ( 1994 ) and others , when comparing our results to observed quantities , we consider only those radio galaxies with prominent emission lines . at @xmath16 and @xmath17 , where there exist high - quality optical spectra allowing classification on the basis of broad h@xmath18 ( laing et al .
1994 ) , there are 15 such radio galaxies and 9 quasars .
the observed radio galaxy fraction is therefore 0.625 , corresponding to @xmath19 for @xmath15 ( figure [ fig : frg ] ; the value is extremely insensitive to the choice of @xmath8 ) .
figure [ fig : rellum ] illustrates the amount by which quasars are expected to be overluminous compared to radio galaxies as a function of @xmath9 and @xmath8 , and for the values determined previously , we find @xmath20 .
this is effectively the same as the difference in luminosities for the two groups of objects , which could therefore be explained if the emission line luminosity is directly proportional to the ionizing luminosity .
its constancy of equivalent width over approximately three orders of magnitude in luminosity ( miller et al .
1992 ) indicates that it is . why then is a similar effect not seen in the luminosity , since both emission lines correlate well with optical luminosity over several orders of magnitude in quasars ?
we believe that the answer is once again due to a bias introduced by scatter .
although the effective ionization parameter of the nlr does not correlate with luminosity ( the range of quasar luminosities is far larger than the observed range of @xmath21 , @xmath22 ; e.g. saunders et al .
1989 ) , the most luminous objects at a given redshift will also have the highest ionization parameters . figure [ fig : linelum ] shows the dependence of and luminosity as a function of ionization parameter for an optically - thick plane - parallel slab .
this was calculated with the photoionization code cloudy version 84.12a ( ferland 1993 ) using the agn continuum described in mathews & ferland ( 1987 ) , although effectively identical results are obtained for power laws with a range of spectral index . over the range of effective ionization parameter typical of the nlr ,
the luminosity is only very weakly dependent on the strength of the ionizing continuum .
whereas a hundredfold increase in @xmath21 increases the luminosity of by a factor of @xmath23 , the luminosity becomes only @xmath24 times larger .
although a single plane - parallel slab is obviously an overly simplistic model for the nlr , there is very little density dependence in the way the line luminosities vary with ionization parameter .
the lessening of the effect at higher redshift is easy to understand , since in the flux - limited 3cr sample this corresponds to higher luminosity and therefore a larger mean opening angle . at such large angles , variations in luminosity of an order of magnitude or less
have little effect on the angle , and hence the bias towards more luminous objects being seen as quasars is smaller .
this is apparent in figure [ fig : rellum ] .
although radio galaxies outnumber quasars 2:1 at low @xmath25 , at @xmath26 the ratio is nearly reversed .
in addition , there may be many objects like 3c 22 ( economou et al .
1995 ; rawlings et al .
1995 ) or 3c 41 ( simpson , rawlings & lacy 1998 ; economou et al .
1998 ) that possess broad h@xmath18 which is difficult to observe from the ground at @xmath27 due to it being redshifted into regions of poor atmospheric transmission or the @xmath28 window , whose strong oh airglow lines hamper the detection of broad features .
the observed radio galaxy fraction of @xmath29% at @xmath26 should therefore be considered an upper limit , and it provides a lower limit to @xmath30 .
this corresponds to @xmath31 , and the ratio would drop below 1.5 if only one third of those objects currently classified as radio galaxies are really quasars .
it should also be noted that measurement errors tend to be rather larger in near - infrared than in optical spectroscopy , and so the study of jackson & rawlings ( 1997 ) is less sensitive to differences in the two classes than are low redshift studies .
jackson & rawlings ( 1997 ) do not offer any explanation as to why they fail to observe a similar effect to that of jackson & browne ( 1990 ) , yet it is naturally explained in our model as a luminosity , rather than a redshift , dependence . although our model might appear to predict that high redshift samples whose luminosities are similar to those of low redshift 3crr objects ( e.g. a sample from the 7c catalogue ) should also display a separation in luminosity with broad / narrow line class , such a prediction also depends on the assumption that the average torus height does not vary with redshift since this also controls the mean opening angle .
we can predict that the difference between the luminosities of radio galaxies and quasars should be a function of quasar fraction ( figures [ fig : frg ] and [ fig : rellum ] ) , and should effectively vanish for samples with a large ( @xmath32% ) quasar fraction .
since our model requires that effectively all the emission is seen directly , it appears to be in contradiction with the recent detection of polarized in radio galaxies by di serego alighieri et al .
these authors calculate that 8090% of the total emission is obscured in radio galaxies , but their calculations assume that the observed featureless continuum is entirely scattered .
it is known that seyfert galaxies possess an extended featureless continuum ( fc2 ; tran 1995 ) and it is reasonable to imagine that radio galaxies do also , thus lowering the estimate of the obscured fraction .
in addition , the emission may be polarized , even if none is obscured , if the scattering region is distributed anisotropically . although di serego alighieri et al .
disfavour this idea , it is not in conflict with observations , and would be expected if the scattering material is associated with the outflow .
the fact that polarized emission is detected in 3c 227 despite our relatively clear view of the nucleus seems to indicate that significant obscuration is not required for the line emission to be polarized .
we also note that the size of the -emitting region that must be obscured is much larger than estimates for the height of the torus derived from the dust evaporation radius and cone opening angle .
while the polarization data are intriguing , we therefore feel that our model is not in conflict with them .
we have used a simple receding torus model , whose free parameters are constrained by observation , to show that low redshift 3crr quasars should be , on average , about twice as luminous in their ionizing continua as radio galaxies of the same radio luminosity .
this difference should also be seen in their , but not their , emission line luminosities , in agreement with observation . for samples with a higher quasar fraction , such as the high redshift 3crr objects ,
the difference in ionizing luminosities between quasars and radio galaxies should be smaller , and there should therefore be less of a difference in their luminosities , again in line with observation .
this model leads to the conclusion that the @xmath05007 emission line , and not the @xmath03727 doublet , is an unbiased indicator of the intrinsic optical ultraviolet luminosity of both quasars and radio galaxies .
the author would like to thank peter eisenhardt for a useful discussion and steve rawlings for a critical reading of the manuscript .
this work was performed by the jet propulsion laboratory , california institute of technology , under a contract with nasa .
laing r. a. , jenkins c. r. , wall j. v. , unger s. w. , 1994 , in bicknell g. v. , dopita m. a. , quinn p. j. , eds , asp 54 : the first stromlo symposium : the physics of active galaxies .
asp , san francisco , p. 201 | we undertake a quantitative investigation , using monte carlo simulations , of the amount by which quasars are expected to exceed radio galaxies in optical luminosity in the context of the ` receding torus ' model .
we compare these simulations with the known behaviour of the @xmath05007 and @xmath03727 emission lines and conclude that is the better indicator of the strength of the underlying non - stellar continuum .
# 1to 0pt#1 galaxies : active galaxies : nuclei quasars : emission lines radio continuum : galaxies |
it has been recently shown @xcite that a constant magnetic field in @xmath1 and @xmath0 dimensions is a strong catalyst of dynamical chiral symmetry breaking , leading to the generation of a fermion dynamical mass even at the weakest attractive interaction between fermions .
the essence of the effect is the dimensional reduction @xmath2 ( i.e. @xmath3 and @xmath4 ) in the infrared dynamics of the fermion pairing in a magnetic field .
the physical reason of this reduction is the fact that the motion of charged particles is restricted in those directions that are perpendicular to the magnetic field .
this is in turn connected with the point that , at weak coupling between fermions , the fermion pairing , leading to the chiral condensate , is mostly provided by fermions from the lowest landau level ( lll ) whose dynamics are ( @xmath5)dimensional . in this paper , we shall further clarify the effect of the dynamical reduction , studying in detail the infrared dynamics in the ( @xmath0)dimensional nambu
jona lasinio ( njl ) model in a magnetic field .
we shall consider both the ordinary ( non supersymmetric ) and supersymmetric versions of the njl model .
in particular , we shall show that in the continuum " limit , when both the strength of the magnetic field and the ultraviolet cutoff go to infinity , both the non supersymmetric and supersymmetric ( weakly coupling ) njl models with @xmath6 colors are reduced to a continuum set of independent ( 1 + 1)dimensional gross - neveu ( gn ) models @xcite , labeled by coordinates @xmath7 in the plane perpendicular to the magnetic field @xmath8 .
the number of colors in the gn models is @xmath9 , where @xmath10 is @xmath11 in the continuum " limit ( here @xmath12 is the ultraviolet cutoff ) .
as will be shown in sec.3 , the factor @xmath13 is proportional to a ( local ) magnetic flux attached to each point in the @xmath14-plane . on the other hand , at strong coupling , the dynamics in the supersymmetric and non supersymmtric njl models are very different .
recall that there is no spontaneous chiral symmetry breaking in the supersymmetric njl model @xcite .
an external magnetic field changes the situation dramatically : chiral symmetry breaking occurs for any value of the coupling constant in this model .
this agrees with the general conclusion of refs .
@xcite that the effect of the catalysis of chiral symmetry breaking by a magnetic field in @xmath0 dimensions is a universal , model independent effect . as was already shown in ref .
@xcite , the dimensional reduction @xmath4 in the dynamics of the fermion pairing in a ( finite ) magnetic field is consistent with spontaneous symmetry breaking .
recall that , due to the mermin wagner
coleman ( mwc ) theorem @xcite , there can not be spontaneous breakdown of continuous symmetries at @xmath15 .
the mwc theorem is based on the fact that gapless nambu goldstone ( ng ) bosons can not exist in @xmath16 dimensions .
however , in a magnetic field , the reduction @xmath4 takes place ( in the infrared region ) only for propagators of charged particles : it reflects the fact that the motion of charged particles is restricted in the directions perpendicular to the magnetic field . on the other hand ,
ng bosons connected with spontaneous chiral symmetry breaking are neutral and therefore their propagators have ( @xmath0)dimensional form @xcite)dimensional form , we understand that the denominator of the propagators depends on energy and all the components of the center of mass momentum , i.e. @xmath17 with @xmath18 . ] .
this in turn implies that the effect of spontaneous chiral symmetry breaking does not contradict the mwc theorem .
the interplay between the gn model and the njl model in a magnetic field established in this paper further clarifies this issue .
as was mentioned above , the njl model is reduced to a set of the gn models only as the strength of the magnetic field goes to infinity .
as we shall show in sec.4 , at finite @xmath19 , the dynamics of the njl model in a magnetic field are in a sense similar to the dynamics in the ( @xmath20)dimensional gn model : the magnetic length @xmath21 plays here the role of the ( physical ) @xmath22parameter which is an infrared regulator .
as was already pointed out in refs.@xcite , there may be interesting applications of this effect in cosmology as well as in particle and condensed matter physics .
the results of the present paper may be particularly relevant for cosmological scenarios based on supersymmetric dynamics @xcite .
the paper is organized as follows . in section 2
we , for completeness , derive the effective action in the gn model . in sections 3 and 4
we establish the connection between the gn model and the njl model in a magnetic field as @xmath23 . in section 5
we consider the dynamics of the supersymmetric njl model in a magnetic field . in section 6
we summarize the main results of the paper and discuss possible applications of these results and as well as the possibility of their extension to inhomogeneous magnetic field configurations . in the appendix some useful formulas and relations are derived .
in this section , for completeness , we shall derive the effective action for the gn model .
the lagrangian density of the gn model is : @xmath24 + \frac{\tilde{g}}{2 } \left [ ( \bar{\psi}\psi)^2+(\bar{\psi}i\gamma^5\psi)^2 \right ] \label{eq1}\ ] ] where @xmath25 and the fermion field carries an additional color " index @xmath26 ( for simplicity , we consider the case of the chiral @xmath27 symmetry ) .
the theory is equivalent to the theory with the lagrangian density @xmath28 - \bar{\psi}(\sigma+i\gamma^5\pi)\psi - \frac{1}{2\tilde{g } } \left(\sigma^2+\pi^2\right ) .
\label{eq2}\ ] ] the euler
lagrange equations for the auxiliary fields @xmath29 and
@xmath30 take the form of constraints : @xmath31 and the lagrangian density ( [ eq2 ] ) reproduces eq.([eq1 ] ) upon application of the constraints ( [ eq3 ] ) .
the effective action for the composite fields @xmath29 and @xmath30 can be obtained by integrating over fermions in the path integral .
it is given by the standard relation : @xmath32 .
\label{eq5}\end{aligned}\ ] ] the low energy quantum dynamics are described by the path integral ( with the integrand @xmath33 ) over the composite fields @xmath29 and @xmath30 . as @xmath34 , the path integral
is dominated by the stationary points of the action : @xmath35 .
we will analyze the dynamics by using the expansion of the action @xmath36 in powers of derivatives of the composite fields .
we begin the calculation of @xmath36 by calculating the effective potential @xmath37 . since @xmath37 depends only on the @xmath38invariant @xmath39 , it is sufficient to consider a configuration with @xmath40 and @xmath29 independent of @xmath41
. then we find from eqs .
( [ eq4 ] ) and ( [ eq5 ] ) : @xmath42 , \label{eq6}\end{aligned}\ ] ] where the integration is done in euclidean region ( @xmath12 is an ultraviolet cutoff ) .
as is known , in the gn model , the equation of motion @xmath43 has a nontrivial solution @xmath44 for any value of the coupling constant @xmath45 .
then the potential @xmath37 can be rewritten as @xmath46 , \label{eq7}\ ] ] where @xmath47 due to the mwc theorem @xcite , there can not be spontaneous breakdown of continuous symmetries at @xmath15 .
the parameter @xmath48 is an order parameter of chiral symmetry breaking only in leading order in @xmath49 ( this reflects the point that the mwc theorem is not applicable to systems with @xmath50 @xcite ) . in the exact gn solution ,
spontaneous chiral symmetry breaking is washed out by interactions ( strong fluctuations ) of would
be ng bosons @xmath30 ( i.e. after integration over @xmath30 and @xmath29 in the path integral ) .
the exact solution in this model presumably corresponds to the realization of the berezinsky
thouless ( bkt ) phase : though chiral symmetry is not broken in this phase , the parameter @xmath48 still defines the fermion mass , and the would
be ng boson @xmath30 transforms into a bkt gapless excitation @xcite .
let us now turn to calculating the kinetic term in @xmath36 .
the chiral @xmath38 symmetry implies that the general form of the kinetic term is @xmath51 where @xmath52 and @xmath53 , @xmath54 are functions of @xmath55 . to find the functions @xmath53 and @xmath54
, one can use different methods .
we utilize the same method as in ref .
@xcite ( see appendix a in that paper ) .
the result is : @xmath56 , \label{eq10 } \\
f^{\mu\nu}_2(\rho^2)&=&-\frac{i}{4 } \int\frac{d^2k}{(2\pi)^2 } tr\left [ s(k)\frac{\partial^2s(k)}{\partial k_{\mu}\partial k_{\nu}}- s(k)i\gamma_5\frac{\partial^2s(k)}{\partial k_{\mu}\partial k_{\nu } } i\gamma_5\right ] , \label{eq11}\end{aligned}\ ] ] with @xmath57 .
the explicit form of these functions is : @xmath58
in this section , we compare the effective actions in the gn model and in the njl model in a magnetic field , and we establish a rather interesting connection between these two models . the analog of the lagrangian density ( [ eq2 ] ) in the njl model in a magnetic field is @xmath59 - \bar{\psi}(\sigma+i\gamma^5\pi)\psi - \frac{1}{2 g } \left(\sigma^2+\pi^2\right ) \label{eq2njl}\ ] ] where @xmath60 , @xmath61 ( the magnetic field is in @xmath62 direction ) . in leading order in @xmath63 , the effective action in the njl model in a magnetic field
is derived in refs .
the effective potential and the kinetic term are ( @xmath52 ) : @xmath64 + o\left(\frac{1}{\lambda}\right ) , \label{eq13}\end{aligned}\ ] ] @xmath65 with @xmath53 and @xmath54 being diagonal tensors : @xmath66 , \nonumber\\ f^{22}_1 = f^{33}_1&= & - \frac{n_c}{8\pi^2 } \left[\ln\frac{\lambda^2}{\rho^2 } - \gamma+\frac{1}{3}\right ] , \nonumber\\ f^{00}_2 = - f^{11}_2&=&-\frac{n_c}{24\pi^2 } \left[\frac{(\rho l)^2}{2 } \zeta\left(2,\frac{(\rho l)^2}{2}+1\right ) + \frac{1}{(\rho l)^2}\right ] , \label{eq15 } \\
f_2^{22 } = f^{33}_2&= & - \frac{n_c}{8\pi^2 } \bigg[(\rho l)^4 \psi \left(\frac{(\rho l)^2}{2}+1\right ) - 2(\rho l)^2\ln\gamma\left(\frac{(\rho l)^2}{2}\right ) \nonumber\\ & & \qquad - ( \rho l)^2 \ln\frac{(\rho l)^2}{4\pi } - ( \rho l)^4 - ( \rho l)^2 + 1\bigg ] .
\nonumber\end{aligned}\ ] ] here @xmath67 is the njl coupling constant , @xmath6 is the number of colors , @xmath68 is the generalized riemann zeta function , @xmath69 , @xmath70 is the euler constant , @xmath71 , and @xmath72 is the magnetic length .
the gap equation @xmath73 is , @xmath19 , where @xmath74 is a minimum of the potential @xmath75 . ]
@xmath76 where the dimensionless coupling constant @xmath77 . in the derivation of this equation
, we used the relations @xcite : @xmath78 as @xmath79 ( @xmath80 ) , we recover the known gap equation in the njl model ( for a review see ref.@xcite ) : @xmath81 this equation admits a nontrivial solution only if @xmath82 is supercritical , @xmath83 ( as eq.([eq2njl ] ) implies , a solution to the gap equation , @xmath84 , coincides with the fermion dynamical mass , @xmath85 ) . as was shown in refs .
@xcite , at @xmath86 , a non trivial solution exists for all @xmath87 .
let us consider the case of small subcritical @xmath82 , @xmath88 , in detail .
a solution is seen to exist for this case if @xmath89 is small .
specifically , for @xmath90 , the left hand side of eq.([eq16 ] ) is positive .
since the first term of the right hand side in this equation is negative , we conclude that a non trivial solution to this equation may exist only for @xmath91 ( @xmath92 ) .
we then find the solution : @xmath93 actually , since eq.([eq21 ] ) implies that condition ( [ eq20 ] ) is violated only if @xmath94 , the expression ( [ eq21 ] ) is valid for all @xmath82 outside that ( scaling ) region near the critical value @xmath95 .
note that in the scaling region ( @xmath96 ) the expression for @xmath97 is different @xcite : @xmath98 } { \ln\lambda^2l^2}. \label{eq22}\ ] ] let us compare relation ( [ eq21 ] ) with relation ( [ eq8 ] ) for the dynamical mass in the gn model .
the similarity between them is evident : @xmath19 and @xmath99 in eq.([eq21 ] ) play the role of an ultraviolet cutoff and the dimensionless coupling constant @xmath45 in eq.([eq8 ] ) .
let us discuss this connection and show that it is intimately connected with the dimensional reduction @xmath100 in the dynamics of the fermion pairing in a magnetic field .
eq.([eq8 ] ) implies that the gn model is asymptotically free , with the bare coupling constant @xmath101 as @xmath102 .
let us now consider the following limit in the njl model in a magnetic field : @xmath23 , @xmath103 . then relation ( [ eq21 ] ) , which can be rewritten as @xmath104 implies that the behavior of the bare coupling constant @xmath82 must be @xmath105 in order to get a finite value for @xmath97 in this limit .
thus in this
continuum " limit , we recover the same behavior for the coupling @xmath82 in the njl model as for the coupling constant @xmath45 in the gn model .
let us now compare the effective potentials in these two models . at first glance
, the expressions ( [ eq6 ] ) and ( [ eq13 ] ) for the effective potentials in these models look very different : the character of ultraviolet divergences in 1 + 1 and 3 + 1 dimensional theories is essentially different .
however , using eqs.([eq17 ] ) and ( [ eq18 ] ) , the expression ( [ eq13 ] ) can be rewritten , for small @xmath89 , as @xmath106 .
\label{eq25}\end{aligned}\ ] ] then , expressing the coupling constant @xmath82 through @xmath48 from eq.([eq21 ] ) , we find that @xmath107
. \label{eq26}\end{aligned}\ ] ] here we used the fact that , because of eq.([eq20 ] ) , the ratio @xmath108 is small near the minimum @xmath109 .
the expressions ( [ eq7 ] ) and ( [ eq26 ] ) for the potentials in these two models look now quite similar .
there is however an additional factor @xmath110 in the expression ( [ eq26 ] ) .
moreover , the field @xmath111 , which depends on the two coordinates @xmath112 and @xmath113 in the gn model , depends on the four coordinates @xmath112 , @xmath113 , @xmath114 and @xmath115 in the njl model . in order to clarify this point
, let us turn to the analysis of the kinetic term ( [ eq14 ] ) in the effective action of the njl model in a magnetic field .
because of the expression ( [ eq21 ] ) for @xmath48 at small @xmath82 , the term @xmath116 dominates in the functions @xmath117 and @xmath118 , around the minimum @xmath109 : @xmath119 up to the additional factor @xmath110 , these functions coincide with those in ( [ eq12 ] ) in the gn model .
on the other hand , the functions @xmath120 and @xmath121 , connected with derivatives with respect to the transverse coordinates , are strongly supressed , as compared to the functions ( [ eq27 ] ) , and the ratios of the functions @xmath120 and @xmath121 to those in ( [ eq27 ] ) go rapidly ( as @xmath122 ) to zero as @xmath123 , @xmath124 . as a result ,
the coordinates @xmath114 and @xmath115 become redundant variables in this limit : there are no transitions of field quanta between different points in the @xmath125plane
. therefore the model degenerates into a set of independent @xmath126dimensional models , labeled by @xmath127 coordinates .
let us show that all these models coincide with a @xmath126dimensional gn model with the number of colors @xmath9 , where the factor @xmath11 was introduced above , in the definition of the continuum " limit ( @xmath23 , @xmath124 ) .
let us put the njl model on a lattice with @xmath128 the lattice spacing of the discretized space - time ( in euclidean region ) .
then its effective action can be written as @xmath129 where @xmath130 , @xmath131 , with @xmath132 , @xmath133 , @xmath134 , @xmath135 , and here the factor @xmath110 was explicitly factorized from @xmath136 .
now , taking into account eqs.([eq7]),([eq12 ] ) and eqs.([eq27 ] ) , ( [ eq28 ] ) , we find that @xmath137 in the continuum " limit with @xmath138 ( here @xmath139 is the ultraviolet cutoff on the lattice ) is anyway arbitrary here , we use the same notation for the cutoff on the lattice as in the proper - time regularization . ] .
the lagrangian density @xmath140 corresponds to the gn model with the number of colors @xmath9 .
note that the symbol @xmath141 here is somewhat formal and it just implies that the gn model occurs at each point in the @xmath125plane .
the physical meaning of this reduction of the njl model in a magnetic field is rather clear . at weak coupling ,
the fermion pairing in a magnetic field takes place essentially for fermions in the lll with the momentum @xmath142=0 .
the size of the radius of the lll orbit is @xmath143=@xmath144 @xcite .
as the magnetic field goes to infinity , this radius shrinks to zero .
then , because of the degeneracy in the lll @xcite , there are @xmath145 states with @xmath142=0 at each point in the @xmath125plane .
this degeneracy factor ( proportional to the ( local ) magnetic flux across a plaquette ) leads to changing the number of colors , @xmath146 , in the gn model.note that since @xmath147 appears analytically in the path integral of the theory , one can give a non - perturbative meaning to the theory with non - integral @xmath147 .
a few comments are in order .
since these gn models are independent , the parameters of the chiral @xmath148 transformations can depend on @xmath149 .
in other words , here the chiral group is @xmath150 . as a result , there are an infinite number of gapless modes @xmath151 in the continuum " limit . since there is no spontaneous breakdown of continuous symmetries at @xmath15 , the fields @xmath151 do not describe ng bosons ( though they do describe gapless bkt excitations ) @xcite .
since the magnetic field depends on @xmath12 in the continuum " limit , it can be considered as an additional parameter ( coupling constant " ) in the renormalization group .
the ratio @xmath152 is arbitrary here .
> from the point of view of the renormalization group , this can be interpreted as the presence of a line of ultraviolet fixed points for the dimensionless coupling @xmath153 .
the values of @xmath153 on the line define the local magnetic flux and , therefore , the number of colors @xmath147 in the corresponding gn models . our consideration of reducing the njl model in a magnetic field to a continuum set of the gn models was somewhat heuristic .
it would be worth deriving this reduction in a more rigorous way , putting the njl model on a lattice in euclidean space and then realizing renormalizations in the continuum " limit . in the next section
, we shall discuss the connection between the gn model and the njl model in a magnetic field in more detail .
in the previous section , the reduction of the njl model in a magnetic field of the infinite strength to a continuum set of the gn models was established .
but what is the connection between the njl and gn models at finite , though large , values of the magnetic field ? in order to answer this question , let us turn to a more detailed discussion of the infrared dynamics within these two models .
the gn model is asymptotically free , with the bare coupling constant @xmath101 as @xmath102 .
therefore there is dimensional transmutation in the model : in the scaling region ( @xmath154 ) the infrared dynamics , with momenta @xmath155 satisfying @xmath156^{-1}\ll 1 $ ] , are essentially independent of either the coupling constant @xmath45 or the cutoff @xmath12 ; the only relevant parameter is the dynamical mass @xmath48 ( which is an analogue of the parameter @xmath157 in qcd ) . because of eq.([eq21 ] ) for @xmath48 , one might expect that a similar dimensional transmutation should take place in the njl model in a magnetic field : for @xmath158 , the infrared dynamics , with @xmath159 , should be essentially independent of the magnetic field and the coupling @xmath67 . the real situation , however , is more subtle .
let us look at the propagators for fermions and @xmath29 and @xmath30 particles at low momenta in the njl model with a magnetic field , in leading order in @xmath63 @xcite : @xmath160^{-1},\label{eq31}\\ { \cal d}_{\pi}(k)&=&\frac{c_{\pi}}{n_c}\bigg[k_0 ^ 2-k_1 ^ 2- \frac{m_{dyn}^2}{|eb| } \ln\left(\frac{|eb|}{\pi m_{dyn}^2}\right){\bf k}^2_{\perp } \bigg]^{-1 } , \label{eq32}\end{aligned}\ ] ] where @xmath161 and where @xmath162 and @xmath163 are some inessential constants ( we consider the case of the weak coupling , when relation ( [ eq21 ] ) is valid ) .
because of relations ( [ eq21 ] ) and ( [ eq23 ] ) , the coefficients of the @xmath164terms in the propagators @xmath165 and @xmath166 are exponentially small . also , in the infrared region , the fermion propagator is independent ( up to power corrections @xmath167 , @xmath168 ) of the magnetic field .
therefore one might think that in this case , like in the gn model , the only relevant parameter for the infrared dynamics is @xmath48 .
however , while the dependence of the propagators of fermions and @xmath29 particles on @xmath19 can indeed be neglected , this dependence is essential in the case of the propagator of the gapless ng mode @xmath30 .
the @xmath169term in @xmath170 provides the ( @xmath0)dimensional character of the ng propagator which , as explained in introduction , is crucial for the realization of spontaneous chiral symmetry breaking in the model . in a sense , the magnetic length @xmath21 plays here the same role as the @xmath22parameter in the ( @xmath20)dimensional gn model @xcite . as @xmath23
, the njl model is reduced to a set of independent gn models . however , at finite values of @xmath19 , the transverse velocity of the ng mode , though small [ @xmath171 ,
is not zero .
therefore there are now transitions of field quanta between different points in @xmath125plane , i.e. interactions occur between the gn models associated with different points @xmath172 . as a result , at finite @xmath19 the chiral group
is reduced to global @xmath148 transformations , and there is only one ng boson @xmath30 for this case .
therefore the reduction of the njl model , described in the previous section , takes place only as @xmath23 . at finite values of the magnetic field ,
the dynamics in the njl and gn models are different : while there is spontaneous chiral symmetry breaking in the njl model , the bkt phase is realized in the gn model @xcite .
the connection between these two sets of dynamics is similar to that between the dynamics of 2dimensional and ( @xmath20)dimensional gn models . in conclusion
, we emphasize that this discussion pertains only to the njl model with a weak coupling constant , when relation ( [ eq21 ] ) is valid . in the case of the njl model with a near
critical @xmath82 , the situation is different : when @xmath96 , ( [ eq22 ] ) is valid .
the difference between these two dynamical regimes reflects the fact that , while at weak coupling the lll dominates , at near
critical @xmath82 all landau levels are relevant @xcite . in the next section
, we shall discuss the dynamics in the supersymmetric njl model .
as is well known , there is no spontaneous chiral symmetry breaking in the supersymmetric ( susy ) njl model at any value of the coupling constant @xcite . here
we shall show that an external magnetic field changes the situation dramatically : chiral symmetry breaking occurs for all ( positive ) values of the coupling constant .
moreover , at weak coupling , the dynamics in susy and ordinary njl models are similar ( and , therefore , intimately connected with the dynamics in the ( @xmath16)dimensional gn model ) .
the action of the susy njl model with the @xmath38 chiral symmetry in a magnetic field is @xmath173 . \label{eq33}\end{aligned}\ ] ] here we utilize the notations of ref.@xcite , except for our choice of metric @xmath174 . in eq.([eq33 ] ) , @xmath175 , @xmath176 and @xmath177 are chiral superfields carrying the color index @xmath178 , i.e. @xmath176 and @xmath177 are assigned to the fundamental and antifundamental representations of the @xmath179 , respectively : @xmath180 ( henceforth we will omit color indices ) .
the vector superfield @xmath181 , with @xmath182 , describes an external magnetic field which is in the @xmath62 direction .
the action ( [ eq33 ] ) is equivalent to the following action : @xmath183- \nonumber\\ & & - \int d^6z\left[\frac{1}{g}hs - qq^cs\right ] - \int d^6\bar{z}\left[\frac{1}{g}\bar{h}\bar{s } - \bar{q}\bar{q}^c\bar{s}\right ] .
\label{eq35}\end{aligned}\ ] ] here @xmath184 , @xmath185 , and @xmath186 and @xmath187 are two auxiliary chiral fields : @xmath188 the euler
lagrange equations for these auxiliary fields take the form of constraints : @xmath189 here @xmath190 is a susy covariant derivative @xcite .
the action ( [ eq35 ] ) reproduces eq.([eq33 ] ) upon application of the constraints ( [ eq37 ] ) . in terms of the component fields ,
the action ( [ eq35 ] ) is @xmath191 .
\label{eq38}\end{aligned}\ ] ] let us consider the effective potential in this model . for this purpose , one can treat all the auxiliary scalar fields as ( independent of @xmath41 ) constants and all the auxiliary fermion fields equal zero ( since the auxiliary fields are colorless , loop diagrams involving them do not contribute to the effective potential in leading order in @xmath63 ) . then , using the euler
lagrange equations for the fields @xmath192 , @xmath193 , @xmath194 , @xmath195 and their conjugates , we find @xmath196 , @xmath197 , @xmath198 , @xmath199 , plus h.c . equations .
then the action becomes @xmath200\varphi - \varphi^{c\dagger}\left[(\partial_{\mu}+iea^{ext}_{\mu})^2 + \rho^2 \right ] \varphi^c \nonumber \\ & & + i\bar{\psi}_d\gamma^{\mu}(\partial_{\mu } -iea^{ext}_{\mu})\psi_d - \sigma\bar{\psi}_d\psi_d - \pi\bar{\psi}_di\gamma^5\psi_d - \frac{\rho^2}{g } \bigg ] , \label{eq39}\end{aligned}\ ] ] where @xmath201 , @xmath202 , and the dirac fermion field @xmath203 is introduced . in leading order in @xmath63 , the effective potential @xmath204
can now be derived in the same way as in the ordinary njl model @xcite .
the difference is that , besides fermions , the two scalar fields @xmath205 and @xmath206 give a contribution to @xmath204 : @xmath207 where @xmath208 as is shown in the appendix , the potential @xmath204 can be rewritten as @xmath209/2 } dx \ln\gamma(x)\bigg]+ \nonumber\\ & + & \frac{n_c}{16\pi^2 l^4 } \bigg [ \ln(\lambda l)^2-\gamma -\ln(8\pi^2 ) \bigg ] + o\left(\frac{1}{\lambda}\right ) , \label{eq43}\end{aligned}\ ] ] where the dimensionless coupling constant is @xmath210 .
as the magnetic field @xmath211 goes to zero ( @xmath80 ) , we recover the known expression for the potential in the susy njl model @xcite : @xmath212 the potential @xmath204 is positive definite , as has to be in a supersymmetric theory .
the only minimum of this potential is @xmath213 corresponding to the chiral symmetric vacuum .
the presence of a magnetic field changes this situation dramatically : at @xmath214 , a non trivial global minimum , corresponding to spontaneous chiral symmetry breaking , exists for all @xmath87 .
moreover , we show below that , as the coupling @xmath215 , the susy njl model becomes equivalent to the ordinary njl model and , therefore , at weak coupling and as @xmath216 , it is reduced to the same continuum set of the @xmath126dimensional gn models . on the other hand ,
the dynamics of the susy and non susy njl models in a magnetic field are very different at strong coupling . the gap equation @xmath73 , following from eq.([eq43 ] ) , is @xmath217 = 0 . \label{eq45}\ ] ] as can be seen from ( [ eq43 ] ) , at @xmath214 the trivial solution @xmath213 to this equation corresponds to a maximum of @xmath75 : @xmath218 .
numerical analysis of equation ( [ eq45 ] ) for @xmath87 and @xmath214 shows that there is a nontrivial solution @xmath219 which is the global minimum of the potential .
the analytic expression for @xmath48 can be obtained at small @xmath82 ( when @xmath220 ) and very large @xmath82 ( when @xmath221 ) . in those two cases ,
the results are : _ a ) _ @xmath90 ( @xmath220 ) .
the gap equation ( [ eq45 ] ) is @xmath222 i.e. @xmath223 = \sqrt{\frac{2|eb|}{\pi } } \exp\left[-\frac{4\pi^2}{|eb|n_cg}\right ] .
\label{eq47}\ ] ] _ b ) _
@xmath224 ( @xmath221 ) .
now , the gap equation ( [ eq45 ] ) gives @xmath225 i.e. @xmath226 the numerical solution to the gap equation ( [ eq45 ] ) for general values of @xmath82 is shown in fig .
let us discuss the case of the weak coupling in more detail . by using the asymptotic series for @xmath227 @xcite , the potential ( [ eq43 ] )
can be rewritten at @xmath228 as @xmath229 .
\label{eq50}\end{aligned}\ ] ] using eq.([eq46 ] ) , the coupling constant @xmath82 can be expressed through @xmath48 , and the potential can be rewritten as @xmath230 .
\label{eq51}\end{aligned}\ ] ] this potential coincides ( up to exponentially small terms ) with that in the ordinary ( weakly coupling ) njl model ( see eq.([eq26 ] ) ) .
now , let us consider the kinetic term in the susy njl model in a magnetic field .
as shown in the appendix , at weak coupling , this term also coincides with that in the ordinary njl model .
therefore , at weak coupling , these two models are equivalent .
note that the equivalence between the susy and non susy njl models becomes explicit only after the renormalization of the coupling constants .
the structure of the ultraviolet divergences in the effective potentials of these models is quite different ( compare eqs .
( [ eq13 ] ) and ( [ eq43 ] ) ) .
this reflects the point that only the infrared ( and not ultraviolet ) dynamics in these models are equivalent , i.e. , in the renormalization group language , these two models are assigned to the same universality class @xcite .
the physical picture underlying this equivalence is clear .
an external magnetic field explicitly breaks supersymmetry : the spectra of charged free fermions and bosons in a magnetic field are essentially different @xcite . while for fermions the spectrum is @xmath231 for bosons it is @xmath232 the crucial difference between them is that while for fermions the energy of the lll is independent of @xmath211 and ( at @xmath233 ) goes to zero as @xmath234 , for bosons the energy of the lll is @xmath235 .
in other words , there is a gap @xmath236 in the spectrum of massless bosons in a magnetic field .
recall that , at weak coupling , the chiral condensate in the njl model occurs because of the fermion pairing in the lll with @xmath233 .
this also happens in the susy njl model , in which the bosonic degrees of freedom become irrelevant at weak coupling .
notice that , in the effective action , the field @xmath111 plays the role of the mass . in particular , the infrared singularities ( as @xmath237 ) in the expressions for the potential and the kinetic term ( see eqs.([eq26 ] ) and ( [ eq27 ] ) ) reflect the absence of a gap in the lll for massless fermions .
we emphasize that the equivalence between these two models takes place at weak coupling only .
when @xmath82 becomes larger than @xmath95 , the dynamics in these models are essentially different . while at @xmath83 in the ordinary njl model , the chiral symmetry is spontaneously broken even without an external magnetic field , there is no spontaneous chiral symmetry breaking in the susy njl model as @xmath238 at any value of @xmath82 ( see eq.([eq49 ] ) ) .
the difference between the dynamical regimes with weak and strong coupling corresponds to the fact that , while at weak @xmath82 the dynamics of the lll of fermions dominates , at @xmath239 all ( fermionic and bosonic ) landau levels become relevant .
in this paper we have studied the infrared dynamics of both ordinary and susy njl models in a magnetic field .
it has been shown that , at weak coupling , the infrared dynamics in these two models are equivalent and , as @xmath23 , the models reduce to a continuum set of @xmath126dimensional gn models . in this paper , as in refs.@xcite , only the case of a homogeneous magnetic field has been considered . to extend the present results to inhomogeneous field configurations ,
we note first that the number of colors in the gn model is @xmath9 , where the factor @xmath240 is proportional to the @xmath241 magnetic flux attached to each point in @xmath125plane .
let us consider a magnetic field @xmath242 , directed in @xmath62 direction but depending on the transverse coordinates .
it is tempting to speculate that in this case , as @xmath243 , the njl model will be reduced to a set of the gn models with the number of colors @xmath244 ( where @xmath245 ) , which is different in different points of the @xmath125-plane
. it would be interesting to check these speculations by studying the njl model in inhomogeneous field configurations .
the results of the present paper are in agreement with the general conclusion of refs.@xcite , that the catalysis of chiral symmetry breaking by a magnetic field is a universal , model independent effect .
this catalysis may have possible applications to cosmology , particle physics and condensed matter physics @xcite .
a specific application of interest would be a linkage between catalysis of chiral symmetry breaking and the existence of very strong primordial magnetic fields in the early universe @xcite .
the results of the present paper may be especially relevant for cosmological models based on supersymmetric dynamics @xcite .
also , the effect of the dimensional reduction by external fields may be quite general and relevant for multi - dimensional field theories .
this research is supported in part by the natural science and engineering research council of canada ( nserc ) .
in this appendix , the effective action for the susy njl model in a magnetic field is derived . besides the expression for the fermion propagator in a magnetic field , which was used in refs.@xcite
, we also need to know the expression for the propagator in a magnetic field for a charged scalar with the mass @xmath246 @xcite : @xmath247 \tilde d(x - y)~ , \label{eqa1}\end{aligned}\ ] ] where the fourier transform of @xmath248 is @xmath249 .
\label{eqa2}\end{aligned}\ ] ] the expression ( [ eq40 ] ) for the effective potential in the susy njl model can be obtained in the same way as for the ordinary njl model in ref.@xcite .
let us show that this expression is equivalent to expression ( [ eq43 ] ) .
eq.([eq40 ] ) can be rewritten as @xmath250 + o\left(\frac{1}{\lambda}\right ) \label{eqa3}\end{aligned}\ ] ] where @xmath251 , \qquad \beta>0 . \label{eqa4}\end{aligned}\ ] ] let us show that @xmath252 using the integral representation for generalized zeta function @xcite , we find @xmath253= \nonumber\\ & = & \lim_{\mu\to-1}\gamma(\mu)\bigg [ 2^{1-\mu}\zeta\left(\mu,\frac{\beta}{2}\right)-\beta^{-\mu } -2^{1-\mu}\zeta\left(\mu,\frac{\beta+1}{2}\right ) - \frac{\mu}{2 } \beta^{-1-\beta}\bigg ] .
\label{eqa6}\end{aligned}\ ] ] in order to find this limit , we need the following identity : @xmath254 \right|_{z=-1}=\frac{q_2-q_1}{2}(q_2 + q_1 - 1 - \ln{2\pi } ) + \int\limits_{q_1}^{q_2 } dq \ln\gamma(q ) .
\label{eqa7}\end{aligned}\ ] ] to derive it , we use the relation @xcite @xmath255 differentiating it with respect to @xmath256 and integrating over @xmath257 , we obtain @xmath258 = -\int\limits_{q_1}^{q_2 } d q \zeta(z+1,q ) - z \int\limits_{q_1}^{q_2 } dq \frac{\partial}{\partial z } \zeta(z+1,q ) .
\label{eqa9}\end{aligned}\ ] ] then , using the identities @xcite @xmath259 we obtain eq.([eqa7 ] ) .
let us show that , at weak coupling , the kinetic term in the effective action of the susy njl model coincides with that in the ordinary njl model . for this purpose
, we shall prove that the contribution of the scalar fields in the kinetic term is suppressed . by using the approach of ref.@xcite
, one finds that the functions @xmath261 and @xmath262 in the kinetic term ( [ eq14 ] ) are now equal to : @xmath263 where @xmath264 are as given in ( [ eq15 ] ) .
the crucial point for us is that , while the fermionic contribution ( [ eq27 ] ) in @xmath265 is divergent as @xmath266 , the contribution of the scalars is finite in this limit .
this conclusion , following directly from eqs.([eqa2 ] ) and ( [ eqa13 ] ) , reflects the fact that there is a gap in the spectrum of massless charged scalars in a magnetic field ( see eq.([eq53 ] ) ) .
since , at weak coupling , the parameter @xmath89 is exponentially small , we conclude that the contribution of the scalars is indeed suppressed at small @xmath82 . in connection with that , note that the dynamics in the ordinary njl model in a magnetic field has been recenltly analyzed in the framework of the renormalization groug approach in d.k .
hong , y. kim , and s .- j .
sin , phys .
d*54 , * 7879 ( 1996 ) .
also note that , because of the presence of the ( neutral ) photon field , the effect of the dimensional reduction is more complicated in qed in a magnetic field ( see refs.@xcite and l.n .
leung , y.j .
ng , and a.w .
ackley , phys .
d*54 , * 4181 ( 1996 ) ) . the fact that in the supercritical phase of the ordinary njl model , @xmath48 increases with @xmath19 has been already pointed out by several authors who discussed possible applications of this effect in nuclear physics and astrophysics : s.p .
klevansky and r.h .
lemmer , _ phys .
_ d*38 , * 3559 ( 1988 ) ; h. suganuma and t. tatsumi , _ ann .
phys . _ * 208 , * 470 ( 1991 ) ; s. schramm , b. mller , and a.j . schramm , _ mod .
_ a*7 , * 973 ( 1992 ) .
turner and l.m .
widrow , _ phys .
_ d*37 , * 2743 ( 1988 ) ; t. vachaspati , phys .
b*265 , * 258 ( 1991 ) ; k. enqvist and p. olesen , phys .
b*319 , * 178 ( 1993 ) ; j. m. cornwall , in _ unified symmetry : in the small and large _ , eds .
b. n. kursunoglu _
( plenum press , new york , 1995 ) ; g. baym , d. bodeker , and l. mclerran , _ phys . rev .
_ d*53 , * 662 ( 1996 ) . | the infrared dynamics in the ( @xmath0)dimensional supersymmetric and non supersymmetric nambu jona lasinio model in a constant magnetic field is studied . while at strong coupling the dynamics in these two models is essentially different , it is shown that the models become equivalent at weak coupling . in particular , at weak coupling , as the strength of the magnetic field goes to infinity , both the supersymmetric and non supersymmetric nambu
jona
lasinio models are reduced to a continuum set of independent ( 1 + 1)dimensional gross neveu models , labeled by the coordinates in the plane perpendicular to the magnetic field .
the relevance of these results for cosmological models based on supersymmetric dynamics is pointed out . |
galaxies appear to form stars in at least two modes ; a quiescent and a starburst mode .
an observational relation revealing these modes is the kennicutt
schmidt relation between the gas surface density ( @xmath1 ) and the star formation rate per surface area ( @xmath2 ) within a galaxy @xcite .
quiescently star - forming disc galaxies typically follow a power law relation , @xmath3 @xcite , whereas starburst galaxies follow a relation with a similar power index , but a @xmath4 times higher normalisation , implying that gas is converted much faster into stars in these galaxies .
a key element for characterizing these modes is hence the timescale with which star formation consumes the gas within a galaxy .
in local disk galaxies , a linear relation exists between the star formation surface density , @xmath2 , and the surface density of molecular hydrogen , @xmath5 @xcite .
a plausible interpretation of this linearity is that star formation is physically tied to the presence of molecular gas , rather than to the total gas surface density . as for the @xmath6 relation ,
the @xmath7 also breaks down for starbursting galaxies such as ultra - luminous infrared galaxies ( ulirgs , * ? ? ?
* ) , where molecular gas is converted into stars more rapidly than for quiescently star - forming galaxies . while the observational picture thus clearly shows that a bursty mode of star formation exists with roughly 15 times shorter gas depletion timescales than for the normal star formation mode , it is less clear what conditions are required to trigger this accelerated star formation . observations also show a redshift - dependent relation between the sfr and stellar mass of galaxies @xcite .
galaxies being significantly more star - forming than predicted by this relation are also often referred to as starbursts @xcite , even though such a selection criterion does not select the exact same population of galaxies as a selection by gas depletion timescales @xcite .
the term _ starburst _ is thus ambiguous , with some definitions relying on gas consumption timescales ( of the ism gas , or only the dense star - forming gas ) , while other definitions rely on the sfr@xmath8-value of the galaxy , or alternatively on the deviation of a galaxy from the mean sfr@xmath9 relation . throughout this paper
we define the term starburst based on the gas consumption timescale derived from the sfr and the gas mass of a galaxy . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] hydrodynamical simulations of isolated major mergers can account for the high star formation rates seen in starbursting ulirgs @xcite .
radiative transfer analysis of major merger simulations has also reproduced their high submillimeter fluxes @xcite .
most of the literature studying major mergers has however employed idealized setups , where the initial conditions were equilibrium disc galaxies and the galaxy orbits have usually been collided on zero - energy orbits .
additionally , ad - hoc choices for the relative orientation of the galaxy discs and for the amount of orbital angular momentum have to be made .
an exception to this idealised approach is the work by @xcite , who analyzed the morphological evolution of a major merger at @xmath10 , finding that the @xmath11 merger remnant had a significant disc surrounding a central bulge .
the hydrodynamical simulation illustris is a large - scale cosmological simulation in which many major mergers occur @xcite .
the simulation has shown interesting successes in describing the evolution of star - forming galaxies @xcite , and in the formation of elliptical and spiral galaxies @xcite .
a short - coming of the simulation is , however , that it did not produce enough starburst galaxies with a very high star formation rate compared to the rest of the galaxy population @xcite .
a potential reason for this deficit of galaxies with large sfrs is that the @xmath12 pc resolution of illustris is not fine enough to resolve the central starbursting regions within galaxies sufficiently well .
the aim of this paper is to study the starbursting gas in high - resolution zoom - in simulations of major mergers identified in the illustris simulation .
these simulations are cosmological , self - consistent simulations , unlike the majority of the major merger calculations in the literature , and hence remove the reliance on idealised initial conditions and prescribed orbital parameters .
we select individual major merger remnants from illustris and repeat the simulation with spatially variable resolution , allowing us to zoom in on the galaxies of interest with higher resolution . with this approach it is possible to study
selected galaxies at a 40 times better mass - resolution than in illustris .
our analysis focusses on the conditions under which starbursting gas with a short consumption timescale appears in galaxy mergers .
we will also discuss whether the reason for the lack of starburst galaxies in the original illustris run is caused by resolution limitations for resolving the starbursting gas ( as suggested in * ? ? ? * ) .
this paper is structured as follows . in section
[ methods ] , we describe our initial conditions and the physical galaxy formation model , and in section [ gassection ] we discuss the gas and star formation properties of the galaxies before , during and after the starburst associated with the major mergers . in section [ orbitsection ] , we assess the orbit of the mergers , and see how it affects the starburst properties of the mergers .
finally , in section [ discussion ] we discuss our results and in section [ conclusion ] we summarize our conclusions .
we use the same physical model as for the auriga simulation project ( first presented in * ? ? ?
this model is closely based on @xcite and @xcite .
the hydrodynamical calculations are carried out with the moving mesh code arepo ( @xcite , see also @xcite ) , and star formation and winds are described with the @xcite model that treats the interstellar medium ( ism ) with a subgrid approach , where the star - forming clouds are not resolved , but instead each star - forming gas cell has an assigned fraction of cold gas and a hot ambient phase based on a simple analytic prescription .
stars form out of the cold phase , and the energy released by supernova type ii is assumed to reheat some of the cold gas and transfer it to the hot gas . together with the evaporation of cold gas trough thermal conduction , this creates a pressurized star forming medium .
star formation occurs in gas cells with a density above @xmath13 @xmath14 .
the star formation timescale of a gas cell is set to @xmath15 , so it is proportional to the local dynamical time .
stellar population particles are used to represent single age stellar populations with an initial mass function from @xcite .
mass loss and chemical enrichment of these stellar population particles are tracked continuously .
when mass is returned to the ism , the mass in various elements ( h , he , c , n , o , ne , mg , si and fe ) is distributed to the 64 nearest gas cells .
kinetic feedback from sn type - ii in the form of winds is created by probabilistically converting star - forming gas cells into wind particles .
the winds have an initial speed that scales with the local dark matter velocity dispersion , which gives a remarkably better fit to the sfr@xmath9 than , e.g. , a constant wind velocity @xcite .
the winds are briefly decoupled from the hydrodynamical interactions after their release , until they reach a gas density slightly below the star formation threshold .
this decoupling and recoupling procedure ensures that the winds escape the star - forming regions without disrupting them , and that the winds actually attain the prescribed mass - loading factor .
when a wind particle recouples , its momentum , thermal energy and metal content is dumped into the nearest gas cell , so that the winds effectively constitute a non - local feedback that occurs at the transition layer between star - forming and non - starforming gas .
black holes are seeded when a halo reaches a mass of @xmath16 , and the adopted seed mass is @xmath17 .
the black holes grow either by bondi - hoyle or eddington - limited accretion ( the smaller of the two rates applies ) .
the feedback _ mode _ of the black hole is determined by the accretion rate @xcite relative to the eddington rate . for high accretion rates the black hole
is assumed to be in the _ quasar mode _ , where thermal feedback heats the gas in the vicinity of the black hole , while in the low - accretion mode , feedback acts mechanically in a _ radio mode _ ,
which is here implemented as a distributed heating of the halo gas .
furthermore , we use a uniform uv background @xcite , and gas cooling rates ( see @xcite for details ) with self - shielding corrections from @xcite .
more details of the physical models will be presented in the main auriga project paper ( grand et al . in prep . ) .
initial conditions are created by selecting galaxies from the illustris simulation @xcite at @xmath11 that experienced a major merger between @xmath18 and @xmath19 . by selecting galaxy mergers in this redshift interval
we ensure that the mergers have time to finish before @xmath11 , so we can study , e.g. , how the relaxed stellar system at @xmath11 is affected by the merger .
we base our selection on the merger trees from illustris .
we have created zoom initial conditions ( ics ) for 4 major mergers using a modified version of the n - genic code @xcite that had also been used to create the ics of the illustris simulation .
the code coarsens the resolution of the initial particle load progressively with distance from the galaxy of interest , while increasing the resolution inside the lagrangian region of the galaxy and its immediate surroundings . in the resulting setup , we distinguish three types of dark matter particles ( dm ) : the high resolution particles , the standard resolution particles ( with resolution very close to the original illustris resolution ) and the low resolution particles ( further away from the target galaxy ) . in the initial particle load ,
the standard resolution particles form a shell around a roughly spherical high - resolution region , with the low resolution particles filling the rest of the simulation box . for each particle type , the displacement field
is sampled in fourier space up to the nyquist frequency of the corresponding particle load .
all waves that were present already in illustris are kept so that the same object forms again , but augmented with additional small - scale power in the initial conditions .
the standard resolution particles in our setup have a mass of @xmath20 ( here @xmath21 is the total dm mass in the box ) , and the high resolution particles have a mass which is lowered by a ` zoom factor ' cubed . the low resolution particles have a variable mass that depends on the distance to the galaxy of interest .
baryons are added to the simulation by putting the cosmologically expected fraction of the mass of each dark matter particle into baryonic gas cells .
the corresponding mesh generating points and dark matter particles are displaced relative to each other such that the center - of - mass of each dark matter and cell pair is unchanged , and the mean distance between dark matter and mesh generating points is maximized .
this is done to avoid artificial pairing effects if a small gravitational softening length is adopted .
the finest dark matter mass resolution level is set by the ` zoom factor ' according to the formula , @xmath22 the factor of 1820/2048 is the ratio between the number of dark matter particles per dimension in illustris , and our standard resolution . for each galaxy
we ran simulations with a zoom factor of 1 , 2 and 3 , which corresponds to mass resolutions 1.4 , 11.4 and 38.5 times finer than in the illustris simulation .
the gravitational softening was chosen to be @xmath23 , where @xmath24 is the width of the cosmological box , and @xmath25 is the number of particles per dimension for the standard resolution particles .
our dark matter mass resolution for the runs with a ` zoom factor ' equal to 3 is therefore 1.7 times coarser than the eris simulation @xcite and @xmath26 times finer than in the milky way simulations presented by @xcite . in our simulation , the softenings are fixed to this physical value for @xmath27 . at @xmath28 the comoving softening is fixed to the ( comoving ) value at @xmath18 .
the main properties of the four major merger simulations are summarized in table [ simoverview ] .
we list the mass of the most - massive galaxy ( @xmath29 ) , and the second most - massive galaxy ( @xmath30 ) participating in the major merger at @xmath31 , which is before the collisions occur
. the mass ratios of all the major mergers are between 1:1.05 and 1:1.51 . before the mergers , the progenitors have masses in the range @xmath32 , and at @xmath11
the merger remnants have reached masses of @xmath33 . at @xmath11 ,
the galaxies reside in host halos with masses of @xmath34 , where @xmath35 is the mass within a sphere centered on the halo containing a mean density of 200 times the critical density of the universe .
all our galaxies therefore have masses close to the milky way at the present epoch , both in terms of @xmath9 and @xmath35 ( according to recent constraints the milky way has a stellar mass of @xmath36 and a virial mass of @xmath37 ; * ? ? ?
each simulation contains two massive galaxies before the merger , while after the merger event one massive galaxy is left .
a significant part of our analysis in this paper relies on merger trees where we track the galaxy in each simulation with the largest stellar mass at @xmath31 immediately before the merger . to this end
we store the i d of each star in the galaxy , and at any other redshift , the main descendant or progenitor of the galaxy is identified with the object that has the largest number of stars in common with the @xmath31 galaxy .
in figure [ allsfr ] , we plot the star formation histories of our simulated galaxies based on the formation time and initial mass of the star particles found at @xmath11 in the remnant galaxies .
the star formation rate ( sfr ) has been calculated in bins of width 28 myr .
each panel shows a galaxy simulated at different resolution , with mass resolutions 1.4 , 11.4 and 38.5 times finer than in the illustris simulation .
common for all simulations is that one or several peaks in the sfr history are present at lookback times between 5 and 8 gyr , which is where the merger happens according to the selection criteria in section [ mergerselection ] . for each star formation history ,
the ` merging time ' , @xmath38 , defined to be the time with the largest black hole accretion rate in the redshift - interval @xmath39 is marked .
furthermore , we define the start and end time ( @xmath40 and @xmath41 ) of the merger phase as 1.5 gyr before and after this merging time , respectively .
compared to the sfr - peak , which appears during the merger ( i.e. with @xmath42 ) , the galaxies evolve quiescently before and after the merger .
an interesting feature is that the high - resolution runs have larger sfr - peaks than the low resolution runs . and
it is not only the merger - induced sfr - peak that is larger for the high - resolution runs , but also the integrated sfr during the merger , which is @xmath43 times larger for the highest compared to the lowest resolution run ( the amount of stellar mass formed within 1.5 gyr of the burst is noted in the upper right corner of each panel ) .
this result that an increased resolution of the simulation leads to more stars formed during a merger - induced starbursts potentially resolves the problem with too little star formation occurring in starbursts in the illustris simulation @xcite . at @xmath11 , the stellar
mass present in the high - resolution runs is also 0.14 - 0.24 dex larger than for the low resolution runs , see table [ simoverview ] .
this difference in stellar mass is smaller than the difference in stellar mass formed during the merger epoch ( @xmath44 dex , the numbers are shown in the upper right part of each panel in figure [ allsfr ] ) .
this is because the galaxies have a more similar star formation history after the merger is finished than during the merger .
the 1605 - 1 , 1605 - 2 and 1605 - 3 simulations , for example , illustrate this in a clear way : for lookback times larger then 5 gyr , the sfr is roughly two times larger for the best resolved simulations compared to the low resolution runs . at smaller lookback times
these differences go away , and at a lookback time of 2 gyr , the sfr is essentially the same in all the simulations . at @xmath11
there is actually a slightly larger sfr in the 1605 - 1 run than in the 1605 - 3 run , which is most likely because the latter simulation used up more of the gas available for star formation earlier in its history . in idealized mergers
, @xcite found a similar result , namely that increasing the resolution in major mergers also increased the peak star formation .
their galaxies showed , however , a more similar behaviour in the pre- and post - burst phases than our cosmological simulations .
various explanations are possible for these differences .
first , it is possible that a self - consistent inclusion of cosmological structure formation as done in our simulations might give more star formation on smaller scales compared to the idealized simulations , where the disc initially follows a smooth mass profile .
second , it is also possible that our hydrodynamical scheme is better at resolving , e.g. , subsonic turbulence than the smoothed particle hydrodynamics method used in the idealized simulations @xcite
. finally , the simulations of @xcite also adopted a different physical galaxy formation model that did not include feedback from agn , which we speculate could change the convergence behaviour of our simulations .
@xcite also studied idealized mergers and found that the star formation peaks are quite sensitive to changes of resolution , the numerical hydrodynamical scheme , and whether or not black hole feedback was enabled .
this is consistent with our interpretation . especially at high redshifts in the range @xmath45 ,
the star formation rates of the galaxies are not fully converged ; the high - resolution simulations typically have up to 2 times higher sfr than in the low - resolution simulations .
similar convergence issues were seen in the @xmath46 box simulations of @xcite performed with the illustris physics model . in section [ sfrrho ]
we will elaborate on this further . to estimate the role of agn feedback we have performed a simulation of the 1605 - 2 galaxy without black hole formation and agn feedback ( _ grey - dashed line _ in the _ lower right panel _ of figure [ allsfr ] ) .
the run without black holes have an up to two times larger sfr during and before the merger .
this is not surprising given that the black hole is responsible for heating and removing gas in the center of galaxies .
in the post - merger regime the ratio between the sfr in the run with and without black holes is larger .
this is expected because a black hole grows rapidly during a merger .
the figure hence reveals that the feedback associated with black holes is crucial for determining the post - merger evolution , but it plays a slightly minor role before and during the merger . before studying the starburst properties of major mergers
we now discuss how we separate the gas in halos into an ism phase , which contains cold gas and where star formation occurs , and a hot phase , which consists of gas heated by accretion shocks and stellar winds from feedback effects associated with star formation . distinguishing these phases
is typically based on the position of a gas cell in the density temperature diagram .
an example is given in figure [ rhot ] ( _ left panel _ ) , where we have constructed a two - dimensional histogram using a @xmath47 grid in the @xmath48@xmath49 plane for the 1330 - 3 galaxy . the logarithmically scaled intensity of the grey color encodes the amount of mass in each cell of the histogram .
the range of mass in the ( grey ) non - empty histogram cells covers five orders of magnitudes .
the behaviour of the gas in this diagram is closely linked to the galaxy formation model .
the thin one - dimensional curve starting at @xmath50 and extending to higher densities is the location of the star - forming gas , which follows the effective equation of state resulting from our subgrid - physics model .
as described in section [ physmodel ] , star formation occurs most rapidly at high densities . formally
, we will base our definition of the hot and the ism gas on the _ the dashed line _ , which takes the form @xmath51 as suggested by @xcite .
we will generally refer to the gas above this relation in the @xmath52 diagram as _ hot gas _ , and the gas below as ism gas ( with mass @xmath53 ) .
the mass of the star - forming gas , which is a subset of the ism gas , is denoted by @xmath54 in the remaining part of this paper .
the hot gas is mostly located at lower densities and higher temperatures than the star - forming gas .
important processes for heating the gas to such high temperatures are accretion shocks , which heat gas to the virial temperature of a halo , and heating by feedback .
the part of the ism which is not star - forming is at similar densities as the hot gas , but with a much lower temperature .
the spatial distribution of the gas phases is shown in the _
central panel_. here the hot gas is shown in _
red _ , the ism gas in _ green _ and the star - forming gas in blue .
a logarithmic scale is used for all of the colors . with
our chosen color - coding the star - forming gas appear _
white _ and is the densest part of the ism gas , making up a disc - like structure
. the hot gas on the other hand dominates outside of the disc . also shown in the figure is the stellar morphology ( _ right panel _ ) .
the image has been constructed based on the @xmath55 , @xmath56 and @xmath57 band luminosities ( they represent _ blue _ , _ green _ and _ red _ , respectively ) of each stellar population particle belonging to the galaxy .
again , the colors represent the logarithm of the luminosity .
the blue ( young ) stars show a ring- or disc - like morphology which closely traces the ism gas .
this is of course expected since stars are formed out of the ism .
another visible feature in the stellar distribution is the presence of a red bulge in a region with very little ism gas .
the most important characteristic of a starburst galaxy is the timescale over which gas is consumed by star formation ( for an overview , see * ? ? ?
to reveal the presence of a bursty star formation mode in our simulations , we show sfr versus @xmath53 in figure [ plotsfrversusmism ] .
also shown are contours with constant gas depletion timescale , @xmath58 . in the literature , similar diagrams revealing the gas depletion timescale are often used to study the star formation laws of observed galaxies @xcite .
the contours with a gas depletion timescale of 1 gyr and 10 gyr correspond to our bursty and normal mode , respectively . of course
the gas consumption timescales characteristic of the two modes depend on the definition of the gas mass . for our simulations
the consumption timescale calculated based on the star - forming gas would for example be around four times shorter than for the total ism gas .
similarly one would get a different timescale if only h@xmath59 or hi gas was used .
some caution should thus be taken when comparing the timescales of our modes with observed consumption timescales . in our analysis ,
the gas quantities are calculated by summing over all ism gas cells within 80 ( physical ) kpc of the center of the main galaxy .
this choice of radius is used to avoid a sudden transfer of mass between the two merging galaxies during their encounters , where the halo finder would classify the primary progenitor as a central galaxy , and the secondary galaxy as a satellite , and therefore momentarily associate a too large stellar mass with the primary galaxy ( @xcite , see also @xcite ) .
a consequence of this definition is that our measurement of sfr and @xmath53 will contain all the gas within both galaxies even before they physically merge . during a merger ,
galaxies trace out a loop in the sfr@xmath53 plane .
this is for example clearly seen for galaxy 1349 - 2 , which goes through the following three phases : * : before the merger the galaxy forms stars quiescently with a gas depletion timescale of @xmath60 yr , and an ism gas mass of @xmath61 .
closer to the merger , new ism gas enters the galaxy and the ism mass increases to a value of @xmath62 , while the ism depletion timescale remains constant . during this step
the galaxy form stars at a gas depletion timescale characteristic for the _ normal star formation mode _ of galaxies . * : closer to the coalescence of the two galaxy cores , the galaxy evolves vertically in the sfr@xmath53 diagram . here
the ism gas mass remains constant , but an increased sfr appears because the gas depletion timescale becomes shorter . as we will see in section [ sfrrho ] ,
the gas depletion timescale shortens because the gas is compressed to high densities .
* : immediately after the cores have merged , ism gas is turned into stars at a short gas depletion timescale of @xmath63 yr . for some time
after the merger the galaxy s gas depletion timescale remains low .
the ism gas mass declines at an approximately constant @xmath64-value of @xmath65 yr from @xmath66 to @xmath67 .
this constant @xmath64-value shows that for some time after the merger , the galaxy effectively has a memory of the merger event . at later times , the @xmath64-value increases again to a value close to ( but slightly shorter than ) the pre - merger value .
this picture is not only valid for galaxy 1349 - 2 , but it also describes the processes happening in the 1349-x , 1526-x and 1605-x - simulations well .
the presence of this evolutionary sequence shows that during a major merger , our galaxy formation models produce a phase of starbursting gas with @xmath68 times shorter gas depletion timescales than the normal mode of star formation .
this is the first time the presence of such starbursting gas is demonstrated in a cosmological simulation when an ism subgrid model like ours is used .
we note that a similar bursty star formation mode has previously been revealed by idealised merger simulations @xcite performed with the ramses simulation code . in section [ discussion ] we will further discuss implications of the starbursts and starbursting gas for hydrodynamical simulations .
the 1330-simulations show however a qualitatively different evolution than described by the above loop .
the post - merger system is left with a larger amount of ism gas than anytime during the merger . in section [ noburst ] , we show that the reason for this is that the sfr - peak of the 1330 - 3 galaxy ( and also for 1526 - 3 ) occurs when the two merging galaxies are several kpc apart , and the gas is therefore not compressed as much as in a nuclear starbursts . in the _ lower right panel _
we again compare the standard 1605 - 2 run to a simulation without black holes .
the two runs behave qualitatively similar with a compression of gas to a shorter gas consumption timescale .
the result that this merger leads to a starburst with a very short gas consumption timescale is therefore not significantly changed by the presence or absence of black hole feedback .
a difference is , however , that the post - merger galaxy ( marked with the @xmath69-symbol ) for the run without black holes has a shorter gas consumption timescale than for the fiducial run .
this is consistent with section [ sfhsubsection ] where we found black holes to be important for shaping the post - merger evolution of the star formation history .
it is possible that the detailed orbit of the black holes in a merger ( as recently modeled by * ? ? ?
* ) can potentially change the exact time , when strong feedback kicks in and heats the star - forming gas .
we speculate that this is more likely to change the duration of a starburst epoch rather than changing whether starbursting gas appear in the galaxy or not . in our star formation model , the timescale on which stars are formed
is closely tied to the local gas density .
the time - evolution of the gas density at which stars are typically formed is further studied in figure [ alldensity.pdf ] , which shows the star - formation rate averaged gas density , @xmath70 here the sum is over all gas cells within 80 kpc of a galaxy . in the figure ,
the density is normalized with respect to the density threshold , @xmath71 , for star formation to occur .
the sfr - weighting effectively makes the density - measurement more sensitive to the very dense gas with high sfrs . for the 1330-x and 1526-x simulations ,
stars are typically formed at gas densities between 3 to 30 times ( i.e. 0.5 - 1.5 dex above ) the star formation threshold .
the two other sets of simulations , 1349-x and 1605-x , show a resolution - dependent behaviour around the burst epoch . the low - resolution galaxies ( 1349 - 1 and 1605 - 1 ) lack very dense star - forming gas , but the simulations with higher resolution contain star - bursting gas with a much shorter star formation timescale . for the 1349 - 3 and 1605 - 3 simulations , the density at which stars are typically formed peaks at around @xmath72 , which corresponds to a star formation timescale of 70 myr . at high densities of @xmath73 ,
the gas in the multi - phase ism model has properties similar to starbursting gas . at such densities
the effective equation of state describing the star - forming gas has a polytropic index , @xmath74 , smaller than @xmath75 , implying that gravitational instabilities set in .
furthermore , above the same density threshold , the star formation timescale is shorter than the cloud evaporation timescale implying that a large fraction of the mass contained in a gas cell is converted into stars before the cold clouds are evaporated by thermal conduction .
when the gas reaches such high densities it will thus form stars much more efficiently than normal star - forming gas .
this instability threshold is marked by the _ dashed horizontal line _ in figure [ alldensity.pdf ] , and it is seen that only the 1349 - 3 and 1605 - 3 simulations reach average densities above this threshold .
it is also worthwhile to point out that an increase in the density at which stars are formed is not only restricted to a short time - interval around the peak of the star formation rate , when the cores of the two galaxies merge .
for all the high - resolution simulations , the sfr - averaged density is elevated by about 0.8 - 1 dex for several gyr after the merger ( compared to the value before the merger ) .
thus , a merger does not only affect the star formation properties of the gas during the coalescence of the cores , but it also impacts the gas in the post - starburst phase .
for the lower resolution run , the gas density at which stars are formed quickly loses its memory about the merger , and shortly after the sfr - peak stars are again formed at similar densities as for the pre - merger galaxy .
such a bursty epoch of star formation ( which is not necessarily coincident in time with a merger ) is also favored by several recent observations of burst cycles in galaxies @xcite .
the presence of very dense starbursting gas in the 1349 - 3 and 1605 - 3 simulations is also reflected in the peak sfr of @xmath76 revealed by figure [ allsfr ] .
this is more than a factor of two higher than for the two high - resolution simulations , 1330 - 3 and 1526 - 3 , that do not contain dense starbursting gas .
similarly , the 1349 - 3 and 1605 - 3 galaxies also reach a shorter gas depletion timescale during the merger according to the sfr@xmath53-plane in figure [ plotsfrversusmism ] .
in section [ sfhsubsection ] we noted that the star formation histories of galaxies are not perfectly converged at redshifts @xmath45 . in figure
[ alldensity.pdf ] , we see that star formation typically occurs in more dense regions , when the resolution of a simulation is increased .
this trend is especially clear at @xmath45 .
this behaviour that star formation occurs at higher densities when the resolution is increased is the reason for the increase in the sfr at increased resolution .
we note that the lines in figure [ alldensity.pdf ] are not so well sampled at high redshift because frequent snapshots were only stored at @xmath10 , which is the primary epoch of interest for our merger studies . to further examine the behavior of the starbursting gas in the major mergers
, we will analyse the morphology of the gas and stars in our galaxies . for each high - resolution simulation with a zoom - factor of 3
we identify the time of the merger induced sfr peak .
we simply select this time as the snapshot with the highest sfr at a lookback time larger than 5 gyr . for each of these snapshots
we plot the density temperature diagram and the spatial distribution of gas and stars ( see figure [ plotdensitymapatmaxsfr ] ) .
all these characteristics have already been introduced in section [ ismexplanation ] .
the density temperature diagram for the galaxies reveals that very dense star - forming gas with @xmath77 only exists for two of the four major mergers .
these two galaxies ( 1349 - 3 and 1605 - 3 ) are , of course , the same galaxies that were found to form stars at high densities in figure [ alldensity.pdf ] .
the stellar morphology reveals that these two galaxies have their sfr peak at the time when the cores of the colliding galaxies merge .
the 1605 - 3 simulation reveals an extremely dense stellar morphology .
the 1349 - 3 simulation also shows two galaxies with merging cores , and in addition , several other galaxies are also visible , and a number of streams mainly caused by the merger .
the gas morphology also shows signs of a nuclear starburst for these systems with a single very dense region with a large amount of star - forming gas ( see _ white blue colors _ in the _ central panels _ in _ row 2 and 4 _ ) .
the two mergers , which do not contain dense star - bursting gas in figure [ plotdensitymapatmaxsfr ] have their sfr peak when their separations are around 10 - 30 kpc according to the stellar light distribution in the _ right panels _
( for simulation 1349 - 3 and 1605 - 3 ) .
again , the star - forming gas closely traces the morphology of the recently formed stars .
the projected gas distributions reveal that the star - forming region of these two galaxies is distributed over a larger area for the nuclear starbursts
. a larger spatial extent of the gas is consistent with the gas being less compressed .
therefore , also the sfr enhancements at the peak are smaller for these galaxies , as revealed by figure [ allsfr ] .
it is natural to ask why some major mergers end up as nuclear starbursts , and others do not .
a key to understand this lies in the orbital parameters of the galaxies , which we study in this section .
most of the existing literature about mergers relies on a range of assumptions about the initial orbits of the galaxies participating in the merger .
typically , galaxies are assumed to be on a keplerian orbit with an energy @xmath78 where @xmath79 is the reduced mass of the system , @xmath80 and @xmath81 and @xmath82 are the masses of the two galaxies participating in the merger .
@xmath83 is the distance between them , and @xmath84 is the angular momentum of the orbit .
elliptical , parabolic and hyperbolic orbits have @xmath85 , @xmath86 and @xmath87 , respectively .
we will now examine how well the orbits of our cosmological mergers fit with such keplerian orbits .
we will track the positions in physical coordinates of the two merging galaxies and the merger remnant in each simulation .
we do this in the rest frame of the @xmath11 merger remnant . to determine the orbital plane , it is necessary to know three points within the plane . as the first two points we will use the positions of our galaxies before the merger at @xmath88 ( for 1330 - 3 we use @xmath89 , because the merger happens slightly later than for the other systems ) . as the third point we use the position of the merger remnant in the first snapshot after the merger occurred .
figure [ orbits1_plane_projection ] shows the galaxy trajectories projected onto the orbital plane . here
the different lines connect the positions of galaxies of our snapshots .
the coordinates used to determine the orbital planes are shown as _ circles _ ( the merger remnant s position ) and _ squares _ ( the positions before the merger ; @xmath89 for 1330 - 3 and @xmath88 for the other galaxies ) .
the orbits for the 1349 - 3 and 1605 - 3 galaxies are clearly head - on mergers , where the galaxies do not circulate each other before they coalesce .
the two other mergers , 1330 - 3 and 1526 - 3 , show a different behaviour , where the galaxies pass each other before they merge .
the two head - on mergers are the systems that lead to nuclear starbursts with extremely dense gas ( see section [ sfrrho ] ) , whereas the other two galaxies do not exhibit strong bursts .
keplerian orbits follow a path in the ( @xmath83 , @xmath90 ) phase - space diagram predicted by eqn .
( [ phasespace ] ) . for a fixed energy @xmath91 and angular momentum @xmath84 ,
it is therefore straightforward to calculate the evolution for the keplerian prediction . to quantify how strongly galaxies deviate from simple keplerian orbits we plot the corresponding phase - space diagram in figure [ orbits ] . for each snapshot of each merger
, we calculate the ( physical ) distance between the two merging galaxies as a function of time .
@xmath90 is obtained by a simple numerical differentiation .
we compare our orbits with a head - on orbit with @xmath92 and @xmath86 .
the speed at which the two galaxies approach each other for such an orbit is the highest possible if the two galaxies participating in the merger were not accelerated by other sources than each other .
we fix the halo mass of each system to the @xmath31 value .
interestingly , the 1349 - 3 is accelerated to a velocity exceeding the @xmath86 energy ( i.e. above the escape velocity in newtonian mechanics ) .
this galaxy experiences additional gravitational kicks from galaxies other than the ones participating in the merger .
in fact , in figure [ plotdensitymapatmaxsfr ] we saw that several galaxies participate in the merger of the 1349-simulation .
it is an important result that in cosmological setups colliding galaxies can approach each other faster than predicted by @xmath86 orbits because of additional velocity kicks from the environment in which the merger takes place .
the presence of gravitational kicks beyond the interaction between the merging galaxies is also visible for the 1330 - 3 simulation , when the galaxies are 260 kpc apart .
the orbit of the 1605 - 3 merger also shows evidence for additional accelerations since the galaxies approach each other at a speed smaller than the @xmath86 orbit until they are at a 150 - 250 kpc separation , at which point they are accelerated to a speed similar to the shown trajectory with @xmath86 .
the 1330 - 3 and 1526 - 3 mergers approach each other at a speed smaller than the @xmath86 orbit .
also visible from the figure is that the mergers that lead to starbursts are those which approach each other at the highest speed . for 1349
- 3 and 1605 - 3 , the galaxies move toward each other with 480 and 320 km s@xmath93 , respectively .
the two characteristics that appear special for simulations with nuclear starbursts are that , ( 1 ) the collisions are head - on , and ( 2 ) the collisions occur at higher velocities than for the galaxies without a nuclear starburst .
a characteristic feature of the ism model adopted in this paper is that the effective pressure of the star - forming gas increases faster with density than for an isothermal gas ( see figure 4 in * ? ? ?
this increased pressure accounts for feedback processes related to star formation .
the approach of not resolving the small - scale structure of the ism explicitly and instead imposing an effective equation of state governing the star - forming gas has been adopted in the illustris simulation as well .
this has helped to make such cosmological hydrodynamical simulations of galaxy formation feasible , and in turn allowed them to contribute to our understanding of the formation of spiral and elliptical galaxies @xcite , the formation of compact quiescent galaxies at @xmath94 @xcite , the colours of satellite galaxies @xcite , and how feedback affects gas accretion and the circumgalactic medium of galaxies ( @xcite and @xcite , respectively ) .
a similar although slightly different in detail prescription of the ism gas has successfully reproduced many galaxy observables in the large - scale cosmological simulation eagle @xcite .
but since both of these approaches do not treat the small - scale structure of the ism realistically , it is expected that the short - time variability in the sfr is generally not completely accounted for . in this study
, we have for the first time produced realistic starburst galaxies within such subgrid frameworks using full cosmological simulations .
we stress that it is an important result of any ism model to be able to produce the enhanced sfrs and short gas depletion timescales as observed in starburst galaxies .
we have here found that this starbursting phase occurs in major mergers .
the compression of gas caused by large - scale gas flows when two galaxies merge is the key driver of the starburst events .
the gas flows succeed in driving starbursts even though we use a heavily simplified ism model that normally does not admit large density fluctuations in the ism . since we have only considered major mergers here
, it is in principle possible that other environmental effects such as minor mergers or filamentary accretion @xcite can also produce starbursting gas .
understanding this will be an interesting subject of future studies . in our low resolution simulations ( with a ` zoom factor ' of 1 ) , the bursty star formation mode with short gas depletion times is essentially absent .
these simulations have a similar resolution as illustris , and it is therefore not surprising that illustris lacks a population of extreme starburst galaxies . with our zoom simulations of major mergers
we have shown that increasing the mass - resolution of the simulations by a factor of 10 to 40 is all what is required for bursty star formation to appear .
this finding is consistent with the speculation in @xcite , where it was suggested that the reason for the dearth of starbursts in illustris is a too low resolution that is insufficient to resolve the sub - kpc starbursting regions .
a completely different approach to form starbursts in hydrodynamical simulations would be to use a different model of the galaxy ism .
@xcite for example use a feedback model that ( attempts to ) describe star - forming clouds inside galaxies .
this creates rapidly fluctuating star formation rates in dwarf galaxies @xcite , which can account for the observed h@xmath95 and far - uv fluxes of local galaxies @xcite . also , the ism has a more clumpy morphology and a larger star formation rate threshold , which also increase the burstiness of these simulations . other examples of more bursty star formation models are @xcite and @xcite . in these models stars
are usually formed in sub - kpc sized clumps of star - forming gas .
we have here shown that it is not strictly necessary to adopt such clumpy ism models in order to produce nuclear starbursts ; the global gas flows in our major mergers in combination with a subgrid prescription of the ism and high resolution are sufficient .
in this paper , we have studied gas properties in a set of high - resolution cosmological zoom simulations of major mergers .
the mergers occur over the redshift range @xmath96 , and the participating galaxies have masses in the range @xmath97 . with our numerical setup
we have re - simulated individual galaxies from illustris with an up to 40 times better mass resolution , but still including a physical treatment of the full cosmological environment surrounding the galaxies of interest . in the present work we have focused on the gas and starburst properties of the major mergers , and our main results are : * high resolution merger simulations reveal a phase of starbursting gas with @xmath9810 times shorter ism gas depletion timescales than normal star - forming galaxies .
this bursty mode of star formation appears when the cores of two merging galaxies coalesce .
a characteristic of the bursty mode is that the star - forming gas is so dense that the ism becomes unstable .
the bursts in our simulations are similar to those observed in ulirgs @xcite .
this is the first time such a bursty star formation mode has been revealed in full cosmological simulations within the framework of the subgrid ism models .
* only two out of four major mergers we studied go through a nuclear starburst episode .
the two other galaxies have their sfrs peak when the two galaxies are still 10 - 30 kpc apart .
our cosmological simulations therefore predict that only a fraction of all gas - rich major mergers go through an intensive nuclear starburst phase .
we find that the major mergers leading to strong nuclear starbursts are head - on mergers , whereas collisions with large impact parameters have their sfr peaks when the galaxies are 10 - 30 kpc apart .
a characteristic of the nuclear starbursts is also that the colliding galaxies have a larger collision velocity than the other systems . in one of the cases ,
the galaxies collide at a speed two times larger than predicted by the @xmath86 parabolic orbit .
we note , however , that these conclusions are only based on a sample of four mergers .
assembling larger samples is clearly desirable to gain better statistics for the behaviour of starburst galaxies in cosmological simulations .
* we argue that it is an important finding that the simple subgrid star formation model , as applied for example in illustris , can produce realistic starbursts with two modes of star formation distinguished by their ism gas depletion timescale .
using such a simple star formation model therefore does not preclude the occurrence of extreme starbursts .
this starbursting mode is , however , only fully revealed at a 10 - 40 times higher mass resolution than used in illustris .
this is still lower than the resolution requirements of simulations that attempt to resolve the ism explicitly .
the main barrier for reproducing the number of starbursting galaxies in illustris thus not appears to lie in limitations of the physics model , but rather in the limited resolution of the simulations ( see also * ? ? ?
most of the literature about major merger simulations relies on idealised set - ups where equilibrium galaxies are collided in isolation ( an exception is @xcite ) .
this involves a large set of ad - hoc assumptions about the structural properties of the involved galaxies and the orbital parameters of the encounter . in this simulation suite
we have taken a different approach and simulated four different major mergers consistently embedded in a @xmath0cdm universe by means of the zoom technique .
this thus opens up the chance to address the physical characteristics of mergers in a realistic cosmological environment , eliminating the parameter freedom that troubles isolated toy simulations . in the future
, we plan to further apply this powerful approach to study the formation , destruction and re - growth of stellar discs before , during and after the occurrence of mergers , respectively
. a related point will be to see how the @xmath11 stellar distribution are affected by mergers at @xmath99 .
we thank the referee for constructive and useful comments .
ms thanks the sapere aude fellowship program .
vs acknowledges the european research council through erc - stg grant exagal-308037 , and the sfb-881 ` the milky way system ' of the german science foundation .
the dark cosmology centre is funded by the danish national research foundation .
the authors like to thank the klaus tschira foundation . | we introduce the ` illustris zoom simulation project ' , which allows the study of selected galaxies forming in the @xmath0cdm cosmology with a 40 times better mass resolution than in the parent large - scale hydrodynamical illustris simulation .
we here focus on the starburst properties of the gas in four cosmological simulations of major mergers .
the galaxies in our high - resolution zoom runs exhibit a bursty mode of star formation with gas consumption timescales 10 times shorter than for the normal star formation mode .
the strong bursts are only present in the simulations with the highest resolution , hinting that a too low resolution is the reason why the original illustris simulation showed a dearth of starburst galaxies .
very pronounced bursts of star formation occur in two out of four major mergers we study .
the high star formation rates , the short gas consumption timescales and the morphology of these systems strongly resemble observed nuclear starbursts .
this is the first time that a sample of major mergers is studied through self - consistent cosmological hydrodynamical simulations instead of using isolated galaxy models setup on a collision course .
we also study the orbits of the colliding galaxies and find that the starbursting gas preferentially appears in head - on mergers with very high collision velocities . encounters with large impact parameters do typically not lead to the formation of starbursting gas .
[ firstpage ] cosmology : theory methods : numerical galaxies : evolution galaxies : formation galaxies : star formation galaxies : starburst . |
three - electron quantum dots are expected to attract a lot of attention in the near future due to several developments , both experimental and theoretical .
first , it was recently demonstrated@xcite that detailed excitation spectra of two - electron quantum dots ( in addition to earlier ground - state measurements@xcite ) can be measured , and theoretically understood , as a function of the externally applied magnetic field .
thus , exploration of the excitation spectra of three - electron quantum dots appears to be a next step to be taken .
second , three - qubit electron spin devices are expected to exhibit enhanced efficiency@xcite for quantum - computing and quantum - information purposes compared to single - qubit and two - qubit ones . in this paper
, we carry out exact diagonalization ( exd ) studies for a three - electron single quantum dot under low and moderate magnetic fields . unlike previous exd studies@xcite that focused mainly on the ground states ( gss ) of circular quantum dots,@xcite we investigate , in addition , the excitation spectra for three electrons in quantum dots with a wide range of anisotropies .
moreover , consideration of anisotropic quantum dots allows us to investigate the structure of the many - body wave functions with respect to strong - correlations effects , such as electron localization and formation of wigner molecules with a linear or zig - zag geometry .
most importantly , we investigate here the feasibility of producing model quantum entangled states ( i.e. , the socalled @xmath2 states@xcite ) , which are often employed in the mathematical treatment of quantum information and which have been experimentally realized with ultracold atoms in linear ion traps.@xcite we note that a main factor motivating our investigations is the different nature of the entangling agent , namely , the electromagnetic field in the case of heavy ions versus the two - body coulomb interaction in the case of electrons .
we further mention other recent proposals in the context of solid state electronic devices for producing three - qubit entanglement . in particular , a scheme based on non - interacting electron - hole excitations in the fermi sea
was investigated in ref . .
unlike our present study that focuses on the effect of the interparticle interaction , however , such _ interaction - free _
entanglement cannot@xcite reproduce the symmetric @xmath2-state [ see eq .
( [ wf3e3212 ] ) below ] . a different proposal@xcite for realizing interaction - free entanglement uses pair - correlation functions to study tripartite entanglement shared among the spins of three fermions in a fermi gas . the exact diagonalization method that we use for the solution of the schrdinger equation corresponding to the hamiltonian of three electrons interacting via a coulomb potential in an anisotropic quantum dot , in conjunction with an analysis employing spin - resolved two - point correlation functions , allows us to gain deep insights into the nature of electronic states and three - qubit entanglement in real solid state devices .
additionally , the exd method provides benchmark results , which could be used for assessment of the adequacy and relative accuracy of certain approximation schemes , including the model heisenberg hamiltonian for three localized electrons arranged in a ring geometry that was most recently used in an investigation of the entangled ground states in a three - spin - qubit system.@xcite
we consider three electrons under zero or low magnetic field @xmath3 in a single quantum dot .
the corresponding many - body hamiltonian is written as @xmath4 where @xmath5 is the dielectric constant of the semiconductor material ( 12.5 for gaas ) .
the single - particle hamiltonian is given by @xmath6 where the last term is the zeeman interaction , with @xmath7 being the effective land factor , @xmath8 the bohr magneton , @xmath9 the perpendicular magnetic field , and @xmath10 the spin projection of an individual electron .
the kinetic contribution in eq .
( [ hone ] ) is given by @xmath11 ^ 2}{2m^ * } , \label{hkin}\ ] ] with @xmath12 being the effective mass ( 0.067@xmath13 for gaas ) and the vector potential @xmath14 being taken according to the symmetric gauge .
the external confining potential is denoted as @xmath15 , where @xmath16 .
the external potential is modeled by an anisotropic 2d oscillator @xmath17 which reduces to a circular parabolic qd for @xmath18 .
the ratio @xmath19 characterizes the degree of anisotropy of the quantum dot , and it will be referred to thereafter as the anisotropy parameter .
results will be presented for three cases : @xmath20 ( circular ) , @xmath21 ( slightly anisotropic ) , and @xmath22 ( strongly anisotropic ) .
we find the eigenstates of the many - body hamiltonian ( [ h3e ] ) using an exact diagonalization method .
accordingly , we expand the many - body wave function as a linear superposition , @xmath23 where @xmath24 denotes a slater determinant made out of the three spin - orbitals @xmath25 , @xmath26 , and @xmath27 . for the spin orbitals , we use the notation
@xmath28 if @xmath29 and @xmath30 if @xmath31 [ and similarly for @xmath26 and @xmath27 ] .
@xmath33 is the maximum number of space orbitals @xmath34 that are considered , with @xmath35 and @xmath36 where @xmath37 and @xmath38 denote up and down spins , respectively .
the space orbitals @xmath34 are taken to coincide with the real eigenfunctions of a 2d anisotropic oscillator , that is , the index @xmath39 and @xmath40 , with @xmath41 being the eigenfunctions of the corresponding one - dimensional oscillators in the @xmath42(@xmath43 ) direction with frequency @xmath44(@xmath45 ) . the parity operator @xmath46 yields @xmath47 , and similarly for @xmath48 .
the total energies @xmath49 and the coefficients @xmath50 s are obtained through a direct numerical diagonalization of the matrix eigenvalue equation corresponding to the hamiltonian in eq.([h3e ] ) .
for the solution of this large scale , but sparse , matrix eigenvalue problem , we have used the arpack computer code.@xcite , with @xmath51 as a function of the magnetic field for @xmath0 non - interacting electrons in a circular quantum dot @xmath52 .
parameters : external confinement @xmath53 mev ; dielectric constant @xmath54 ; effective mass @xmath55 , effective land coefficient @xmath56 .
the labels @xmath57 denote the quantum numbers for the total spin and the total angular momentum .
different landau bands are denoted by the different @xmath58 values .
the @xmath59 indices are not indicated , since the multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 .
, width=302 ] the exd wave function ( [ wf3e ] ) preserves by construction the third projection @xmath59 of the total spin , since only slater determinants with a given @xmath59 value are used in the expansion .
the exact diagonalization automatically produces eigenfunctions of the square , @xmath62 , of the total spin @xmath63 .
the corresponding eigenvalues @xmath64 are calculated with the help of the expression @xmath65 a slater determinant and the operator @xmath66 interchanges the spins of electrons @xmath67 and @xmath68 provided that their spins are different ; @xmath69 and @xmath70 denote the number of spin - up and spin - down electrons , respectively , while @xmath71 denotes the total number of electrons .
since the spin orbitals @xmath72 s are orthogonal , the coulomb matrix elements between two slater determinants are calculated using the slater rules,@xcite and the necessary two - body matrix elements between space orbitals @xmath73 are calculated numerically . we have found that this method produces numerically stable results in comparison with algebraic expressions.@xcite , with @xmath51 as a function of the magnetic field for @xmath0 non - interacting electrons in a anisotropic quantum dot with anisotropy parameter @xmath74 .
parameters : external confinement @xmath75 mev ; @xmath76 mev ; dielectric constant @xmath54 ; effective mass @xmath77 , effective land coefficient @xmath56 .
the labels @xmath57 denote the quantum numbers for the total spin and the total angular momentum in the corresponding circular quantum dot .
although the total angular momentum is not a good quantum number for an anisotropic quantum dot , we retain the labels @xmath78 here in order to facilitate comparison with the circular case in fig .
[ enspnint ] .
the @xmath59 indices are not indicated , since the multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 .
, width=302 ]
in this section , we study the ground - state and excitation spectra as a function of an increasing magnetic field @xmath9 with an emphasis on the role of correlation effects and the influence of the anisotropy . , with @xmath51 as a function of the magnetic field for @xmath0 interacting electrons in a circular quantum dot @xmath52 .
parameters : external confinement @xmath79 mev ; dielectric constant @xmath80 ; effective mass @xmath55 ; effective land coefficient @xmath56 .
the labels @xmath57 denote the quantum numbers for the total spin and the total angular momentum .
the @xmath59 indices are not indicated , since the multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 . , width=283 ] to better understand the importance of correlations , we first display in fig.[enspnint ] the spectra in the absence of the coulomb interaction ( non - interacting electrons ) and for the case of a circular quantum dot .
these energy spectra can be determined simply as @xmath81 , where @xmath82 are the darwin - fock energies for a single electron.@xcite the main trend is the formation of three - particle landau bands ( each with an infinite number of states ) that tend for @xmath83 to the asymptotic energy levels @xmath84 , @xmath85 . note that , for large magnetic fields ( @xmath83 ) , the reference energy @xmath86 , with @xmath51 , tends to @xmath87 . in this limit ,
the states @xmath88 , belonging to the same landau band @xmath58 , become degenerate in energy , converging to the corresponding familiar landau level ( with index @xmath58 ) .
apart from an overall constant , the picture in fig .
[ enspnint ] is the same as that found in the phenomenological `` constant - interaction ''
model.@xcite an important property is the absence of crossings between individual levels within each landau band .
a consequence of this is that the ground state at any @xmath9 has the same quantum numbers as the one at @xmath89 , i.e. , it has total spin @xmath90 and total angular momentum @xmath91 .
the absence of crossings within each landau band is a characteristic property of non - interacting electrons , and it is independent of the anisotropy of the external confinement .
this point is illustrated in fig .
[ enspninteta ] where the non - interacting three - electron spectra are plotted for the case of a quantum - dot with moderate anisotropy ( @xmath21 ) .
[ for the single - electron energies @xmath82 in an elliptic quantum dot , see ref . and ref .
an inspection of fig .
[ enspninteta ] shows that the anisotropy has an effect mainly for small values of the magnetic field ( by lifting the degeneracies at @xmath89 ) . on the formation of the landau bands at higher @xmath9
, the anisotropy has practically no effect , and , in particular , it can not induce level crossings within each landau band ) .
another property of the non - interacting spectra is the existence of several degenerate levels ( not shown in figs .
[ enspnint ] and [ enspninteta ] ) associated with the excited states .
we have searched for such degeneracies by inducing a small lifting of them through the artificial use of a very weak coulomb repulsion specified by @xmath92 .
for example , in the circular case [ see fig .
[ enspnint ] ] , we found that the state @xmath93 is doubly degenerate , while the state @xmath94 is degenerate with two other @xmath95 states .
these additional states move higher in energy as the strength of the coulomb interaction increases .
we further found that the lifting of degeneracies is sufficiently strong for larger coulomb repulsions with @xmath96 that all the curves in figs .
[ enspcir ] , [ ensp0.7 ] , and [ ensp0.5 ] below are simple ( i.e. , the additional states present in the non - interacting case have been pushed much higher and fall outside the energy window shown ) .
turning on the interaction introduces correlation effects that lead to important modifications of the non - interacting spectra shown in figs.[enspnint ] and [ enspninteta ] . fig .
[ enspcir ] displays the corresponding spectra for the same circular quantum dot as in fig .
[ enspnint ] , but in the presence of a coulomb repulsion with @xmath80 ( gaas ) .
of course , a first effect is the increase in the total energy , but the main difference from the non - interacting case in fig .
[ enspnint ] is the presence of crossings between levels within the same landau band . as a result ,
within the plotted range of magnetic fields , the ground - state total - spin quantum number remains @xmath90 at the first ground - state crossing ( at point @xmath9 ) , and then it changes to @xmath97 ( at the second ground - state crossing at point @xmath98 ) . at the same time , the total angular momentum changes from @xmath91 , to @xmath99 , and then to @xmath100 , respectively . as long as the effective land coefficient @xmath56 , which is the case for the results presented in this section , this threefold alternation in the spin and angular momentum quantum numbers repeats itself ad - infinitum .
we note that experimental observation of this threefold alternation may be forthcoming , since quantum dots with a vanishing land coefficient have been recently fabricated@xcite and were used already to measure two - electron excitation spectra .
the crossings of the curves associated with the three different pairs of quantum numbers @xmath101 , @xmath93 , and @xmath94 form a small triangle ( labeled as abc ) , which is located about @xmath102 t. anticipating the results for non - circular dots below , we note that this triangle tends to collapse to a single point with increasing anisotropy . , with @xmath51 as a function of the magnetic field for @xmath0 interacting electrons in an elliptic quantum dot with intermediate anisotropy ( anisotropy parameter @xmath21 ) .
parameters : external confinement @xmath75 mev , @xmath103 mev ; dielectric constant @xmath80 ; effective mass @xmath77 ; effective land coefficient @xmath56 .
the labels @xmath57 denote the quantum numbers for the total spin and the total angular momentum in the corresponding circular quantum dot .
the @xmath59 indices are not indicated , since the multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 .
note the shrinking of the abc triangle compared to the @xmath20 case shown in fig .
[ enspcir ] .
, width=283 ] another prominent difference between the spectra of non - interacting ( fig .
[ enspnint ] ) and interacting ( fig .
[ enspcir ] ) electrons pertains to the degeneracies at @xmath89 between the @xmath97 and @xmath90 states that are lifted in the interacting - electrons case ; compare in particular the curves with quantum numbers @xmath93 and @xmath104 with the @xmath105 one .
in contrast , the original degeneracies at @xmath89 of the @xmath90 states are unaffected by the interelectron interaction ; compare the curves @xmath106 and @xmath107 , as well as the ones labeled @xmath108 and @xmath109 .
however , these @xmath90 degenaracies at @xmath89 are lifted as a result of an increasing anisotropy of the quantum dot , as seen in fig .
[ ensp0.7 ] . , with @xmath51 as a function of the magnetic field for @xmath0 electrons in an elliptic quantum dot with strong anisotropy ( anisotropy parameter @xmath22 ) .
parameters : external confinement @xmath110 mev , @xmath111 mev ; dielectric constant @xmath80 ; effective mass @xmath55 ; effective land coefficient @xmath56 .
the single labels denote the quantum numbers for the total spin .
the @xmath59 indices are not indicated , since the multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 .
note the collapse of the triangle abc compared to the cases with @xmath20 ( fig .
[ enspcir ] ) and @xmath21 ( fig .
[ ensp0.7 ] ) , and the appearance of a triple - point crossing . , width=283 ] next , we explore the effect of increasing the anisotropy of the quantum dot . in particular , keeping the same strength for the coulomb interaction @xmath112 , we present two representative anisotropy cases , i.e. , @xmath21 ( intermediate anisotropy , see fig .
[ ensp0.7 ] ) , and @xmath22 ( strong anisotropy closer to a quasilinear case , see fig .
[ ensp0.5 ] ) . inspection of the results for the case of intermediate anisotropy ( fig .
[ ensp0.7 ] ) , reveals that compared to fig .
[ enspcir ] the spectra are distorted , but they maintain the overall topology of the circular dot . as a result , we have been able to use the same pairs of labels in naming the different curves , even though the second label does not have the meaning of an angular momentum ( the total angular momentum is not conserved for @xmath113 ) .
there are two main differences from the circular case : i ) the degeneracies at @xmath89 between the @xmath90 states are lifted , and ii ) there is a marked rounding of all the @xmath90 curves in the beginning , so that they do not intersect the vertical energy axis at sharp angles as is the case with fig .
[ enspcir ] .
this initial rounding and bending of the energy curves due to the anisotropy has been experimentally observed@xcite in two - electron quantum dots . in the case of strong anisotropy ( fig .
[ ensp0.5 ] ) , the spectra have evolved to such an extent that only little relation to the circular case can be traced , and as a result we use a single label signifying the total spin to distinguish them .
an important feature that emerges is that the three curves with lowest energies ( two @xmath90 and one @xmath97 curve ) form a band that is well separated from the other excited states .
the existence of such an isolated lowest - energy band is important for validating simple two - qubit and three - qubit models introduced in quantum computation and quantum information theory.@xcite another remarkable feature of the strong - anisotropy case is the appearance of a non - trivial triple - point crossing lying on the ground - state curve ( see arrow in fig . [ ensp0.5 ] ) , which is created from the collapse of the abc triangle between the two @xmath90 and the one @xmath97 lowest - in - energy curves ( compare figs . [ enspcir ] and [ ensp0.7 ] ) .
this low - energy non - trivial triple point [ forming within the lowest landau band ( @xmath114 ) ] is due to the effect of the coulomb interaction , and it is to be contrasted to other trivial triple - point crossings at much higher energies arising from the intersection of the lowest landau band with the @xmath115 and @xmath116 higher landau bands , and which are present even in the non - interacting limit [ see , e.g. , the triple crossing at ( 2.0 t ; 14.2 mev ) in fig . [ enspnint ] ] .
it would be of interest to analyze whether the recently observed@xcite triple - point crossings in deformed quantum dots are non - trivial or trivial in the sense described above . before leaving this section
, we note that the spin multiplets @xmath60 and @xmath61 are degenerate in energy when @xmath56 , which was the case for the energy spectra presented in figs . 1 - 5 .
at a given magnetic field , this degeneracy is naturally lifted when @xmath117 ; however , the final total energies can be easily calculated by adding the zeeman term @xmath118 to the spectral curves displayed in these figures .
furthermore , for a given pair @xmath119 , the zeeman term does not influence the intrinsic structure of the many - body exd wave function [ i.e. , the expansion over constituent slater determinants , see eq .
( [ wf3e ] ) ] , and thus taking @xmath56 does not effect the results for electron densities , conditional probability distributions , and von neumann entropies presented below .
when the confining potential lacks circular symmetry , charge localization is reflected directly in the single - particle electron densities . indeed , electron localization is visible in figs .
[ den12k125 ] and [ den12k3k1 ] , which display the electron densities for @xmath0 electrons in an anisotropic quantum dot with @xmath22 . fig .
[ den12k125 ] illustrates the evolution of electron localization with increasing magnetic field in the case of a weaker coulomb repulsion @xmath112 .
one sees that already at @xmath89 , the electron density is shaped linearly for all practical purposes . however , the three peaks of the localized electrons are rather weak , which contrasts with the case of @xmath120 t [ fig .
[ den12k125](b ) ] , where the three electron peaks are sharply defined . fig .
[ den12k3k1 ] [ in conjunction with fig .
[ den12k125](a ) ] illustrates the strengthening of electron localization as a function of increasing coulomb repulsion , i.e. , decreasing dielectric constant @xmath5 , from a value of 12.5 [ fig .
[ den12k125](a ) ] to @xmath121 [ fig .
[ den12k3k1](a ) ] and then to @xmath122 [ fig . [ den12k3k1](b ) ] . in this last case [ fig .
[ den12k3k1](b ) ] , the three electrons are almost fully localized , with orbitals that exhibit practically zero mutual overlap . since we keep the average frequency , @xmath123 , approximately constant ( i.e. , @xmath124 mev ) for all anisotropy cases studied in this paper , decreasing the dielectric constant is equivalent to increasing the wigner parameter@xcite @xmath125 . at zero magnetic field
, @xmath125 is widely used as a universal parameter to indicate the strength of correlations , since it provides the relative strength of the coulomb repulsion with respect to the quantum kinetic energy , i.e. , @xmath126 with the characteristic length @xmath127 .
for the numerical values of @xmath125 associated with the cases studied here , see fig .
[ enteta12k ] . in the previous section
, we saw that already the electron densities provide partial information about the formation of a linear wigner molecule within an elliptic quantum dot .
indeed , from the charge distributions in figs .
[ den12k125 ] and [ den12k3k1 ] , one can infer that the electrons are localized in three separate positions @xmath128 , @xmath129 , and @xmath130 .
if the electrons were spinless , this situation could be approximately reproduced by a single slater determinant denoted as @xmath131 . however , to probe the spin distribution of the electrons , the exact - diagonalization charge densities do not suffice ; one needs to consider spin - resolved two - point correlation functions , defined as @xmath132 with the exd many - body wave function given by equation ( [ wf3e ] ) . using a normalization constant @xmath133
we further define a related conditional probability distribution ( cpd ) as @xmath134 having the property @xmath135 .
the spin - resolved cpd gives the spatial probability distribution of finding the remaining electrons with spin projection @xmath136 under the condition that the third electron is located ( fixed ) at @xmath137 with spin projection @xmath138 ; @xmath136 and @xmath138 can be either up @xmath139 ) or down ( @xmath140 ) . before examining such cpds , evaluated for numerically determined exd wave functions , it is instructive to consider on a qualitative level the spin structure of the wave functions that can be formed from three localized spin - orbitals only .
in particular , we focus on the case with a total spin projection @xmath1 , when the most general three - orbital wave function is given by the superposition of three slater determinants , i.e. , by the expression @xmath141 with the normalization @xmath142 . unlike the circles used earlier to indicate spinless electrons , the arrows in eq .
( [ wfspabc ] ) indicate the spin projections of the individual spin - orbitals .
the general states ( [ wfspabc ] ) are a superposition of three slater determinants and have attracted a lot of attention in the mathematical theory of entanglement .
indeed , they represent a prototypical class of three - qubit entangled states known as @xmath2-states.@xcite for general coefficients @xmath143 , @xmath144 , and @xmath145 , the states ( [ wfspabc ] ) are not eigenfunctions of the square of total spin @xmath62 ( while the exact - diagonalization wave functions in eq .
( [ wf3e ] are always good eigenfunctions of @xmath62 ) .
however , the special values of these coefficients that lead to good total - spin quantum numbers can be determined.@xcite in particular , using the notation @xmath146 ( where the index @xmath67 is employed in case of a degeneracy ) , one has @xmath147
( i.e. , @xmath148 ) , @xmath149 ( i.e. , @xmath150 , @xmath151 ) , @xmath152 ( i.e. , @xmath153 , @xmath154 , @xmath155 ) . for completeness , we list the case for three fully spin - polarized localized electrons ( which of course is not a @xmath2-state ) .
@xmath156 the wave functions with projections @xmath157 and @xmath158 are similar to the above , but with inverted single - particle spins . before proceeding further
, we note that the term @xmath2-state is some times reserved for the symmetric form @xmath159 in eq .
( [ wf3e3212 ] ) .
this symmetric @xmath2-state has been experimentally realized in linear ion traps.@xcite as we show below , quantum dots offer the means for generating in addition the less symmetric forms given by eqs .
( [ wf3e12121 ] ) and ( [ wf3e12122 ] ) .
such nonsymmetric three - qubit states are sometime denoted@xcite as @xmath160-states ( with a prime ) . in this paper
, we do not make use of this distinction , and we refer to both symmetric and nonsymmetric forms simply as @xmath2-states . in fig .
[ cpdk1 ] , we present several spin - resolved cpds associated with the exd ground state at @xmath89 and strong anisotropy @xmath22 , which is a @xmath161 state [ see fig .
[ ensp0.5 ] ] . although the exd expansion in eq .
( [ wf3e ] ) consists of a large number of slater determinants built from delocalized harmonic - oscillator orbitals , the cpd patterns in fig .
[ cpdk1 ] reveal an intrinsic structure similar to that of the wave function @xmath162 in eq.([wf3e12121 ] ) , which is made out of only three localized spin - orbitals .
in particular , when one requires that the fixed electron has a down spin and is located at the center of the quantum dot , the spin - up electrons are located on the left and right with equal weights [ fig .
[ cpdk1](a ) ] . keeping the down spin - direction , but moving the fixed electron to the right
, reveals that the spin - up electrons are located on the left and the center with equal weights [ fig .
[ cpdk1](b ) ] . considering a spin - up direction for the fixed electron and placing
it on the right reveals that the remaining spin - up electron is distributed on the left and the center of the quantum dot with unequal weights ; approximately 4 ( left ) to 1 ( center ) following the square of the coefficients in front of the determinants @xmath163 ( @xmath150 ) and @xmath164 ( @xmath165 ) in the wave function @xmath162 [ see eq . ( [ wf3e12121 ] ) ] . similarly , considering a spin - up direction for the fixed electron and placing
it on the right reveals that the spin - down electron is distributed on the left and the center of the quantum dot with unequal weights approximately 1 ( left ) to 4 ( center ) , in agreement with the weights of the slater determinants in eq .
( [ wf3e12121 ] ) . in section [ seceta12s12 ] , we investigated the intrinsic structure of the ground - state three - electron wave functions with total spin @xmath90 and for the case of a strong anisotropy @xmath166 . in this section , we analyze a case of the first - excited exd wave function with total spin @xmath90 and for the same strong anisotropy @xmath22 , again at @xmath89 t and for strong interelectron repulsion @xmath122 .
we denote this state as @xmath167 . in fig .
[ cpdk1exci12](a ) , we display the electron density ( ed ) for this second @xmath90 state , while in figs .
[ cpdk1exci12](b , c , d ) , we display spin - resolved cpds for the same state . from the charge density ,
we conclude that the three electrons form a sharply defined linear wigner molecule . the spin - resolved cpd with a spin - down fixed electron placed on the right [ see fig .
[ cpdk1exci12](b ) ] is similar to that in fig .
[ cpdk1](b ) . however , the two spin - resolved cpds with a _ spin - up _ fixed electron placed on the right [ see figs .
[ cpdk1exci12](c ) and [ cpdk1exci12](d ) ] are quite different from the corresponding cpds in figs .
[ cpdk1](c ) and [ cpdk1](d ) .
in fact , in both cases , only one single hump appears to the left of the fixed electron , located at the center for the remaining spin - up electrons [ fig .
[ cpdk1exci12](c ) ] , or on the left for the remaining spin - down electrons [ fig .
[ cpdk1exci12](d ) ] .
this indicates that the intrinsic structure of the @xmath167 wave function is close to that of @xmath168 in eq .
( [ wf3e12122 ] ) , with @xmath153 and @xmath169 . in sections [ seceta12s12 ] and [ seceta12s12exci ] , we investigated the intrinsic structure of the many - body three - electron wave functions with total spin @xmath90 and for the case of a strong anisotropy @xmath166 . in this section
, we analyze a case of an exd wave function with total spin @xmath97 and for the same strong anisotropy @xmath22 , again at @xmath89 t. in particular , we analyze the intrinsic structure of a @xmath170 wave function that is the second excited state for these parameters . in fig .
[ cpdk1exci32 ] , we display spin - resolved cpds for this @xmath97 excited state . a remarkable feature is that for a fixed electron placed on the right all three cpds , @xmath171 [ fig . [ cpdk1exci32](a ) ] , @xmath172 [ fig . [ cpdk1exci32](b ) ] , and @xmath173 [ fig . [ cpdk1exci32](c ) ] coincide .
this indicates that the intrinsic structure of the @xmath170 wave function is close to that of @xmath159 in eq .
( [ wf3e3212 ] ) , with all three coefficients equal to each other , i.e. , @xmath174 . taking into account the @xmath173 cpd with the fixed electron at the center of the quantum dot ,
it is clear that the geometric arrangement of the three localized electrons is linear .
arrangements that are more complicated than the linear geometry can emerge , however , for a range of different parameters , as is discussed in section [ sec0.7 ] below .
in this section , we analyze a case of an exd wave function with total spin @xmath97 and for the intermediate anisotropy @xmath21 . in particular , we analyze the intrinsic structure of a @xmath170 wave function that is the ground state at a magnetic field @xmath175 t ( see fig . [ ensp0.7 ] ) . in fig .
[ cpdk125eta0.7 ] , we display spin - resolved cpds for this ground state .
a remarkable feature is that for a fixed electron placed on the right all three cpds , @xmath171 [ fig .
[ cpdk125eta0.7](a ) ] , @xmath172 [ fig . [ cpdk125eta0.7](b ) ] , and @xmath173 [ fig . [ cpdk125eta0.7](c ) ] coincide .
this indicates that the intrinsic structure of the @xmath170 wave function is close to that of @xmath159 in eq .
( [ wf3e3212 ] ) , with all three coefficients equal to each other , @xmath174 . however , these cpds , as well as the @xmath173 cpd with the fixed spin - down electron at the center [ fig .
[ cpdk125eta0.7](d ) ] , are broader along the @xmath43-direction compared to the cpds associated with the linear molecular arrangement in fig .
[ cpdk1exci32 ] .
this suggests that , for an intermediate anisotropy ( @xmath21 ) , the intrinsic structure of @xmath170 is more complicated .
indeed , as demonstrated in fig .
[ cpdk125eta0.7down ] where the fixed spin - down electron is successively placed away from the @xmath42-axis at ( 0 , 20 nm ) and at ( 0 , -20 nm ) , the intrinsic structure corresponds to a superposition of two molecular isomers , each one described by a three - orbital wave function @xmath159 , but with the three localized spin - orbitals located on the vertices of two isosceles triangles , each one being a mirror reflection ( relative to the @xmath42-axis ) of the other .
the base of the first isosceles triangle lies at -6 nm [ fig .
[ cpdk125eta0.7down](a ) ] and that of the second one at 6 nm [ fig . [ cpdk125eta0.7down](a ) ] off the @xmath42-axis ( in the @xmath43-direction ) .
the two - triangle configuration discussed for three electrons above may be seen as the embryonic precursor of a quasilinear structure of two intertwined `` zig - zag '' crystalline chains .
double _ zig - zag crystalline chains may also be related to the _ single _ zig - zag wigner - crystal chains discussed recently in relation to spontaneous spin polarization in quantum wires.@xcite it is interesting to inquire of how this two - triangle structure is reflected in the spatial distribution of the electron densities . indeed , in fig .
[ den0.7k125](a ) , we display the electron density associated with the ( 3/2,1/2 ) ground state at @xmath175 t. we note in particular the absence of a third peak at the center of the quantum dot . instead
, two rather small peaks appear at ( 0,20 nm ) and ( 0,-20 nm ) , in agreement with the two - triangle internal structure revealed by the cpd analysis .
we further display in fig .
[ den0.7k125](b ) the corresponding electron density for the ( 1/2,1/2 ) ground state at @xmath89 and for a strong coulomb repulsion @xmath176 at the intermediate anisotropy @xmath21 . as a result of the enhanced electron localization , the electron density exhibits pronounced peaks whose locations form a clearly defined diamond ; this indicates again the presence of a two - triangle internal configuration.@xcite the detailed interlocking of the two triangular configurations is further revealed in the spin - resolved cpds that are displayed in fig .
[ cpdk1eta0.7 ] . from the cpds in figs .
[ cpdk1eta0.7](a ) and [ cpdk1eta0.7](b ) , it can be concluded that one triangle is formed by the points @xmath177 nm , @xmath178 nm , and @xmath179 nm , while the second one ( its mirror ) is formed by the points @xmath180 nm , @xmath181 nm , and @xmath182 nm . the @xmath173 [ fig .
[ cpdk1eta0.7](c ) ] and @xmath171 [ fig . [ cpdk1eta0.7](d ) ] cpds with the fixed electron on the right at ( 43,0 ) nm are similar to those in figs .
[ cpdk1](b ) and [ cpdk1](c ) , respectively , with the difference that the central hump is clearly a double one .
this indicates that each triangular configuration is associated with a wave function of the form @xmath162 given in eq .
( [ wf3e12121 ] ) . naturally , the regime of a linear configuration versus a two - triangle one depends on both the strength of the interaction and the anisotropy .
detailed studies of the phase boundary between these two intrinsic structures are , however , left for a future investigation .
the many - body wave functions for @xmath0 electrons analyzed in the previous sections are highly entangled states , since they can not be reduced to a single slater determinant . for special ranges of the dot parameters
, we showed that they acquire the same internal structure as the prototypical @xmath2-states . in this section , we demonstrate that the degree of entanglement can be further quantified through the use of the von neumann entropy @xmath183 for indistinguishable fermions which ( in analogy to the two - electron case @xcite ) is defined as @xmath184 where @xmath185 is a constant ( see below for choosing its value ) and the single - particle density matrix is given by @xmath186 and is normalized to unity , i.e. , @xmath187 .
the greek indices @xmath188 ( or @xmath189 ) count the spin orbitals @xmath190 that span the single - particle space ( of dimension @xmath191 ; see section [ exd ] ) .
note that , in keeping with previous literature on two electrons,@xcite the logarithms are taken to base two . naturally , for calculating numerically the matrix elements @xmath192 , we use the expansion ( [ wf3e ] ) to get @xmath193 where the following conventions for the indices @xmath194 ( or @xmath195 ) apply : @xmath196 and @xmath197 between slater determinants that enter in eq .
( [ sdij ] ) simply equal @xmath198 or vanish .
the single - particle density @xmath199 in eq .
( [ densmat ] ) is in general non - diagonal .
thus we further perform numerically a diagonalization of @xmath199 , and we use the new diagonal elements @xmath200 to straightforwardly calculate the von neumann entropy in eq .
( [ entvn ] ) . as was discussed in refs .
, the von neumann entropy provides a natural measure of entanglement in the case of interacting indistinguishable fermions . in this case
, the entanglement is related@xcite to quantum correlations that are intrinsic to the many - body wave function , i.e. , @xmath183 quantifies the fact that strongly correlated states comprise a larger number of significant slater determinants compared to weakly correlated ones . accordingly
, one expects that @xmath183 increases when the many - body correlations increase ( i.e. , when @xmath125 and @xmath9 increase ) .
this was the case indeed for the @xmath201 quantum dot,@xcite but we have found that it also holds true for the @xmath0 quantum dot , as can be seen from fig .
[ enteta12k ] and fig .
[ enteta12 ] . in the case when the many - body wave function reduces to a single slater determinant , i.e. , when the expansion coefficients reduce to @xmath202 , all the matrix elements @xmath203 vanish except three diagonal ones ( corresponding to three fully occupied spin orbitals ) which are equal to 1/3 ; then @xmath204 .
since the entanglement due to the pauli exchange principle by itself can not be used as a resource for quantum - information processing,@xcite we take the constant @xmath185 in eq .
( [ entvn ] ) to be @xmath205 and as a result the von neumann entropy for a single slater determinant vanishes in our convention . as a function of the dielectric constant @xmath5 [ or equivalently the wigner parameter @xmath125 ; see eq .
( [ wigpar ] ) ] for @xmath0 electrons in an anisotropic quantum dot with strong anisotropy ( anisotropy parameter @xmath22 ) .
parameters : external confinement @xmath110 mev , @xmath111 mev ; effective mass @xmath55 .
the single labels @xmath206 and @xmath207 denote the quantum numbers for the total spin . the ground state ( gs ) , and the first ( i ) and second ( ii ) excited states are indicated .
the horizontal arrow indicates the direction of increasing correlations . according to our convention , the von neumann entropy for a single determinant vanishes .
although the energy gaps between the three exd states diminish with decreasing @xmath5 ( they are quasidegenerate for @xmath122 ) , the relative energy ordering remains unchanged in the plotted range .
, width=283 ] in fig .
[ enteta12k ] , we plot the von neumann entropy for the three lowest exd states with @xmath1 as a function of @xmath5 ( @xmath125 ) for @xmath0 electrons in an anisotropic quantum dot for a strong anisotropy with anisotropy parameter @xmath22 . it is apparent that the von neumann entropy increases for all three states as @xmath125 increases ( @xmath5 decreases ) and the electrons become more localized . at @xmath208 ( corresponding to weaker correlations ) , the von neumann entropies for the three states are clearly non - vanishing , indicating that these exd states are far from being close to a single slater determinant . on the other hand ,
it is natural to expect that the exd states will reduce to single slater determinants at the non - interacting limit . to check this expectation ,
we have carried out an exd calculation for the same qd parameters described in the caption of fig .
[ enteta12k ] , but with a very large @xmath209 in order to approximately mimick the non - interacting limit . in this latter case
, we found that indeed the ground state [ with @xmath210 is practically a single slater determinant made out from the three spin orbitals @xmath211 , @xmath212 , and @xmath213 [ the lowest - in - energy spatial orbital @xmath214 being doubly occupied ; see section [ exd ] for the meaning of indices @xmath215 and @xmath216 ) .
we also found that the corresponding @xmath217 is practically zero .
however , due to the @xmath22 anisotropy , one has @xmath218 , which gives rise to a high degree of energy degeneracy among excited slater determinants with good total spin .
for example , the slater determinant @xmath219 is degenerate to the determinant @xmath220 .
in this situation , a small @xmath221 interaction is sufficient to produce strong mixing of the degenerate slater determinants , and as a result the corresponding @xmath217 values for excited states were found to be non - vanishing .
these findings are reflected in fig .
[ enteta12k ] where , for @xmath208 ( weakest coulomb repulsion in the plotted range ) , the @xmath217 value for the exd ground state is noticeably lower than the values for the two excited states . as was demonstrated in section [ seceta12 ] , at zero magnetic field and strong coulomb repulsion ( e.g. , @xmath122 ) , the three electrons are well separated and localized , and their exd wave functions are equivalent to the forms given in eq .
( [ wf3e12121 ] ) ( gs ) , eq .
( [ wf3e12122 ] ) ( first excited state , i ) , and eq .
( [ wf3e3212 ] ) ( second excited state , ii ) .
these forms are special cases of the general form in eq .
( [ wfspabc ] ) for which another measure of entanglement , called the tangle and specifying the reduced tripartite entanglement among the three localized spin qubits,@xcite can be applied .
the tangle can be calculated@xcite from the coefficients a , b , and c , and it was found that it vanishes for all cases covered by the general form in eq .
( [ wfspabc ] ) . in this respect ,
the von neumann entropy for three well separated electrons studied here exhibits qualitatively a very different behavior , since the values of @xmath217 at @xmath122 are all different , as seen from fig .
[ enteta12k ] .
in particular , we note that in this case the exd value of @xmath217 for the i state is lower than that of the gs state ; this naturally reflects the fact that the first excited exd state in this limit is effectively composed of only two slater determinants [ see eq .
( [ wf3e12122 ] ) ] compared to the three slater determinants associated [ see eq .
( [ wf3e12121 ] ) ] with the exd ground state . as a function of the magnetic field for @xmath0 electrons in an anisotropic quantum dot with strong anisotropy ( anisotropy parameter @xmath22 ) .
parameters : external confinement @xmath110 mev , @xmath111 mev ; dielectric constant @xmath80 ; effective mass @xmath55 .
the single labels @xmath206 and @xmath207 denote the quantum numbers for the total spin .
the vertical arrows indicate the discontinuous jump in the von neumann entropy of the ground state at @xmath222 t , where the ground - state quantum numbers change character , first from ( @xmath223 ) to ( @xmath224 ) , and then immediately to ( @xmath225 ) . the ground state ( gs ) , and the first ( i ) and second ( ii ) excited states are indicated , both to the left and to the right of the vertical arrows . according to our convention , the von neumann entropy for a single determinant vanishes .
, width=283 ] in fig .
[ enteta12 ] , we plot the von neumann entropy for the three lowest exd states with @xmath1 as a function of the magnetic field for @xmath0 electrons in an anisotropic quantum dot with the same parameters as those for the energy spectra in fig .
[ ensp0.5 ] ( strong anisotropy with anisotropy parameter @xmath22 , as also was the case with fig . [ enteta12k ] ) .
it is apparent that the von neumann entropy increases for all three states as the magnetic field increases and the electrons become more localized .
an interesting feature is the discontinuous jump ( around @xmath222 t ) in the von neumann entropy of the exd ground state .
this jump is illustrated by the vertical arrows and is associated with the triple abc point in fig .
[ ensp0.5 ] .
this discontinuity in the ground - state @xmath217 arises from the sudden change in the intrinsic structure ( in term of constituent slater determinants ) of the ground state , as the latter changes its quantum numbers first from ( @xmath223 ) to ( @xmath224 ) , and then again immediately to ( @xmath225 ) at the triple point .
we have presented extensive exact - diagonalization calculations for @xmath0 electrons in anisotropic quantum dots , and for a broad range of anisotropies and strength of inter - electron repulsion .
we have analyzed the excitation spectra both as a function of the magnetic field and as a function of increasing anisotropy .
a main finding was the appearance of triple - crossing points in the ground - state energy curves for stronger anisotropies .
analysis of the intrinsic structure of the many - body wave functions through spin - resolved conditional probability distributions revealed that for all examined cases ( including those with parameters corresponding to currently fabricated quantum dots ) the electrons localize forming wigner molecules . for certain ranges of dot parameters ( mainly at strong anisotropy ) , the wigner molecules acquire a linear geometry , and the associated wave functions with a spin projection @xmath1 are similar to the socalled @xmath2-states that are a prototype of entangled states .
for other ranges of parameters ( mainly at intermediate anisotropy ) , the wigner molecules exhibit a more complex structure consisting of two mirror isosceles triangles .
this latter structures can be considered as an embryonic unit of a zig - zag wigner crystal in quantum wires .
99 c. ellenberger , t. ihn , c. yannouleas , u. landman , k. ensslin , d. driscoll , and a.c .
gossard , phys .
lett . * 96 * , 126806 ( 2006 ) .
zmbuhl , c.m .
marcus , m.p .
hanson , and a.c .
gossard , phys .
. lett . * 93 * , 256801 ( 2004 ) .
y. nishi , p.a .
maksym , d.g .
austing , t. hatano , l.p .
kouwenhoven , h. aoki , and s. tarucha , phys .
b * 74 * , 033306 ( 2006 ) .
s. tarucha , d.g .
austing , t. honda , r.j .
van der hage , and l.p .
kouwenhoven , phys .
lett . * 77 * , 3613 ( 1996 ) .
m. ciorga , a.s .
sachrajda , p. hawrylak , c. gould , p. zawadzki , s. jullian , y. feng , and z. wasilewski , phys .
b * 61 * r16315 ( 2000 ) .
f. meier , j. levy , and d. loss , phys .
* 90 * , 047901 ( 2003 ) .
r. woodworth , a. mizel , and d.a .
lidar , j. phys . : condens .
matter * 18 * , s721 ( 2006 ) .
gorbachev and a.i .
trubilko , laser phys . lett . * 3 * , 59 ( 2006 ) .
t. vrtesi , phys .
a * 75 * , 042330 ( 2007 ) .
b. rthlisberger , j. lehmann , d.s .
saraga , ph .
traber , and d. loss , arxiv:0705.1710v1 ( 2007 ) .
p. hawrylak , phys .
lett . * 71 * 3347 ( 1993 ) .
s.a . mikhailov and n.a .
savostianova , phys .
b * 66 * , 033307 ( 2002 ) . szafran _
et al . _ [ b. szafran , f.m .
peeters , s. bednarek , and j. adamowski , phys .
b * 69 * , 125344 ( 2004 ) ] have studied the ground states of three fully spin - polarized electrons in anisotropic quantum dots at high magnetic fields .
they used a configuration - interaction scheme with a multicenter basis of floating gaussians centered around different sites . unlike the present work , spin - dependent cases at low magnetic fields were not considered by szafran _
w. dr , g. vidal , and j.i .
cirac , phys .
a * 62 * , 062314 ( 2000 ) . v. coffman , j. kundu , and w.k .
wootters , phys .
a * 61 * , 052306 ( 2000 ) .
roos , m. riebe , h. hffner , w. hnsel , j. benhelm , g.p.t .
lancaster , c. becher , f. schmidt - kaler , and r. blatt , science * 304 * , 1478 ( 2004 ) .
beenakker , c. emary , and m. kindermann , phys . rev .
b * 69 * , 115320 ( 2004 ) .
lehoucq , d.c .
sorensen , and c. yang , _
arpack users guide : solution of large - scale eigenvalue problems with implicitly restarted arnoldi methods _
( siam , philadelphia , 1998 ) .
a. szabo and n.s .
ostlund , _ modern quantum chemistry _
( mcgraw - hill , new york , 1989 ) . when the orbitals @xmath226 are eigenfunctions of a 2d oscillator , analytic expressions for the coulomb matrix elements in eq .
( [ vabgd ] ) have been reported by , e.g. , s.m .
girvin and t. jach [ phys .
b * 28 * , 4506 ( 1983 ) ] for a circular oscillator , and ref . for an anisotropic one . c.g .
darwin , proc .
cambridge philos .
* 27 * , 86 ( 1930 )
. v. fock , z. phys . *
47 * , 446 ( 1928 ) .
kouwenhoven , d.g .
austing , and s. tarucha , rep .
. phys . * 64 * , 701 ( 2001 ) .
a.v . madhav and t. chakraborty , phys .
b * 49 * , 8163 ( 1994 ) .
j. kyriakidis and s.j .
penney , phys .
b * 71 * , 125332 ( 2005 ) .
j. kyriakidis , m. pioro - ladriere , m. ciorga , a.s .
sachrajda , and p. hawrylak , phys .
b * 66 * , 035320 ( 2002 ) .
j. kyriakidis and g. burkard , phys .
b * 75 * , 115324 ( 2007 )
. d.g . austing , g. yu , c. payette , j.a .
gupta , m. korkusinski , and g.c .
aers , phys .
( a ) * 204 * , 508 ( 2007 ) . c. yannouleas and u. landman , phys .
* 82 * , 5325 ( 1999 ) ; ( erratum ) * 85 * , 2220 ( 2000 ) r. pauncz , _ the construction of spin eigenfunctions : an exercise book _ ( kluwer academic / plenum publishers , new york , 2000 ) .
l . cao and m. yang , j. phys .
b * 36 * , 4245 ( 2003 ) .
klironomos , j.s .
meyer , and k.a .
matveev , europhys . lett . * 74 * , 679 ( 2006 )
. g. piacente , i.v .
schweigert , j.j .
betouras , and f.m .
peeters , phys .
b * 69 * , 045324 ( 2004 ) . for the two - triangle configuration at high @xmath9 and for the case of the @xmath227 fully spin - polarized state , see also ref . .
r. paskauskas and l. you , phys .
a * 64 * , 042310 ( 2001 ) .
f. buscemi , p. bordone , and a. bertoni , phys .
a * 73 * , 052312 ( 2006 ) . for exd calculations of the von neumann entropy in a two - electron anisotropic quantum dot , see t. ihn , c. ellenberger , k. ensslin , c. yannouleas , u. landman , d.c .
driscoll , a.c .
gossard , int . j. mod
b * 21 * , 1316 ( 2007 ) . c. yannouleas and u. landman , rep .
( 2007 ) , in press .
k. eckert , j. schliemann , d. bruss , and m. lewenstein , ann . of phys .
( new york ) * 299 * , 88 ( 2002 ) .
ghirardi and l. marinatto , phys .
a * 70 * , 012109 ( 2004 ) . | exact - diagonalization calculations for @xmath0 electrons in anisotropic quantum dots , covering a broad range of confinement anisotropies and strength of inter - electron repulsion , are presented for zero and low magnetic fields .
the excitation spectra are analyzed as a function of the strength of the magnetic field and for increasing quantum - dot anisotropy .
analysis of the intrinsic structure of the many - body wave functions through spin - resolved two - point correlations reveals that the electrons tend to localize forming wigner molecules . for certain ranges of dot parameters ( mainly at strong anisotropy ) , the wigner molecules acquire a linear geometry , and the associated wave functions with a spin projection @xmath1 are similar to the representative class of strongly entangled states referred to as @xmath2-states . for other ranges of parameters ( mainly at intermediate anisotropy ) ,
the wigner molecules exhibit a more complex structure consisting of two mirror isosceles triangles .
this latter structure can be viewed as an embryonic unit of a zig - zag wigner crystal in quantum wires .
the degree of entanglement in three - electron quantum dots can be quantified through the use of the von neumann entropy . |
spin - dependent transport processes in layered structures are described on the basis of the following simple picture . after entering in the k - th layer @xmath0 , the incident current ( which was spin polarized along the axis described by the unite vector @xmath1 in the layer @xmath2 )
first aligns @xcite along the axis @xmath3 . inside the ferromagnetic layer ,
the population of spin up ( @xmath4 ) and spin down ( @xmath5 ) is then not conserved due to spin - flip scattering , and some of the down spins relax to the up direction . in this picture
, the states of the sub - system @xmath0 are described by the variables @xmath6 where @xmath7 is the entropy , @xmath8 is the magnetization of the layer k without current , @xmath9 is the number of conduction electrons with spin parallel to the unit vector @xmath3 and @xmath10 is the time derivative of the magnetization in the layer k without current .
if @xmath11 and @xmath9 are independent variables , the conservation of the magnetic momentum reads @xmath12 where @xmath13 is the bohr magneton and @xmath14 is the land factor . in order to write the conservation laws ,
the spin - flip scattering mechanism is described as a chemical reaction transforming a spin down into a spin up along the axis @xmath3 . in this context , the reaction rate @xmath15 is introduced as the number of chemical events per unit of time @xcite , @xcite .
let @xmath16 and @xmath17 be the current of particles flowing from the layer @xmath18 to the layer @xmath19 due respectively to the electrons with spin in the direction @xmath20 and to the electrons with spin in the direction @xmath21 .
the conservation of the particles is then described by @xmath22 where @xmath23 is the spin flip probability of the alignment process @xcite . in the case of ballistic alignment @xmath24 with @xmath25 the angle between @xmath26 and @xmath27 . introducing the polarized current @xmath28 , and the normal current @xmath29 defined by @xmath30 the eq .
( [ contn1 ] ) can be put into the form : @xmath31 \\
\frac{dn^k_{-}}{dt}\,&=&\,\ i^{k-1 \to k}_{-}- \,i^{k \to k+1}_{-}\,+\,\left[\gamma(k-1;k)i^{k-1 \to k}_{p}\ , + \dot{\psi}^k\right ] \label{contn2 } \end{array}\right.\ ] ] with @xmath32 inserting eq .
( [ contn2 ] ) into eq .
( [ contm0 ] ) the conservation of the magnetization reads : @xmath33\right ) \label{contm}\ ] ] the problem of the spin transfer between the polarized current and the magnetic layer is hence solved if the polarized current @xmath28 and the reaction rates @xmath34 can be describe as functions of the experimentally accessible parameters , the current @xmath29 , the electric field @xmath35 and the kinetic coefficients .
this task is typically performed by the application of the first and second laws of the thermodynamics .
the system @xmath36 is open to heat transfer , to chemical transfer , and to mechanical work due to the magnetization and magnetic fields .
let us define the heat and chemical power by @xmath37 and the mechanical power by @xmath38 .
the first law of the thermodynamics applied to the layer @xmath36 gives @xmath39 where @xmath40 . furthermore , with using the canonical definitions @xmath41 the energy variation is : @xmath42 in the present work we limit our analysis to the isothermal case , @xmath43 .
the entropy variation of the sub - layer is deduced from the two last equations , after introducing the conservation laws : @xmath44 where the total chemical potential is @xmath45 .
we have furthermore defined @xmath46 and @xmath47 .
the chemical affinity of the reaction , defined by @xmath48 has also been introduced .
the entropy being an extensive variable , the total entropy variation of the system is obtained by summation over the layers 1 to @xmath49 where the layer 1 is in contact to the left reservoir @xmath50 and the layer @xmath49 is in contact to the right reservoir @xmath51 .
letting @xmath52 the total entropy variation is : @xmath53^{r^l \to 1 } - [ \ldots]^{\omega \to r^r } \nonumber\\ & & + \ , \sum_{k=1}^{\omega}(\vec{h}^k\,-\,\vec{h}^{ext } ) \ , \vec{m_{0}}^k \,+ \sum_{k=1}^{\omega } \left(- \frac{\partial e^k}{\partial \dot{\vec{m}_{0}^k } } \right ) \ , \frac{d \dot{\vec{m}_{0}^k}}{dt } \nonumber\\ & & + \ , \sum_{k=2}^{\omega}\frac{1}{2 } \left(\tilde{a}^{k-1}-\tilde{a}^k\,+\,2 \gamma(k-1;k ) \ , \tilde{a}^k \right)\,i^{k-1 \to k}_{p } \nonumber\\ & & + \ , \sum_{k=2}^{\omega } \frac{1}{2 } ( \mu_{0}^{k-1 } \,- \ , \mu_{0}^{k})\ , i^{k-1 \to k}_{n } \,+\ , \sum_{k=1}^{\omega } \tilde{a}^{k } \ , \dot{\psi}^k \label{entropytot}\end{aligned}\ ] ] where the two first terms in the right hand side of the equality stand for the heat and chemical transfer from the reservoirs to the system @xmath54 .
the variation of entropy takes the form @xmath55 where @xmath56 are generalized forces and @xmath57 are the conjugated generalized fluxes .
the variation of entropy is composed by an external entropy variation @xmath58 and by an internal entropy variation @xmath59 . by applying the second law of thermodynamics
@xmath60 we are leading to introduce the kinetic coefficients @xmath61 such that @xmath62 . by identification with the expression ( [ entropytot ] ) , the kinetic equations are obtained : @xmath63 = \left[\begin{array}{ccccc } l_{nn } & l_{np } & l_{nc } & l_{nm } & l_{n\dot{m } } \\ l_{pn } & l_{pp } & l_{pc } & l_{pm } & l_{p\dot{m } } \\ l_{cn } & l_{cp } & l_{cc } & l_{cm } & l_{c\dot{m } } \\ l_{mn } & l_{mp } & l_{mc } & l_{mm } & l_{m\dot{m } } \\ l_{\dot{m}n } & l_{\dot{m}p } & l_{\dot{m}c } & l_{\dot{m}m } & l_{\dot{m } \dot{m } } \\ \end{array}\right ] \left[\begin{array}{c } \frac{1}{2 } ( \mu_{0}^{k-1 } \,- \ , \mu_{0}^{k})\\ \frac{1}{2 } \left(\tilde{a}^{k-1}-\tilde{a}^k\,+\,2 \gamma(k-1;k ) \ , \tilde{a}^k \right )
\\ \tilde{a}^{k } \\ \vec{h}^k\,-\,\vec{h}^{ext}\\ -\left ( \frac{\partial e^k}{\partial \dot{\vec{m}_{0}^k } } \right)\\ \end{array}\right ] \label{kinetic}\end{aligned}\ ] ] the indices @xmath64 and @xmath65 stand respectively for the normal and polarized transport processes ( see section iii ) , the indices @xmath66 stands for the spin - flip scattering chemical reaction and the indices @xmath67 and @xmath68 account for the dynamics of the magnetization ( see section iv ) .
the kinetic coefficients are state functions : @xmath69 and the symmetrized matrix is positive : @xmath70 .
furthermore , according to onsager relations , the kinetic coefficients are symmetric or antisymmetric @xmath71 we assume in the following that cross effects between electronic transport and ferromagnetic transport are negligible , so that @xmath72 if @xmath73 and @xmath74 .
note that a polarized current is directly produced by a non - uniform magnetization state , through the coefficient @xmath75 in eq ( [ kinetic ] ) .
the physical meaning of the kinetic coefficients is described in the two following sections .
we focus in this section on the first three equations of ( [ kinetic ] ) which describes the electric transport with spin polarization . after performing the continuum limit
, we have : @xmath76 = \left[\begin{array}{ccc } -l_{nn } & -l_{np } & l_{nc } \\
-l_{pn } & -l_{pp } & l_{pc } \\
-l_{nc } & -l_{pc } & l_{cc } \\ \end{array}\right ] \left[\begin{array}{c } \frac{\partial \mu_{0}}{\partial z}\\ \frac{\partial a}{\partial z } \,-\,2 \tilde{\gamma } a \\ { a } \\ \end{array}\right ] \label{kineticelec}\end{aligned}\ ] ] where @xmath77 and the choice of symmetric coefficients @xmath78 and @xmath79 is motivated by the fact that we neglect in this work the direct effects of the magnetic field on the charge carriers @xcite .
in the framework of the two - channel approximation @xcite , the coupling between the two conduction bands is neglected ( i.e. there is no cross effect @xcite between the currents @xmath80 , and @xmath81 ) : @xmath82 where @xmath83 is the mean conductivity of the two spin channels .
the conductivity asymmetry @xmath84 of the two channels is given by @xmath85 equation eq ( [ kineticelec ] ) leads then to the set of equations : @xmath76 = \frac{\sigma_{0}}{e}\,\left[\begin{array}{ccc } 1 & -\beta & 0 \\ -\beta & 1 & 0 \\ 0 & 0 & \frac{e}{\sigma_{0}}l_{cc } \\ \end{array}\right ] \left[\begin{array}{c } \frac{\partial \mu_{0}}{\partial z}\\ \frac{\partial a}{\partial z } \,-\,2 \tilde{\gamma } a\\ { a } \\ \end{array}\right ] \label{gmrarray}\end{aligned}\ ] ] with @xmath86 and @xmath87 . in the stationary state , @xmath88 , and assuming that @xmath89 , @xmath90 and @xmath91 are approximately independent of z the diffusion equation of the chemical affinity is deduced ( see appendix ) :
@xmath92 where the spin diffusion length @xmath93 is given by @xmath94 the `` domain wall '' diffusion length @xmath95 is given by @xmath96 and the parameter k is given by @xmath97 note that the chemical affinity a is equal to the difference of the chemical potentials of the two conduction bands @xmath98 . if we assume no rotation of the spin polarization axis , @xmath99 ( which implies antiparallel magnetic configuration at the interface ) , then equation eq ( [ diff2 ] ) is the well known diffusion equation describing the so - called `` spin accumulation '' or `` spin depletion '' effect responsible for the giant magnetoresistance @xcite , @xcite , @xcite , @xcite . a straightforward calculation
( see appendix ) leads to the giant magnetoresistance of the interface : @xmath100
in this section we focus on the magnetic transport equation without electric current : @xmath101 . from eq
( [ entropy ] ) the entropy variation reduces to : @xmath102 so that the application of the second law of thermodynamics yields , @xmath103 where the kinetic coefficients @xmath104 are the coefficients of the inverse matrix @xmath105 . note
that in adiabatically closed systems , @xmath106 , @xmath107 and @xmath108 are state functions ( i.e depend only of the state variables ( s , @xmath109 , @xmath110 ) , and not on @xmath111 ) .
since the kinetic coefficients are also state functions , the first equation in ( [ kineticm ] ) shows hence that @xmath112 depends on @xmath111 .
we are then leading to impose @xmath113 in order to satisfy the second equation in ( [ kineticm ] ) , which gives the magnetic kinetic energy @xcite . the coefficient @xmath114 can be identified to the magnetic mass , and the first equation in ( [ kineticm ] ) gives the total magnetic force @xmath115 acting on the system : @xmath116 equation ( [ newton ] ) rewrites : @xmath117 where we have identified the gilbert friction coefficient @xcite , @xcite to @xmath118 .
the theorem of the kinetic momentum gives the equation of the dynamics : @xmath119 where @xmath120 is the gyromagnetic ratio and the magnetic gibbs potential @xcite is defined by @xmath121 .
equation ( [ dynamicsm ] ) is the well known gilbert equation @xcite , @xcite , and can be put into the following landau - lifshitz form . in the case of uniform magnetization
we have @xmath122 , where @xmath123 is the saturation magnetization , eq ( [ dynamicsm ] ) rewrites @xmath124 where @xmath125 is here the gradient operator on the surface of a unite sphere .
the phenomenological parameters h and g are linked to the gyromagnetic ratio @xmath126 and the gilbert damping coefficient @xmath127 by the relations @xcite @xmath128
let us assume an interface composed by an incident current @xmath129 of conductivity asymmetry @xmath130 , polarized in the direction @xmath131 which enters in a ferromagnetic layer ( f ) polarized in the direction @xmath132 with conductivity asymmetry @xmath84 .
the transfer of magnetic moments is describes by the term @xmath133 .
equation ( [ contn2 ] ) rewrites @xmath134 where @xmath135 and the expressions of @xmath136 and @xmath137 are derived in the appendix .
the change of the magnetic moment of the layer due to the polarized current is given by equation ( [ contm ] ) and ( [ ll ] ) : @xmath138 where the first , second and third term in the right hand side are respectively the precession term ( or transverse relaxation ) , the longitudinal relaxation term , and the spin transfer due to spin polarized conduction electrons .
@xmath139 in order to estimate the effect of the injection of spin polarized current , the equation ( [ polgll ] ) is applied to the case of monodomain ferromagnet with applied field oriented at the angle @xmath25 from a single anisotropy axis ( see fig . [ angles ] ) .
if the vector @xmath140 makes an angle @xmath141 from the anisotropy axis , the gibbs energy density can be written in the following form @xcite : @xmath142 where @xmath143 is the reduced applied field defined with the anisotropy field @xmath144 , @xmath145 is the anisotropy constant , s is the section and @xmath146 is the out - of - plane coordinate of the vector @xmath140 . due to the cylindrical geometry , @xmath147 . before injecting the current , the angle @xmath148 is given by the equilibrium condition @xmath149 .
the precessional term can be neglected in ( [ polgll ] ) ( low frequency response and/or high damping limite @xcite ) , and @xmath150 . experiments and samples are described in ref .
@xcite , @xcite , @xcite .
ni nanowires are obtained by the method of electrodeposition in track etched membrane templates
. a micro - contact is realized , and the magnetoresistance of a single nanowire is measured .
the wires are about 70 nm diameter and 6000 nm length and the magnetic energy is dominated by the zeeman energy term and the shape anisotropy ( or magnetostatic term ) , very close to that of an infinite cylinder .
the anisotropy field is calculated to be @xmath151 t. the effect of the spin - polarized current was evidenced experimentally by injecting a strong current of about @xmath152 at a fixed value of the external field @xmath153 smaller than the field @xmath154 where the switching occurs without current .
the magnetization switch occurs at the angle @xmath155 .
the maximum distance @xmath156 where the jump of the magnetization can still be observed corresponds then to the variation of the angle @xmath157 needed to shift the magnetization up to the unstable state . for steady states , inserting @xmath158 , @xmath159 , eq .
( [ polgll ] ) leads to @xmath160 where @xmath161 is the polar vector perpendicular to @xmath140 .
the parameter c is defined by the relation @xmath162 where the activation volume @xmath163 of magnetization @xmath123 was estimated to be @xmath164 , @xmath165 @xcite , and @xmath166 @xcite , @xmath167 @xcite .
we obtain @xmath168 .
all parameters in eq .
( [ result ] ) are known if the magnetization reversal mode , which describes the irreversible jump , is known . in some few theoretical models of magnetization reversal @xcite , the functions @xmath169 and @xmath170 are analytical . in the framework of the present empirical approach ,
the experimental data are fitted by the relation deduced from a curling reversal mode in an infinite cylinder @xcite , @xcite : @xmath171 the single adjustable parameter @xmath172 is defined by the geometrical parameter k @xcite , by the exchange length @xmath173 , @xcite and by the radius of the wire r. the experimental points @xmath169 are fitted in fig .
[ hsw ] .
we obtained @xmath174 ( which corresponds to r of about 60 nm ) .
the relation between the angle of the applied field @xmath25 and the angle of the magnetization @xmath175 is : @xmath176 the curve @xmath156 , evaluated from eq .
( [ result ] ) by numerical resolution with a polarization in the direction of the wire axis @xmath177 is plotted in fig .
[ dhswicurrent ] and fig .
[ dhswi ] , together with the experimental data . a strong discrepancy from the linear curve of @xmath178 at small current pulses can be observed . above a critical current corresponding to about @xmath179
the linear fit gives a parameter @xmath180 , which is in accordance with the rough evaluation of eq .
( [ cparameter ] ) .
this critical current below which the linear regime failed in fig .
[ dhswi ] could be interpreted following ref@xcite and ref@xcite as the current needed in order to excite spin waves or other magnetization inhomogeneities @xcite .
the curve given by eq .
( [ result ] ) can then be plotted without adjustable parameter ( fig .
[ dhswi ] ) .
the divergence at @xmath181 is due to the numerical resolution of eq .
( [ result ] ) ( numerator and denominator tend to zero at @xmath182 angle ) .
@xmath183 @xmath184
a systematic thermokinetic description of a metallic ferromagnetic layer open to electronic spin polarized reservoirs has been performed . at constant temperature , assuming the two current approximation and neglecting direct action of the magnetic field on charge carriers , five coupled transport equations account for the complexity of the system .
the approximation of the explicit uncoupling of the transport processes leads to the known results about gmr and landau - lifshitz - gilbert equations for magnetization dynamics . within this approximation and on the basis of the conservation equation of the magnetic moment
, the description of both polarized current and magnetization dynamics leads to a generalized landau - lifshitz - gilbert equation .
the application of this model to experimental data about current - induced magnetization reversal is performed .
the existence of a critical current indicates that the kinetics of magnetization inhomogeneity also plays an important role .
however , the comparison with experimental data shows that the derived thermokinetic generalized landau - lifshitz - gilbert equation provides a description of the basic mechanism responsible for the effect of polarized - current - induced - magnetization reversal .
this work is directly inspired by the synthesis performed by professor c. gruber on thermokinetics , and i thank him for his corrections and comments .
i thank professor j .- ph . ansermet for permanent support .
special thanks to derek kelly for his participation to the experimental work .
the appendix is structured in three parts . in the first part ( a )
, the equation of for the difference of chemical potentials @xmath185 is derived from equation ( [ gmrarray ] ) in the case of steady states . in part ( b )
the equation is applied to the simplified case of gmr or spin accumulation , where the polarization axis is assumed constant through the interface .
the gmr of the interface is deduced . in part ( c ) , the equation is applied in the framework of the experimental study of polarized - current - induced - magnetization - reversal , where an abrupt charge of the polarization axis occurs at the interface . @xmath76 = \frac{\sigma_{0}}{e}\,\left[\begin{array}{ccc } 1 & -\beta & 0 \\ -\beta & 1 & 0 \\ 0 & 0 & \frac{e}{\sigma_{0}}l_{cc } \\ \end{array}\right ] \left[\begin{array}{c } - \frac{\partial \mu_{0}}{\partial z}\\ - \frac{\partial a}{\partial z}\,+\ , 2 \tilde{\gamma } a \\ { a } \\ \end{array}\right ] \label{gmrarraya}\end{aligned}\ ] ] where @xmath206 , @xmath207 , @xmath66 and @xmath208 are constants .
the electric field e(z ) is defined by @xmath209 so that @xmath210 . under the condition of continuity of the currents of the two spin channels at the interface ( no surface scattering ) : @xmath211 we have @xmath212 .
the spin polarized current on the left side of a single interface ( b=0 ) is deduced : in the case of an interface composed by an incident current of conductivity asymmetry @xmath84 , polarized in the direction @xmath131 , entering in a ferromagnetic layer polarized in the direction @xmath140 with conductivity asymmetry @xmath217 , the equation ( [ a(z ) ] ) and ( [ mu0 ] ) still hold in the left and right hand sides of the interface .
however , the change of the polarization axis at the interface leads to modify the continuity equation of the current of the two spin channels @xmath218 .
the integration constant now reads m. n. baibich , j. m. broto , a. fert , f. nguyen van dau , f. petroff , phys .
lett . * 61 * , 2472,(1988 ) and g. binash , p. grnberg , f. saurenbach , w. zinn , phys .
b * 39 * , 4828 ( 1989 ) l. berger j. appl . phys . * 55 * , 1954 ( 1984 ) l. berger phys .
b * 54 * , 9353 ( 1996 ) and j. appl . phys . *
81 * , 4880 ( 1997 ) c. slonczewski , j. magn .
magn . mat . *
159 * , l1 ( 1996 ) and j. magn .
magn . mat .
* 195 * , l261 ( 1999 ) ya .
b. bazaliy , b. a. jones , shou - chen zhang , phys .
b * 57 * , r3213 ( 1998 ) .
freitas , l. berger , j. appl .
57 ( 1985 ) , 1266 , and c .- y
. hung , l. berger , j. appl .
* 63 * , 4276 ( 1988 ) e. salhi and l. berger , j. appl
. phys . * 76 * , 4787 ( 1994 ) .
m. tsoi , a.g .
m. jansen , j. bass , w .- c .
chiang , m. seck , v. tsoi , p. wyder , phys .
lett . * 80 * , 4281 ( 1998 ) .
j. z. sun , j. magn .
magn . mat . *
202 * , 157 ( 1999 ) e.b .
myers , d.c .
ralph , j.a .
katine , r.n .
louie , r.a .
buhrman , science * 285 * , 867 ( 1999 ) .
j - e wegrowe , d. kelly , y. jaccard , ph .
guittienne , j - ph ansermet , europhys .
* 45 * , 626 ( 1999 ) .
j - e wegrowe , d. kelly , y. jaccard , j - ph ansermet , submitted e.c.g .
stueckelberg , p.b : scheurer , `` thermocintique phnomnologique galilenne '' birkauser verlag , basel and stuttgart , 1974 c. gruber , `` thermodynamics of systems with internal adiabatic constraints : time evolution of the adiabatic piston '' , eur . j. phys . * 20 * 259 ( 1999 ) m. johnson and r. h. silsbee , phys
lett . * 55 * , 1790 ( 1985 ) and m. johnson and r.h .
silsbee phys .
b * 37 * , 4959 ( 1987 ) .
a proper description of this alignment process ( ballistic vs. adiabatic alignment , or the so - called `` domain - wall - scattering '' problem ) is described e.g. in the references @xcite a.c .
smith , j. janak , r. b. adler , `` electronic conduction in solids '' , mcgraw - hill physical and quantum electronics series , 1967 , appendix l. i. prigogine , `` introduction to thermodynamics of irreversible processes '' , j. wiley and sons 1962 .
p. levy and s. zhang , phys .
* 79 * , 5111 ( 1997 ) , j. b. a. n. van hoof , k. m. schep , a. brataas , p. j. kelly , g. e. w. bauer , phys .
b * 59 * , 138 ( 1999 ) , a. brataas , g. tatara , g. e. w. bauer , phys . rev .
b * 60*,3406 ( 1999 ) and a. brataas , g. tatara , g. e. w. bauer , phylos . mag .
b * 78 * , 545 ( 1998 ) . a description of what can be expected for spin polarized hall effect can be found in j.e .
hirsch phys .
lett . * 83 * , 1834 ( 1999 ) .
a. fert , i.a .
cambell , j. phys .
f : metal phys . * 6 * , 849 ( 1996 ) .
this approximation corresponds to the approximation performed from equation ( 9 ) to equation ( 11 ) in reference @xcite .
van son , h. van kampen , p. wyder , phys
* 58 * , 7113 ( 1987 ) .
s. zhang , p. m. levy , a. fert , phys .
b * 45 * , 8689 ( 1992 ) .
t. valet , a. fert , phys .
b * 48 * , 7099 ( 1993 ) . m.a .
wongsam , r. w : chanterell , j. magn .
* 152 * , 234 ( 1996 ) t. l. gilbert , phys
. rev . * 100 * , 1243 ( 1955 ) h. suhl , ieee trans . mag . * 34 * , 1834 ( 1998 ) .
w. f. brown jr .
`` micromagnetics '' , interscience publishers 1963 w. t. coffey , yu .
p. kalmykov , j. t. waldron , `` the langevin equation '' , world scientific series in contemporary chemical physics vol . 11 , 1996 ( page 337 ) a. aharoni , j.appl.phys .
* 10 * , 70s ( 1959 ) and `` introduction to the theory of ferromagnetism '' , clarendon press , oxford 1996 and references therein j .- e .
wegrowe , d. kelly , a. franck , s. e. gilbert , j .- ph .
ansermet , phys .
* 82 * , 3681 ( 1999 ) .
wegrowe , s. e. gilbert , v. scarani , d. kelly , b. doudin , j .- ph .
ansermet , ieee trans .
magn . * 34 * , 903 ( 1998 ) .
l. piraux , s. dubois , c. marchal , j.m .
beuken , l. filipozzi , j.f .
despres , k. ounadjela , a. fert , j. magn .
mat . * 156 * , 317 ( 1996 ) .
w. t. coffey , d.s.f .
crothers , j. l. dormann , yu .
p. kalmykov , e.c .
kennedy , w. wernsdorfer , phys.rev.lett .
* 80 * , 5655 ( 1998 ) .
w. wernsdordfer et al .
rev . lett . * 77 * , 1873 ( 1996 ) , w. wernsdordfer et al .
b * 55 * , 11552 ( 1997 ) .
v. kambersky , canadian journal of physics * 48 * , 2906 ( 1970 ) , and czech .
b * 22 * , 572 ( 1971 ) | in order to describe the recently observed effect of current induced magnetization reversal in magnetic nanostructures , the thermokinetic theory is applied to a metallic ferromagnet in contact with a reservoir of spin polarized conduction electrons .
the spin flip relaxation of the conduction electrons is described thermodynamically as a chemical reaction .
the diffusion equation of the chemical potential ( or the giant magnetoresistance ) and the usual landau - lifshitz - gilbert ( llg ) equation are derived from the entropy variation .
the expression of the conservation laws of the magnetic moments , including spin dependent scattering processes , leads then to the generalized llg equation with spin polarized current .
the equation is applied to the measurements obtained on single magnetic ni nanowires .
an unexpected and spectacular effect due to spin polarization of conduction electrons in metallic ferromagnets , the giant magnetoresistance ( gmr ) , appeared with the first transport studies on magnetic nanostructures @xcite .
the spin diffusion length of conduction electrons being of some few tens of nm , the relaxation of the conduction electron spins becomes observable when the magnetization can be controlled over this typical length .
some predictions about the inverse effect , namely the influence of spin polarized current on the dynamics of the magnetization , were also proposed .
berger predicted the existence of some surprising phenomena due to the action of spin polarized conduction electrons on domain walls @xcite or spin waves @xcite in magnetic thin films .
slonczewski predicted the rotation of the magnetization due to polarized current in multilayered systems @xcite , and bazaliy et al .
derived from microscopic considerations a generalized landau - lifshitz - gilbert equation @xcite . all the above mentioned approaches
are microscopic and based on the ballistic approximation . from an experimental point of view ,
freitas and berger , hung and berger @xcite , and salhi and berger @xcite show the action of a high current density on domain walls in thin films .
recent experiments on nanostructured samples bring new evidence for the interpretation in terms of the action of the spin of the conduction electrons .
tsoi et al .
@xcite show an effect of a high current density on spin waves generation in co / cu multilayers , sun reported on current - driven magnetic switching in manganite @xcite , and myers et al . reported an effect of current induced switching in magnetic multilayer device @xcite . in a recent work , we have evidenced an effect of current - induced magnetization reversal in magnetic nanowires , @xcite , @xcite where the reversal of the magnetization is induced by a high current at an applied field 20% smaller than the normal reversal field .
the ballistic approximation is however difficult to justify in all these experiments .
a phenomenological approach based on the thermokinetic theory @xcite , @xcite of a metallic ferromagnet in contact with a reservoir of spin polarized conduction electrons is proposed .
in contrast to the pioneering works of johnson and silsbee @xcite on spin polarized current , the spin - flip scattering is introduced here as a chemical reaction .
this formulation allows the conservation laws of the spin polarized conduction electrons and the ferromagnetic order parameter to be treated ( section i ) .
this phenomenological approach models the different effects able to take place in the system by a set of coupled transport equations ( section ii ) .
the differential equation of the chemical potential is derived leading to the known formula of the gmr ( section iii ) . for a closed ferromagnetic system without electric current ,
the common landau - lifshitz - gilbert equation is derived ( section iv ) .
the open system with both ferromagnetization and spin polarized conduction electrons leads to the description of the generalized landau - lifshitz - gilbert equation with polarized current ( section v ) .
the consequences in terms of current - induced magnetization reversal are deduced .
the model is applied on the data obtained on ni nanowires . |
in 1974 , c. n. yang@xcite pointed out that a fundamental gauge force is completely and intrinsically characterized by a non - integrable phase factor that an elementary particle of charge @xmath0 picks up when it moves from a to b points : @xmath1 here @xmath2 is the gauge potential and @xmath3 stands for path ordering . today
it has been fully established that the fundamental forces of weak , electromagnetic , and strong interactions between the elementary particles are all dictated by ( [ nonit ] ) as gauge forces , which is a great triumph of physics in the twentieth century .
the world of condensed matter physics is within the realm of a particular fundamental force , namely , the electromagnetic force . in the presence of millions of other electrons and ionic atoms ,
an electron moving in a solid is no longer a free ` bare ' electron in the vacuum .
nevertheless , the electron still remains ` free ' as a bloch electron in a periodic array of the ionic atoms as if the latter is ` transparent ' , only with its energy spectrum being altered into the so - called energy bands .
if the lattice vibration and the screened coulomb repulsion from other electrons can be treated perturbatively , the bloch electron world in the condensed matter , known as the landau s fermi liquid state , is well described by the quantum many - body theory which was a great achievement in 50.@xcite but nature is more complex and richer . in some real materials ,
the band mass of a bloch electron can become much heavier or its bandwidth is much reduced as compared to the strength of the local coulomb interaction between the electrons .
this is precisely the proposal made by p. w. anderson@xcite in 1987 in explaining high-@xmath4 superconductivity discovered in the cuprate compounds .
the proposed model for the cuprates is extremely simple .
that is , the single - band bloch electrons in the copper - oxide layers of the cuprate will experience a strong on - site coulomb repulsion @xmath5 as compared to a relatively narrow bandwidth .
this hubbard - type model is a great simplification of a more realistic description of the copper - oxides , justified at reasonable energy and length scales from quantum chemistry .
an entirely new physics is conjectured@xcite to arise in the infrared regime .
but the problem is that the well - established perturbation - based quantum many - body theory@xcite is expected to completely fail in this strong correlation limit .
the challenge to theorists is how to identify a right mechanism , and at the same time , to find an appropriate mathematical description for such a perturbation inoperative regime , in order to accommodate a great amount of anomalous properties observed in cuprate superconductors.@xcite the story that we will present below is that for such an innocent - looking large-@xmath5 hubbard model , its infrared novelty can be fully captured by an exotic emergent gauge force in the same fashion that the elementary particles experience the fundamental forces in nature .
that is , novel physical properties observed in experiments may be essentially attributed to some relevant low - energy collective degrees of freedom with an intrinsic gauge structure . in this sense ,
the emergent gauge force in the cuprates is for real , which is precise and robust when the length scale is larger than the lattice constant of the copper - oxide unit cells and the energy is much less than a characteristic scale caused by the hubbard @xmath5 for two electrons staying at the same site ( i.e. , the so - called mott gap @xmath6@xmath7@xcite ) . here
it is crucial to distinguish such an intrinsic gauge force from a fictitious gauge degree of freedom arising from a formal decomposition of an electron operator mathematically , e.g. , in the so - called slave - particle formulations.@xcite the latter represents a counter - effect against breaking up the electron degrees of freedom and disappears if one stays with the electronic description . in other words , the spinless holon of charge @xmath8 and neutral spinon of spin-@xmath9 that appear in the fractionalizationof an electron carry opposite gauge charges such that the electron as a composite object is still gauge - neutral . in particular , these latter gauge fluctuations will be suppressed or ` higgsed ' when the holons are condensed and/or spinons are in singlet pairing@xcite [ i.e. , the so - called resonating - valence - bond ( rvb ) state@xcite ] .
by contrast , the intrinsic gauge force discussed in the following will exist generally as a precise property@xcite hidden in the @xmath10-@xmath11 model ( the large-@xmath5 hubbard model ) , which will dictate how the physical degrees of freedom are fractionalized mathematically .
in particular , the electrons can always feel the effect of such a gauge force , even if the subsystems ( partons ) of fractionalization have experienced off - diagonal - long - range - orders ( odlros ) , like holon condensation and/or spinon rvb pairing . as such
, it is fundamentally different from the aforementioned gauge fluctuations@xcite associated with the slave - particle fractionalization , which can be effectively eliminated by the odlros of the partons .
this novel gauge force plays a critical role@xcite in driving superconducting phase transition of the electronic system .
consequently the system exhibits non - bcs superconducting and non - fermi - liquid normal - state behavior because of such emergent gauge fields.@xcite
) , is equivalent to an emergent gauge force .
it is picked up by a hole moving from a to b along a given path @xmath12 , and its novel quantum interference effect will be responsible for the essential physics of a doped mott insulator . , scaledwidth=60.0% ] let us start by outlining an exact theorem for the large-@xmath5 hubbard model , i.e. , the @xmath10-@xmath11 model on a bipartite lattice , which holds true for arbitrary dimensions , electron concentration , and temperature.@xcite it states that when a charge carrier moves from a to b , it will always pick up a phase factor as a product of signs : @xmath13 as illustrated in fig . 1
, here @xmath14 is path - dependent , and thus is a typical _ non - integrable _ phase factor ; each @xmath15 sign faithfully records the doublet ( @xmath16 or @xmath17 ) spin backflow at each step of hopping to the nearest - neighbor site ( a neutral spin must be simultaneously exchanged positions with the charge according to the @xmath10-@xmath11 model ) ; such a non - integrable phase factor is known as the phase string effect,@xcite which appears in the single - particle propagator , total energy expression , and the partition function of the system .
for example , the partition function of the @xmath10-@xmath11 model can be generally expressed by@xcite
@xmath18 \label{z}\ ] ] where @xmath19 represents the closed loops of the whole charge carriers and spins ; the positive weight @xmath20>0 $ ] depends on all the detailed coupling constants as well as the dimensionality , temperature , and electron concentration ; finally an additional sign factor @xmath21 represents conventional fermionic statistical signs resulted from exchanges between the doped charge carriers.@xcite near half - filling , such statistical signs are much sparser than the phase string signs ( [ ps ] ) picked up by the charge carriers over the set of closed loops @xmath19 .
the important implication is as follows .
the phase string factor ( [ ps ] ) as a non - integrable phase factor proves that the long - wavelength physics of the large-@xmath5 hubbard model or the @xmath10-@xmath11 model is completely and intrinsically dictated by a gauge force , just like the general definition ( [ nonit ] ) for the fundamental forces of nature .
such a gauge force is emergent by nature because it disappears when the hubbard @xmath5 is reduced to much smaller than the bandwidth , where the perturbative quantum many - body theory works and the system is well described by the landau s fermi liquid theory . in this weak coupling limit , the solid state is conventional with the electrons behaving as normal fermions , and may be continuously connected to the fermi gas limit of the bloch electrons .
but the strong coupling limit , as characterized by the opening up of a mott gap , signals@xcite the failure of the quantum many - body theory and the landau s paradigm . here a new paradigm in the condensed matter
emerges.@xcite it is a universe in which individual electrons themselves are no longer elementary but ` emergent ' , while the basic constituent degrees of freedom are collective in nature , including both charge and spin , and the fundamental force becomes a long - range gauge force.@xcite the nature of such an emergent gauge force can be understood by noting that the non - integrable sign factor ( [ ps ] ) leads to exotic quantum interference effect for a charge particle traversing a quantum spin background from a to b. namely , the spin degrees of freedom will strongly influence the charge degree of freedom nonlocally via ( [ ps ] ) .
a very illustrative example will be given below .
_ vice versa _ , the spin degrees of freedom will be strongly influenced by ( [ ps ] ) as well in the presence of a finite density of the charge carriers in order to optimize the _ total _ energy of the system .
in other words , the gauge force as introduced by ( [ ps ] ) will mediate a long - range _ mutual _ entanglement between the charge and spin degrees of freedom , leading to all the novelty in the different concentration , temperature , etc .
, regimes of a complex phase diagram . such an emergent gauge structure is described mathematically by a mutual chern - simons gauge fields.@xcite the non - integrable phase factor appearing in ( [ z ] ) indicates that the fermion signs of the electrons are completely altered by the strong interaction .
in fact , the fermion signs totally disappear at half - filling limit , with @xmath22=@xmath23 , which is the simplest but most striking example of how the strong interaction can completely change the electron statistics .
sparse signs pop up with the introduction of doped holes via @xmath22 and @xmath24 , which are much reduced at low doping as compared to the original fermionic signs associated with the underlying electrons
. the simplest nontrivial case is doping one hole into the half - filling mott insulator . here
one has @xmath23 and the phase string factor @xmath22 plays singularly the most important role .
generally , one expects that when the spatial distance between the given a and b points in fig .
1 is sufficiently long , the non - integrable phase factor @xmath25 in ( [ ps ] ) can result in a severe destructive quantum interference from different paths with @xmath25 differing by @xmath26 . as an analogy to anderson localization,@xcite the single hole doped in such a mott insulator will be self - localized@xcite ( see below ) .
this is probably one of the most striking consequences of the non - integrable phase factor identified for the @xmath10-@xmath11 model : in sharp contrast to anderson localization , the present system is translation invariant without the presence of disorder and the entire novelty comes from the emergent gauge force as the strong correlation effect .
summing up all the paths with singular @xmath25 at long distance is a difficult analytic task .
however , to reveal the novel quantum interference effect of the non - integrable phase factor ( [ ps ] ) , one may consider a ladder system with one direction of the sample very long , but finite along the other direction , instead of a large two - dimensional square lattice .
as a matter of fact , even for a two - leg ladder , i.e. , two one - dimensional chains coupled together , there are more than one path to connect any a and b lattice sites , and consequently , @xmath25 becomes non - integrable phase factor to play the crucial role of quantum interference so long as the ladder is long enough along the chain direction .
fortunately for the ladder systems , the density matrix renormalization group ( dmrg ) method provides an accurate and powerful numerical approach to this problem.@xcite a systematic dmrg study on the one - hole problem in large - scale @xmath10-@xmath11 ladder systems , with the leg number ranging from 1 to 5 , has been carried out recently@xcite .
the results have unequivocally demonstrated that the single hole is indeed self - localized once the leg number is larger than one . by contrast , once the non - integrable phase factor @xmath25 is turned off by slightly modifying@xcite the @xmath10-@xmath11 hamiltonian , the self - localization disappears and the doped hole recovers a well - defined bloch quasiparticle behavior .
so far no approximation has been made about the model and one has already known deeply about its novel nature by identifying the non - integrable phase factor ( [ ps ] ) .
in essence , the non - integrable phase factor of @xmath22 and @xmath24 in ( [ z ] ) is topological rather than geometrical , in the sense that it is completely specified by the numbers of hole - spin exchanges , @xmath27 , and hole - hole exchanges , @xmath28 , for a given set of closed paths @xmath19
. therefore , such a sign structure resembles the _ altered _ statistical signs . in other words , the familiar fermion signs in a weakly interacting electron system , which is responsible for the fermi liquid behavior , is now replaced by new statistical signs encoded by @xmath22 and @xmath24 .
one may introduce a unitary transformation @xmath29 to reexpress the ground state of the @xmath10-@xmath11 model by @xmath30 here the so - called mutual duality transformation @xmath29 will properly incorporate the precise phase string ( [ ps ] ) , which is statatical in nature as discussed above . on the other hand , @xmath24 in ( [ hh ] ) will be incorporated into @xmath31 by fermionic statistical signs ( see below ) . as the ground state after the duality transformation , @xmath31 is presumably much smoother such that a perturbative treatment becomes possible .
note that the phase string @xmath32 in ( [ ps ] ) is closed - path - dependent as well as hole - spin - exchange - dependent .
it thus resembles the mutual statistics .
one may require @xmath29 to keep track of the nonlocal phase shift in ( [ ps ] ) as a generalized berry s phase : @xmath33 where @xmath34 represents the mult - coordinates of the doped holes transversing the set of closed paths @xmath19 .
such a @xmath29 in the coordinate space can be written as @xmath35 where @xmath36 is the complex coordinate with the subscripts @xmath37 and @xmath38 denoting the hole and down spin sites , respectively . in the second - quantization formalism
, one has @xmath39 with @xmath40 here , @xmath41 in @xmath42 defines the number operator of the hole created at site @xmath43 by the electron annihilation operator @xmath44 .
each doped hole introduces a nonlocal phase shift @xmath45 , which is defined by @xmath46 in which @xmath47 @xmath48 and @xmath49 denotes the background spin number at site @xmath50 . in this new world of mottness , at the electron density close to the so - called half - filling with each lattice site occupied by one electron , the relevant degrees of freedom are collective ones .
the charge carriers are at those sites where the electron numbers deviate from the half - filling ; and the neutral spins are at half - filled sites .
the doubly occupied sites are in a high - energy sector which only exist in virtual processes to mediate the so - called antiferromagnetic superexchange coupling between the nearest neighboring neutral spins in the large-@xmath5 limit . generally speaking
, the original single electrons do not directly exist as they are strongly coupled to each other as the hilbert space is reduced by the no double occupancy constraint .
the mathematical question is how those low - energy degrees of freedom behave quantum mechanically ; and the physical question is how the system responds to an external experimental probe .
the goal is to search for a correct class of ground state wave functions when the conventional perturbative approach is totally invalidated .
with the singular sign structure accurately embedded / regulated in @xmath29 in the ground state @xmath51 , a mean - field - type treatment of the non - singular @xmath52 is possible . at finite doping ,
it may be generally constructed in the following form @xcite @xmath53at half - filling , with @xmath54 , @xmath52 recover the so - called bosonic rvb state @xmath55 , which was first proposed by liang , doucot , anderson@xcite to accurately describe the antiferromagnetic state of the heisenberg model .
note that at finite doping , the no double occupancy constraint in ( [ frac1 ] ) is naturally implemented because holes are created in pairs via @xmath56 on the single - occupancy `` vacuum '' state @xmath55 .
the hole pairing amplidue @xmath57 is taken as a smooth variational parameter . at finite doping , the single - occupancy `` vacuum '' state @xmath55 will be motified self - consistently by doping .
in general , it may be written as @xmath58 with @xmath59 enforcing the single - occupancy constraint @xmath60 on the bosonic rvb state @xmath61 with @xmath62 . here
the rvb pairing amplitude @xmath63 is another variational parameter , which will decides the rvb pairing size depending on the doping concentration .
by explicitly incorporating the non - integrable phase factor or the sign structure identified for the @xmath10-@xmath11 model , a new class of variational wave functions has been constructed in ( [ unitary ] ) . in the following , we discuss the general features and consequences of such new type of ground states , which is clearly distinct from a conventional bcs wave function for superconductivity and a fermi liquid state for normal state .
one may straightforwardly recast the ground state ansatz ( [ unitary ] ) into a direct product state : @xmath64by decomposing the electron annihilation operator @xmath65 as follows@xcite @xmath66 the ground state ( [ gsansatz ] ) is composed of three ` fractionalized ' subsystems .
here @xmath67where the bosonic wavefunction @xmath68 defines a bose - condensed holon state with a _ _ bosonic__creation operator @xmath69 acting on a vacuum @xmath70 ; and @xmath71which describes a backflow spinon state with the ` bcs'-like pairing amplitude @xmath72 , where @xmath73 denotes a _ fermionic _ creation operator acting on a vacuum @xmath74 ; note that the bosonic rvb state @xmath75 is already given in eq .
( [ phirvb ] ) .
finally , both ( [ gsansatz ] ) and ( [ decomp ] ) involve a projection operator , which is defined by @xmath76 in which @xmath77 will further enforce @xmath78 , such that each @xmath79-spinon always coincides with a holon as @xmath80 ( here @xmath81 and @xmath82 with @xmath83 . by applying @xmath84 ,
the physical hilbert space is restored in ( [ gsansatz ] ) .
so we see that the original ground state ansatz ( [ unitary ] ) is actually simplified as a direct product state of three subsystems in ( [ gsansatz ] ) . here
we do not see the trace of electrons directly anymore . instead , they are all fractionalized into three types of pastons , the holons created by @xmath85 , the backflow fermionic spinons created by @xmath86 , and bosonic spinons created by @xmath87 . in particular , all of these partons are in the odlro states of their own in ( [ bgs ] ) , ( [ phia-0 ] ) and ( [ phirvb ] ) , respectively , where the holons are bose condensed , @xmath79-spinons and @xmath88-spinons are in rvb paired states .
it is well known in condensed matter physics that when a many - body quantum system exhibits an odlro by spontaneously breaking a global symmetry , it will become ` rigid ' against fluctuations attempting to destroy the odlro to restore the symmetry .
for example , a crystal with a translational symmetry breaking naturally exhibits the rigidity in a conventional sense ; a superconductivty odlro leads to the expulsion of external magnetic fields , etc .
in other words , with the three parton subsystems exhibiting their own odlros , the conventional gauge fluctuations associated with the fractionalization ( which would otherwise play a crital role to confine the partons should the decomposition is unphysical ) will get suppressed , which in turn justifies the present fractionalization scheme given in ( [ decomp ] ) .
it is important to note that the emergent gauge structure due to the non - integrable phase ( sign ) factor ( [ ps ] ) is an intrinsic one , which can not be ` higgsed ' by the odlros in the parton subsystems mentioned above .
such an emergent gauge structure should be distinguished from the usual gauge fluctuations due to fractionalization , which are ` higgsed ' by the odrlos to self - consistently support the fractionalization in ( [ decomp ] ) as already pointed out in the introduction .
therefore , these partons are not simply described by some mean - field equations at finite doping . as a matter of fact
, the fractionalization is precisely dictated by the emergent sign structure as discussed above .
the quantum interference of these non - integrable phases from different paths will be characterized by a pair of emergent gauge fields , known as the mutual chern - simons gauge fields as discussed in sec .
such a topological gauge force between the partons can not be gauged away by the unitary transformation @xmath29 and will play important roles in shaping the transformed ground state @xmath89 in ( [ unitary ] ) . in general , in the parton representation ( [ gsansatz ] )
, the emergent mutual chern - simons gauge fields will couple to the holons and spinons such that even though the ground state is approximately expressed in a direct product form , the mutual influences have been factored into each subsystem@xcite . in other words ,
the gauge force is essential in determining the ground states of the subsystems and their doping depdendences . in the following , we briefly outline some key features , while one is referred to refs . for more details .
+ _ antiferromagnetic state , superconducting state , and the pseudogap physics _ + at half - filling ,
all the electrons are reduced to neutral spins in the low - energy sector below the mott gap _ without _ the play of the fermionic statistics .
the ground state ( [ unitary ] ) or ( [ gsansatz ] ) is naturally turned into an antiferromagnetic state with a true magnetic odlro in @xmath90 .
@xmath90 in this limit describes the ground state of the heisenberg model very accurately as a variational state @xcite , which serves as the reference state for us to understand the finite - doped mott insulator . at finite doping , the charge degree of freedom
is introduced by doped holes in a background of neutral spins , and thus the particular fractionalization in sec .
iva emerges due to the phase string like sign structure ( [ ps ] ) .
a superconducting state arises once the localized doped holes in the dilute limit ( cf .
ii ) undergo a delocalization transition beyond some critical doping .
the superconducting state of the doped mott insulator is a natural ground state of pure electrons , without needing an extra ` gluon ' like phonon in a bcs superconductor . in the latter ,
a fermi liquid state is a natural ground state for purely electronic degrees of freedom , which sets in as a ` normal state ' once the cooper pairing mediated by phonons is turned off . in this sense ,
the superconducting ground state ( [ gsansatz ] ) is a stable infrared fixed point state , which also controls all the anomalous pseudogap behavior at finite temperature . in other words , the basic correlations exhibited in high - temperature ` normal state ' regimes are already encoded in the ground state , in a form of electron fractionalization . as shown in ( [ gsansatz ] ) , the fundamental degrees of freedom are fractionalized into holons and two - component spinons in the ground state , which may be regarded as the ` parent state ' .
the so - called lower pseudogap phase ( lpp ) is formally very similar to the superconducting state in that the holons remain condensed , while both the @xmath88- and @xmath79-spinons are still rvb - paired .
but with unpaired @xmath88-spinons thermally excited in @xmath75 at @xmath91 , the superconducting phase coherence in the ground state ( [ gsansatz ] ) is disordered in the lpp , because these thermally excited spinons automatically carry vortices , due to the novel sign structure as can be explicitly seen in ( [ unitary3 ] ) and ( [ phif])@xcite .
namely , the lpp state is a phase - disordered superconducting state , whose anomalous physical properties are thus intimately tied to the non - bcs structure of the superconducting state itself . here
the spin pseudogap behavior is manifested by @xmath75 @xcite .
the bosonic rvb pairing of the @xmath88-spinons accurately depicts the tendency for the local spins to develop antiferromagnetic correlations with decreasing temperature , which explains the high - temperature spin pseudogap behavior over a wide range . on the other hand
, the holon condensation would simply mean that the doped charges ( holons ) gain coherence .
but due to the altered statistical sign structure mentioned above , there generally exists a novel ` mutual entanglement ' between the holon and @xmath88-spinon subsystems , via the mutual chern - simons gauge fields .
consequently , the spin excitations ( the @xmath88-spinons ) strongly affect the charge condensate by creating supercurrent vortices , i.e. , the spinon - vortices . in the lpp , they disorder the superconducting phase coherence , resulting in a large non - drude resistivity and strong nernst effect as the characteristics of non - gaussian - like superconducting fluctuations in the lpp @xcite . self - consistently , the holon condensation in the lpp induces a spin gap @xmath92 in the spinon excitation spectrum and destroys the aflro . at a sufficiently low temperature ,
with the thermally excited spinon - vortices greatly reduced in number due to the spin gap , the confinement of them into vortex - antivortex pairs eventually becomes possible in the lpp , in a fashion of kosterlitz - thouless - type transition , which results in a true superconducting phase coherence below @xmath93 as controled by the spin gap @xmath92 .
another prediction for the ansatz ( [ gsansatz ] ) is the existence of a distinct pseudogap phase known as the lpp - ii state @xcite .
such a state can be realized when the bcs - like pairing of the @xmath79-spinons in @xmath94 of ( [ phia-0 ] ) is destroyed ( say , by strong magnetic fields at low @xmath95 ) _ before _ the occurrence of phase disordering by thermally excited @xmath88-spinon - vortex excitations .
correspondingly , the cooper pairing amplitude vanishes to result in a non - superconducting normal state , at least in the magnetic vortex core region .
note that the @xmath79-spinon in the superconducting phase and the lpp state is charge - neutral as well as gauge - neutral , immune from the mutual chern - simons gauge force between the @xmath88-spinons and holons@xcite .
it is gapped in the lpp such that its contributions to spin susceptibility and specific heat vanish at low temperature .
but in the lpp - ii , the fermi pockets of the fermionic @xmath79-spinons give rise to quantum oscillation , a pauli susceptibility and a linear-@xmath95 specific heat just like in a typical fermi liquid .
we have presented an example of emerging physics in a strongly correlated system of condensed matter . in the low - energy world of emergence
, the underlying physical law has been ` changed ' from a conventional description of bloch electrons with strong local repulsion to a novel one that is described by collective degrees of freedom , i.e. , some ` new ' elementary particles , mutually interacting through an emergent gauge force . for a physicist living in this low - energy world without being able to get access to energies higher than the mott gap , he /
she could not experimentally identify individual electrons as the basic constituents , and thus no hubbard model could be deduced as the underlying physical law .
instead , the greatest triumph that he / she could possibly achieve is to ` discover ' the new elementary particles and a topological gauge structure as dictated by ( [ ps ] ) , which constitute the fundamental and complete physical law of this new universe .
this emergent world of gauge force is protected by the mott gap , and is categorically different from the condensed matter universe that we know of at a ` higher energy ' level . given the exactness of the mapping between the two worlds , it provides a perfect example to illustrate the idea of ` more is different ' in condensed matter physics.@xcite
this work was initiated during the workshop in celebration of c n yang s 90th birthday and the 15th anniversary of institute for advanced study at tsinghua university ( iastu ) .
we would like to thank the colleagues and students from iastu for the earlier collaborations and discussions related to the present paper .
the support from nsfc and most grants is also acknowledged . c. n. yang , phys .
lett . * 33 * , 445 ( 1974 ) ; t. t. wu and c. n. yang , phys .
d * 12 * , 3845 ( 1975 ) .
a. a. abrikosov , l. p. gorkov , and e. dzyaloshinskii , _ method of quantm field theory in statistical physics _ , ( prentice - hall , englewood cliffs , 1963 ) .
d. pines and p. nozieres , _ theory of quantum liquids _ , ( benjamin , new york 1965 ) .
p. w. anderson , science * 235 * , 1196 ( 1987 ) .
p. w. anderson , _ the theory of superconductivity in the high @xmath4 cuprates _ , ( princeton univ .
press , princeton , 1997 ) . for a review , see , p. a. lee , n. nagaosa , and x. g. wen , rev .
phys . * 78 * , 17 ( 2006 ) . c. ye , p. cai , r. z. yu , x. d. zhou , w. ruan , q. q. liu , c. q. jin , and y. y. wang , nat
. commun . 4 , 1365 ( 2013 ) so far , for simplicity
, the whole discussion has been focused on the large-@xmath5 limit .
but i wish to point out that essentially the same physics will persist over to a finite-@xmath5 so long as the splitting of the bloch band into the lower and upper hubbard bands by @xmath5 holds true , which will be presented elsewhere.@xcite | we illustrate how a completely new world of gauge force emerges from a conventional condensed matter system in a rigorous way .
a characteristic energy scale ( mott gap ) separates such an exotic universe from the ordinary one that we condensed matter physicists are more familiar with at higher energies .
the governing physical law is no longer about individual electrons but concerns fractionalized particles , i.e. , partons , as the new collective modes resulted from strong correlation among the electrons .
novel phenomena in this low - energy universe are clearly distinguished from landau s fermi liquid described by the perturbative quantum many - body theory . |
presently , it is well known that information has become the most important resource leading to many major paradigm shifts . the realization has now emerged that the processing of information can not be separated from the underlying fundamental physics and that the physical aspects of information processing must be taken into consideration .
this has led to a new understanding of the information processing , cryptography , and the methods and techniques used for communication @xcite that is based on the rules of quantum theory @xcite .
a venue where information plays a central role is in the playing of games for which the established branch of mathematics called game theory binmore , rasmusen , osborne provides the necessary analytic tools , methods , and solution concepts .
any analysis of games is based on some fundamental assumptions relating to what information is available to each participating player at a particular stage of the game .
the finding that physical aspects can have crucial role in the information processing has natural implications for playing of games and considerations of the underlying fundamental physics become important also for game theory .
such questions have led to the creation of the research area of quantum games meyerdavid , ewl , ew , vaidman , benjaminhayden , enkpike , caves , johnson , marinattoweber , iqbaltoor1,duli , du , piotrowski , iqbaltoor3,flitneyabbott1,iqbaltoor2,piotrowski1,shimamura1 , flitneyabbott2,dunplayer , hanetal , iqbalweigert , mendes , cheontsutsui , iqbalepr , nawaztoor , ozdemira , cheon , shimamura , chenwang , schmidetal , ichikawatsutsui , cheoniqbal , ozdemir , flitneygreentree , prevedel , iqbalcheon , ichikawatsutsuicheon , ramzan , flitneyhollenberg , aharon , bleiler , ahmedbleilerkhan , qiang , qiang1,chappella , chappelld , iqbalabbott , chappell , iqbalcheonabbott , kolenderski , chappellb , chappellc , pawela , pawela1,brunner , frackiewicz , iqbalbayes , ramanathan , buscemi , mohammad . what these studies have shown us
is that quantum strategies result in outcomes that can defy our classical intuition .
the speed with which new quantum technologies are emerging suggest that soon it would be usual to take full advantage of quantum theory and using quantum strategies in order to beat an opposing player at some realistic game that uses quantum technology .
also , there are suggestions that quantum games can potentially provide new insights into the rise of complexity and self - organization at the molecular level where the rules are dictated by quantum mechanics @xcite . in the area of quantum games a well known quantization scheme for bimatrix games
was proposed by eisert et al .
@xcite . in this scheme ,
the players s strategies or actions , are particular local unitary transformations performed on an initial entangled state @xmath0 in @xmath1 hilbert space .
after the players actions , the quantum state passes through an unentangling gate and thereafter is called the final state @xmath2 .
this state is subsequently measured using a stern - gerlach type apparatus generating four quantum probabilities @xcite .
players payoff relations are expressed in terms of payoff entries of the corresponding bimatrix and the resulting quantum probabilities .
some experimental realizations of quantum games are described in refs .
du , prevedel , kolenderski .
one of the most appealing features of quantum games is the possibility that strategies can influence each other and form collective strategies .
the measurement basis for the final state @xmath3 is defined by associating the pure classical strategies of the players with the four basis vectors of the two - qubit quantum state .
players payoff relations contain four quantum probabilities obtained by projecting the final quantum state onto the basis vectors and are expressed in terms of players local unitary transformations and the projection postulate of quantum mechanics .
consideration of the four quantum probabilities in the player payoff relations leads to the question of whether the payoff relations in the quantum game can also be understood terms of player mixed strategies in the classical game .
this is reminiscent of the situation when quantum probabilities are factorizable where one can express the quantum probabilities in terms of player mixed strategies .
it turns out that factorizability of quantum probabilities presents a set of strong relations . in this paper
, we present two approaches in obtaining games with non - factorizable probabilities . in the first approach
, factorizability is controlled by an external parameter whereas , in our second approach , we replace the factorizability relations with two weaker relations stating that quantum probabilities are normalized and that players strategies can be expressed in terms of the quantum probabilities .
these relations are weaker as these are obeyed by factorizable quantum probabilities but the converse is not the case .
player payoff relations under the weaker relations allow us to obtain an extension of the classical symmetric bimatrix game with non - factorizable probabilities .
we then explore the nash equilibria in the symmetric bimatrix game with non - factorizable probabilities .
consider a symmetric bimatrix game @xcite @xmath4 where @xmath5 @xmath6 @xmath7 and @xmath8 are real numbers .
assume that alice s and bob s mixed strategies are @xmath9 @xmath10,$ ] respectively , at which their payoffs can be written as @xmath11 a nash equilibrium ( ne ) consists of the pair @xmath12 of strategies such that no player has any motivation to unilaterally deviate from it .
the game s nash inequalities take the form @xmath13 which can be written as @xmath14 or@xmath15 ( p^{\ast } -p ) & \geq & 0 , \notag \\ \left [ ( \alpha -\beta -\gamma + \theta ) p^{\ast } + ( \beta -\theta ) \right ] ( q^{\ast } -q ) & \geq & 0 , \label{nashclassgame}\end{aligned}\]]from which in the following , we also refer to this classical mixed strategy description of the matrix game ( [ game matrix ] ) simply as the classical game .
now consider the game ( [ game matrix ] ) when played in eisert et al.s quantization scheme .
it uses two qubits to play the quantum game whose quantum state is in @xmath1 dimensional hilbert space . in view of the game ( [ game matrix ] ) , a measurement basis for quantum state of two qubits
is chosen as @xmath16 @xmath17 @xmath18 @xmath19 .
an entangled initial quantum state @xmath0 is obtained by using a two - qubit entangling gate @xmath20 i.e. @xmath21 where @xmath22 and @xmath23 @xmath24 $ ] is a measure of the game s entanglement . for a separable or product game
@xmath25 whereas for a maximally entangled game @xmath26 . eisert et al .
chose the game to be maximally entangled i.e. the initial state is a maximally entangled state @xmath27 .
players then perform their local actions @xmath28 and @xmath29 from the set of unitary transformations @xmath30 where @xmath31 and @xmath32 .
these actions change the initial maximally entangled state @xmath0 to @xmath33 which then passes through an untangling gate @xmath34 and the state of the game changes to the final state i.e. @xmath35 .
the state @xmath2 is then measured in the basis @xmath16
@xmath36 @xmath37 @xmath38 and using the quantum probability rule player payoffs are obtained as @xmath39 the ne for the quantum game is a pair @xmath40 of local unitary transformations that are obtained from the inequalities @xmath41 that is , the ne consists a pair of unitary transformations , corresponding to the two players , such that neither player is left with any motivation to deviate from it . for @xmath42 @xmath43 @xmath44 @xmath45 the matrix ( [ game matrix ] ) gives the game of prisoners dilemma for which eisert et al . reported that a pareto - optimal ne is obtained as @xmath46 at which the players payoffs are @xmath47
we notice that the payoff relations ( [ ewlpayoffs ] ) can be written as @xmath48 where @xmath49 are four quantum probabilities obtained using the quantum probability rule i.e. projecting the final quantum state @xmath2 of the game to the four basis vectors @xmath50 @xmath51 @xmath52 @xmath53 .
the probabilities @xmath54 are normalized i.e. @xmath55 comparing the payoff relations in the classical game ( [ payoffs ] ) and in the quantum game ( [ qpayoffs ] ) , we notice that the payoff relations in the quantum game can be reduced to the payoff relations in the classical game when probabilities @xmath54 are factorizable .
probabilities @xmath56 are factorizable when for a given set of values assigned in range @xmath57 $ ] to probabilities @xmath58 we can find two probabilities @xmath59 $ ] such that @xmath54 can be written in terms of @xmath60 and @xmath61 as @xmath62 if this is the case for the probabilities @xmath54 then we can associate variables @xmath60 and @xmath61 to the players alice and bob , respectively , so that the payoff relations in the quantum game ( [ qpayoffs ] ) are interpreted as a mixed - strategy classical game i.e. @xmath63 with @xmath60 and @xmath61 satisfying the factorizability relations ( factorizability ) .
note that from the definition of factorizability in ( factorizability ) , @xmath54 represent not just four numbers in @xmath64 $ ] that add up to @xmath65 but the entirety of such four numbers that can be generated by the quantum mechanical setup used for playing a quantum game .
quantum probabilities @xmath54 , however , may not be factorizable in the sense described in ( [ factorizability ] ) .
that is , the measurement stage of the quantum game can result in such @xmath54 for which one can not find @xmath59 $ ] so that eqs . _
_ ( factorizability ) are satisfied . viewing a quantum game from this probabilistic viewpoint
, it then seems natural to ask how removing the factorizability relations ( [ factorizability ] ) may affect the outcome of the game .
the setting of a quantum game allows sets of non - factorizable probabilities and in this paper we consider role of non - factorizable probabilities in the playing of a game .
this question can also be stated as follows .
considering eqs .
( [ qpayoffs],[qpayoffs1 ] ) we can describe the factorizable game as the one for which @xmath66 where the probabilities @xmath54 are related to players s strategies @xmath67 and @xmath61 via strong factorizability relations ( [ factorizability ] ) .
now we ask what if the relationship between @xmath9 @xmath61 and @xmath54 are weaker than described by factorizability and how this is going to affect the solution of the game . to address this , in the following we introduce two approaches towards considering non - factorizable probabilities .
our first approach considers an external parameter @xmath68 that determines whether the probabilities @xmath54 @xmath69 are factorizable or not . for this , we consider the following probability distribution @xmath70 it can be confirmed that for @xmath68 in the range @xmath71 $ ] we have @xmath72 and that probabilities @xmath54 are normalized according to the eq .
( [ normalization ] ) .
when @xmath73 the distribution ( [ firstapproach ] ) reduces to the factorizable distribution of the eq .
( [ factorizability ] ) .
however , when @xmath68 is non - zero , the probability distribution ( [ firstapproach ] ) is not factorizable . using the payoff relations ( [ non - factor - payoffs ] ) , we now obtain the payoffs for alice and bob for the distribution ( [ firstapproach ] ) as @xmath74 \notag \\ & & + \gamma \left [ ( 1-k)q(1-p)+kp(1-q)\right ] + \theta \left [ 4k(1-k)pq+(1-p)(1-q)\right ] , \notag \\ \pi _ { b}(p , q , k ) & = & \alpha ( 2k-1)^{2}pq+\gamma \left [ ( 1-k)p(1-q)+kq(1-p)% \right ] \notag \\ & & + \beta \left [ ( 1-k)q(1-p)+kp(1-q)\right ] + \theta \left [ 4k(1-k)pq+(1-p)(1-q)\right ] .\end{aligned}\]]the ne strategy pair @xmath12 is then obtained from the inequalities @xmath75 q^{\ast } + ( \beta -\theta ) -k(\beta -\gamma ) \right\ } ( p^{\ast } -p ) & \geq & 0 , \notag \\ \left\ { \left [ \alpha ( 1 - 2k)^{2}-\beta -\gamma + \theta \left\ { 1 + 4k(1-k)\right\ } \right ] p^{\ast } + ( \beta -\theta ) -k(\beta -\gamma ) \right\ } ( q^{\ast } -q ) & \geq & 0.\end{aligned}\]]now , for the prisoners dilemma game considered above , we have @xmath42 @xmath76 @xmath44 @xmath45 and these reduce the above ne conditions to @xmath77 \right\ } ( p^{\ast } -p ) & \geq & 0 , \notag \\ \left\ { -1-p+k\left [ 5 - 8p(1-k)\right ] \right\ } ( q^{\ast } -q ) & \geq & 0.\end{aligned}\]]for @xmath73 these relations give the ne in the classical factorizable game i.e. @xmath78 at which the players payoffs are obtained as @xmath79 .
now consider the case when @xmath80 for which we find the ne being @xmath81 and the players payoffs are obtained as @xmath82 .
these payoffs are identical to the ones that are obtained in the maximally entangled quantum game using eisert et al s scheme @xcite with both players playing the quantum strategy @xmath83 that is defined in eq .
( [ qstrategy ] ) . however , we notice that the probability distribution in eq .
( firstapproach ) is not as general as a quantum probability distribution can be within a quantum game .
this is because for ( [ firstapproach ] ) to be a quantum probability distribution it is required to obey only the normalization constraint ( [ normalization ] ) .
although the probability distribution ( [ firstapproach ] ) is normalized it also obeys other restrictions because of its particular form and the way it is defined .
this can also be stated as follows .
whereas the probability distribution ( firstapproach ) is normalized , not every normalized probability distribution will have this form .
that is , there can be quantum probability distributions that can not be written in the same form as the distribution ( [ firstapproach ] ) .
however , the probability distribution ( firstapproach ) demonstrates the effect of a non - factorizable probability distribution on the outcome of a game .
it also produces the classical game as a special case . at this stage
, we notice that when @xmath54 are factorizable in the sense described by eqs . _ _ ( [ factorizability ] ) ,
the players strategies @xmath60 and @xmath61 can also be expressed in terms of probabilities @xmath56 as @xmath84 and using eqs .
_ _ ( [ upsilons ] ) these can be written as @xmath85 as @xmath60 and @xmath61 are the players strategies , they should be independent of each other .
this is ascertained by assuming @xmath86 to be a constant .
when quantum probabilities @xmath54 are factorizable , the above quantum game can be interpreted in terms of its playing using coins as follows .
consider a pair of biased coins that are tossed together .
either coin can land in the head @xmath87 or the tail @xmath88 state and for the pair we define @xmath89 as the probabilities of the coins landing in the @xmath90 , @xmath91 , @xmath92 , @xmath93 states , respectively . here
, for instance , @xmath94 is the probability that the first coin ( or alice s coin ) lands in the @xmath95 state whereas the second coin ( or bob s coin ) lands in the @xmath96 state . in eq .
( [ p&q ] ) , @xmath60 is then the probability , in the joint toss of two coins , that the first coin lands in the @xmath95 state and @xmath61 is the probability that the second coin lands in the @xmath95 state .
when @xmath60 and @xmath61 are defined as the players strategies , a player plays his / her strategy by changing the bias of the coin to which he / she is given access to . also , note that for factorizable quantum probabilities using eqs .
( [ factorizability],[p&q ] ) we can write @xmath97 that immediately suggests an interesting situation that we consider in the following .
we replace the factorizability relations ( [ factorizability ] ) relating to @xmath9 @xmath61 and @xmath54 with the relations given by the eqs .
( [ p&q ] ) and the normalization ( [ normalization ] ) .
we then note that using ( [ p&q],[normalization ] ) the players payoff relations ( [ qpayoffs ] ) can be re - expressed in terms of @xmath60 and @xmath61 . with such re - expression of the players payoffs
, we can then study the impact on nash equilibria of the game .
as factorizability relations are not assumed to hold , the probabilities @xmath54 can become non - factorizable and are thus not describable according to the eqs .
_ _ ( [ factorizability ] ) . in other words ,
for our game - theoretic scenario , we are interested in finding the impact of dropping the assumption of factorizability of @xmath54 on the nash equilibria of a game , while the relations ( [ p&q ] ) and the normalization constraint on quantum probabilities ( [ normalization ] ) remain valid .
to find this impact we proceed as follows . using normalization constraint for quantum probabilities ( [ normalization ] ) and the defining relations ( [ p&q ] ) , we re - express players s payoff relations ( [ qpayoffs ] ) in the quantum game in terms of @xmath60 and @xmath61 . for this , from eqs . _ _ ( [ p&q ] ) we write @xmath98 and note that by substituting from eqs .
_ _ ( [ upsilons2&3 ] ) the eq . _ _ ( [ normalization ] ) , describing the normalization of @xmath99 , takes the form @xmath100 now , from eqs . _ _ ( [ upsilons2&3 ] ) we substitute for @xmath101 and @xmath102 and from eq . _ _ ( [ upsilon4 ] ) we substitute for @xmath103 into the payoff relations ( [ qpayoffs1 ] ) to obtain @xmath104 eqs . _ _ ( [ re - def - eqs ] ) can be considered as a re - expression of the players payoff relations in the quantum game under the relations ( [ p&q ] ) along with the normalization constraints ( [ normalization ] ) . as the right sides of eqs .
_ _ ( [ re - def - eqs ] ) are functions of @xmath60 and @xmath105 whereas @xmath86 is considered a constant , we can , therefore , write once again @xmath106 however , the sense of these relations is different from that of the relations ( [ qpayoffs1 ] ) . relations ( [ qpayoffs1 ] ) assume factorizability ( [ factorizability ] ) of @xmath54 whereas relations ( [ qpayoffs2 ] ) assume the payoff relations in the quantum games ( qpayoffs ) , the relations ( [ p&q ] ) along with the normalization ( normalization ) .
also , note that in eqs .
_ _ ( [ re - def - eqs ] ) only @xmath107 and @xmath61 are controlled by the players alice and bob , respectively , as @xmath108 is considered a constant .
recall that the players local unitary actions @xmath28 and @xmath109 transform a maximally entangled initial state @xmath0 to @xmath2 and @xmath99 are obtained by projecting @xmath2 to the basis in four - dimensional hilbert space .
as @xmath110 and @xmath111 , therefore , we can write @xmath112 that is , @xmath60 and @xmath61 , as they appear in eqs . _ _ ( [ re - def - eqs ] ) , are functions of @xmath28 and @xmath29 .
this is significant because : \a ) payoff relations ( [ re - def - eqs ] ) are reduced to the players payoff relations ( [ payoffs ] ) for the classical game by replacing @xmath60 with @xmath113 and @xmath61 with @xmath114 in ( [ re - def - eqs ] ) and then allowing @xmath54 to become factorizable according to eqs . _ _ ( [ factorizability ] ) .
\b ) when @xmath54 are factorizable and , therefore , eqs .
_ _ ( re - def - eqs ) are reduced to the classical mixed - strategy payoff relations ( [ payoffs ] ) , then @xmath60 and @xmath61 can be associated to the unitary transformations @xmath28 and @xmath29 , respectively , that is @xmath115 \c ) the presence of @xmath86 within the two eqs . _
_ ( re - def - eqs ) is reminiscent of the measure of entanglement @xmath23 in the initial state when the quantum game is played in eisert et al.s scheme . in view of the above considerations , and the eqs .
_ _ ( [ re - def - eqs ] , [ qpayoffs2 ] , [ assoc ] ) , we notice that the right sides of the payoff relations ( [ re - def - eqs ] ) depend on @xmath60 and @xmath61 and we can thus write the payoff relations ( [ re - def - eqs ] ) as @xmath116 and consider these as the defining payoff relations for another game .
note that for quantum game the players strategies are unitary transformations @xmath117 and @xmath29 , whereas for the new game with the payoff relations ( [ newgame ] ) the players strategies are @xmath59 $ ] .
a game is defined by its payoff relations and in ( [ newgame ] ) we have a game that makes no reference to quantum operations . the game represented by the payoff relations ( [ newgame ] ) can be called a non - factorizable game or a game that permits non - factorizable probabilities .
we remind reader that in obtaining this game we have essentially replaced the factorizability relations given by eqs .
_ _ ( [ factorizability ] ) by the defining relations for @xmath60 and @xmath61 given by eqs . _ _ ( [ p&q ] ) and the normalization of quantum probabilities given by eq . _ _ ( normalization )
. now , the relations given by eqs . _ _ ( [ p&q ] , [ normalization ] ) together are weaker than the factorizability relations given by eqs .
_ _ ( [ factorizability ] ) as when the factorizability relations ( factorizability ) hold , the relations given by eqs .
_ _ ( [ p&q ] , [ normalization ] ) also hold .
this is because , as can be seen from eqs .
( [ factorizability ] ) , factorizable set of probabilities are normalized and as it follows from factorizability relations that @xmath60 and @xmath61 can then be expressed in terms of the quantum probabilities @xmath54 according to eqs .
_ _ ( p&q ) .
however , the converse is not true as eqs . _ _ ( [ p&q],normalization ) together can not give us the factorizability relations ( factorizability ) .
with reference to eqs .
_ _ ( [ p&qusingus ] ) , we note that @xmath118 and @xmath119 and can then write @xmath120 where @xmath86 is assumed a constant . by writing the payoff relations ( [ newgame ] ) we re -
express them in the quantum game ( [ ewlpayoffs ] ) in terms of @xmath9 @xmath10 $ ] both of which can be described in terms of @xmath28 and @xmath29 . here
we show how the mixed - strategy classical game , as described by the payoff relations ( [ payoffs ] ) , can be located within the game ( newgame ) with non - factorizable probabilities .
as considered in a ) above , for the case when players s strategies are unitary transformations @xmath121 and @xmath29 , the payoff relations ( [ newgame ] ) for the game with non - factorizable probabilities are reduced to the payoff relations ( payoffs ) for the classical game by firstly replacing @xmath60 with @xmath122 and @xmath61 with @xmath114 in ( re - def - eqs ) and then allowing the probabilities @xmath54 to become factorizable according to eqs . _ _ ( [ factorizability ] ) .
the mixed - strategy classical game is thus located within the game with non - factorizable probabilities under the assumption of factorizability and the relations for @xmath60 and @xmath61 expressed in terms of the probabilities @xmath56 .
note that the right sides of the payoff relations ( [ newgame ] ) for the game with non - factorizable probabilities are obtained from the right sides of payoff relations ( [ re - def - eqs ] ) in the quantum game . by replacing @xmath123 by @xmath124 where @xmath118 and @xmath119 , we have replaced players s local quantum strategies @xmath28 and @xmath29 by the variables @xmath60 and @xmath61 , respectively . as @xmath125 and @xmath126 , the players strategies can now be interpreted as non - local or global .
the nash equilibria for the game ( [ newgame ] ) can now be found , as usual , from the inequalities @xmath127 that gives @xmath128 by comparing nash inequalities ( [ nashqgame ] ) in this game with nonfactorizable probabilities with the nash inequalities ( [ nashclassgame ] ) in the classical game , we notice that only pure nash equilibria can exist in the game with non - factorizable probabilities . when @xmath129 we have @xmath130 whereas when @xmath131 we have @xmath132 . at these nash
equilibria the players payoffs are obtained as @xmath133 which for @xmath134 give @xmath135 and @xmath136 . however , for the same nash equilibria with @xmath137 we obtain @xmath138
eisert et al.s scheme for quantization of a bimatrix game involves four quantum probabilities . when these probabilities are factorizable in a specific sense , as described by eqs .
_ _ ( [ factorizability ] ) , the quantum game attains the interpretation of a mixed - strategy classical game .
the factorizability relations are quite strong and allow expressing player strategies in terms of the four probabilities .
this paper presents two approaches in obtaining bimatrix games with non - factorizable probabilities .
our first approach discusses a non - factorizable probability distribution in which the non - factorizability is controlled by an external parameter @xmath68 . in our second approach
we replace the relations describing factorizability , and given by eqs .
( factorizability ) , with two weaker relations as described by eqs .
_ _ ( [ normalization ] , [ p&q ] ) .
the weaker relations state that the four quantum probabilities are normalized and that players strategies can be expressed in terms of the four quantum probabilities .
these relations are weaker than the factorizability relations because when the factorizability relations hold , these relations also hold . whereas the converse is not true i.e. when these two relations hold the factorizability may not hold .
we then analyze the players payoff relations that are subject to these weaker relations .
as the two relations are weaker than the factorizability relations , we name the resulting game as the game with non - factorizable probabilities .
our analysis of a symmetric bimatrix game shows that such a game can only have pure nash equilibria when probabilities are non - factorizable and the value assigned to the probability @xmath86 decides the players payoffs at these nash equilibria .
we note that the relations @xmath110 and @xmath139 as obtained from the factorizability constraints ( factorizability ) are apparently the simplest expressions connecting the players strategies @xmath60 and @xmath61 to the quantum probabilities @xmath54 @xmath140 .
however , this does not mean that these are the only possible expressions that are consistent with the factorizability constraints ( [ factorizability ] ) .
there can be other possible cases , for instance , when @xmath141 and @xmath142 i.e. both @xmath60 and @xmath61 are functions of all four quantum probabilities @xmath54 @xmath143 .
in such a case , the factorizability constraints will take the following form @xmath144 , \text { } \epsilon _ { 3}=% \left [ 1-p(\epsilon _ { i})\right ] q(\epsilon _ { i}),\text { } \epsilon _ { 4}=% \left [ 1-p(\epsilon _ { i})\right ] \left [ 1-q(\epsilon _ { i})\right ] , \label{factorizability1}\ ] ] and ,
depending on how the functions @xmath141 and @xmath145 are defines , the analysis will lead to different outcome of the quantum game .
essentially , in this paper we use the factorizability constraints ( [ factorizability ] ) to obtain such functions that express players strategies in terms of the probabilities @xmath54 and there can be more than one possible ways of doing that and each way has to respect the requirement that when @xmath54 are factorizable , now in the sense of eqs .
( [ factorizability1 ] ) , the quantum game reduces itself to the classical mixed - strategy game .
zdemir , s. k. , shimamura , j. , imoto , n. , quantum advantage does not survive in the presence of a corrupt source : optimal strategies in simultaneous move games , _ phys . _
_ _ a _ * 325 * , 104 - 111 ( 2004 ) . cheon , t. , game theory formulated on hilbert space , _ aip _ _ conf .
_ _ proc_. volume * 864 * , 254 - 260 ,
_ quantum _ _ computing : _ _ back _ _ action _ ( 2006 ) , indian institute of technology , kanpur , india .
arxiv : quant ph/0605134 kolenderski , p. , sinha , u. , youning , l. , zhao , t. , volpini , m. , cabello , a. , laflamme , r. , jennewein , t. , aharon - vaidman quantum game with a young - type photonic qutrit , _ phys . _
_ rev . _ _ a _ * 86 * , 012321 ( 2012 ) .
iqbal , a. , chappell , j. m. , li , q. , pearce , charles e.m . , abbott , d. , a probabilistic approach to quantum bayesian games of incomplete information , _ quantum information processing _ * 13 * , 2783 - 2800 ( 2014 ) . | a quantum game - theoretic setting enables the analysis of strategic interaction among agents with access to quantum resources .
one of the original motivations for re - expressing quantum algorithms , and quantum communication protocols , in terms of competing quantum agents was to provide new insights into their workings .
looking closely , one finds that quantum strategic interactions utilize quantum probabilities and such probabilities are realizable in quantum mechanical experiments that allow implementation of unitary transformations , quantum measurements etc .
this paper presents a perspective on a quantum game - theoretic framework involving quantum probabilities , which are not subject to the strong factorizability constraints that classical probabilities are found to possess .
we suggest that a quantum version of a classical game can be obtained more directly if we replace the factorizability relations for probabilities in a classical game with two weaker constraints that are also obeyed by the quantum probabilities .
the solutions of symmetric bimatrix games with such ` non - factorizable ' probabilities are then analyzed and compared to the solutions that appear in the standard version of a quantum game . |
the last couple of years have witnessed an impressive series of achievements in the field of large - scale structure , thanks to new large surveys of galaxies and clusters of galaxies .
the enthusiasm for new results on the clustering of galaxies and clusters has been strenghtened by the unprecedented possibility to couple these to the anisotropies in the cosmic microwave background over an overlapping range of scales ( see contributions by melchiorri and cooray , this volume ) . in this brief review
i have tried and provide a general guide for the non - specialist through some of the large - scale structure results .
clearly , such a review is far from being complete , although the references indicated should allow the reader to find further links to the available literature on the subject ( before february 2002 ) .
i therefore apologize to those colleagues whose work has not been adequately covered .
the currently popular model for the origin and evolution of structure in the expanding universe is the cold dark matter ( cdm ) model @xcite , whose global features provide a framework which is remarkably consistent with a large number of observations .
the `` cosmology 2000 '' version of the model ( often referred to in the recent literature as the concordance model ) , which takes into account the independent evidences for a flat geometry ( from the angular power spectrum of anisotropies in the cosmic microwave background @xcite ) and an accelerated expansion ( from the luminosity - distance relation of distant supernovae , used as `` standard candles '' @xcite ) is one where cdm , in the form of some kind of weakly - interacting non - relativistic particles ( see pertinent articles in this volume ) , contributes about 25 - 30% of the total density , with the remaining 70% provided by a `` dark energy '' associated to a _
cosmological constant_. i will comment at the end of this review on how comfortable we should feel in front of the number of `` unseen '' ingredients of this model . here
we shall use the model as it is , in fact `` just a model '' , i.e. a physically motivated machinery which works remarkably well when confronted with a variety of observations .
choosing cdm ( or any other model ) means specifying a _ transfer function _ @xmath5 .
this can be thought of as describing a linear amplifier is the fourier wavenumber , i.e. the inverse of a 3d spatial scale @xmath6 , measured in @xmath7 , with @xmath8 being the hubble constant in units of @xmath9 .
most recent determinations indicate @xmath10 with about 10% error , see w. freedman contribution to this volume . ] which filters the primordial spectrum of fluctuations ( typically of the scale - invariant form @xmath11 generally predicted by inflation ) to produce the shape of power spectrum we can still observe today on large [ @xmath12 scales , @xmath13 @xcite .
one of the nice features of the cdm spectral shape in any of its variants is to naturally lead to a _
hierarchical _ growth of structures , where larger entities are continuously formed from the assembly of smaller ones @xcite . within the _ gravitational instability _
picture , the formation of galaxies and larger structures is completely driven by the gravitational field of the dark matter , with our familiar _ baryonic _
matter representing only a tiny bit of the mass ( @xmath14 of the total energy density ) .
the lighting - up of galaxies and other luminous objects depends then on how the baryons cool within the dark matter haloes and form stars , ending up as the only directly visible peaks of a much larger , invisible structure .
this increasing complexity in the physics involved in this cascade of processes is reflected by the limits in the predicting power of current detailed models of galaxy formation .
predictions from purely gravitational n - body experiments concerning the overall clustering of the dark mass can be regarded as fairly robust @xcite .
more complex semi - analytical calculations addressing the history of galaxy formation have seen exciting progress during the last few years @xcite , but they clearly still depend on a large number of not fully constrained parameters . direct measurements of large - scale structure are a classical test - bench for cdm models and they have , for example , been the reason for rejecting the original einstein - desitter ( @xmath15 ) version of the model , whose transfer function is inconsistent with the observed balance of large- to small - scale power @xcite . the main problem in the game is that true direct measurements of mass structure ( as e.g. through peculiar velocities or gravitational lensing cosmic shear ) , are not trivial : most observations have necessarily to use radiating objects as tracers of the mass distribution , and thus need to go through the uncertainties mentioned above to allow meaningful comparison to model predictions @xcite .
since the 1970 s , redshift surveys of galaxies have represented the primary way to reconstruct the 3d topology of the universe @xcite .
year 2000 has seen the completion and public release @xcite of the first 100,000 galaxy redshift measurements by the anglo - australian 2df galaxy redshift survey , the largest complete sample of galaxies with measured distances to date @xcite .
this survey includes all galaxies with blue magnitude @xmath2 brighter than @xmath16 , mainly over two areas covering @xmath17 square degrees in total , to an effective depth of about @xmath18 ( @xmath19 ) .
its immediate precursors @xcite reached a similar depth , but over much smaller areas : for comparison , the las campanas redshift survey ( lcrs @xcite ) , measured a total of 16,000 redshifts , against the 250,000 that will eventually form the full 2df survey .
a plot of the galaxy distribution within the two main sky regions of this survey is shown in fig.[2df_cones ] . here
one can appreciate in detail the wealth of structures typical of the distribution of galaxies : clusters , superclusters ( filamentary or perhaps sheet - like ) and _ voids _
, i.e. regions of very low galaxy density @xcite .
i will discuss the main clustering results from this survey in the following sections . in a parallel effort ,
the sloan digital sky survey ( sdss ) is covering a large fraction of the northern sky with a uniform five - band ccd survey ( @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 ) , plus measuring redshifts for one million galaxies over the same area @xcite . the photometry reaches a red magnitude @xmath25 ; the redshift survey is limited to galaxies brighter than @xmath26 , resulting in a depth similar to that of esp , lcrs and 2df , but over a very large area . early data over 462 square degrees have been recently released .
the sdss represents the largest and most comprehensive galaxy survey work ever conceived : in addition to the redshift survey , the multi - band photometry is going to be of immense value for a number of studies , as estimating _ photometric redshifts _ @xcite to much larger depth , or selecting samples with well - defined colour / morphology properties .
a relevant example of such applications has been the discovery of several high - redshift ( @xmath27 ) quasars , including the @xmath28 case for which the first possible detection of the long - sought gunn - peterson effect , essentially the fingerprint of the `` dark - ages '' , has been recently reported @xcite
. another important application will be the selection of about @xmath29 `` red luminous '' galaxies with @xmath30 , that will be observed spectroscopically providing a nearly volume - limited homogeneous sample out to @xmath31 , to study the clustering power spectrum on extremely large scales @xcite .
both the 2df and sdss redshift surveys rely upon the large multiplexing performances of fiber - fed spectrographs , that allow the light from several hundred galaxies over a field of view of 1 - 2 degrees to be conveyed into the same slit on the spectrograph .
this specific technology , in various forms , has been the key to the explosion of the redshift survey industry in the 1990 s , bringing the efficiency from the @xmath32 redshifts / night for galaxies brighter than blue magnitude @xmath33 of the 1970 s , to the current @xmath34 redshifts / night to @xmath35 ( see e.g. @xcite for a more accurate account ) .
clusters of galaxies are complementary tracers of large - scale structure ( see e.g. @xcite ) . especially before the current era ,
when @xmath36 galaxy redshifts are becoming available over comparable volumes , groups and clusters have represented the most efficient alternative to map very large volumes of the universe , exploring in this way the gross structure and its statistical properties in the weak clustering regime .
x - ray selection represents currently the most physical way to identify large homogeneous samples of clusters of galaxies ( see also discussion in @xcite ) .
clusters shine in the x - ray sky due to the _ bremsstrahlung _ emission produced by a hot plasma ( @xmath37 kev ) trapped within their potential wells .
the bolometric emissivity ( i.e. the energy released per unit time and volume ) of this thin gas is proportional to its density squared and to @xmath38 .
such dependence on @xmath39 makes clusters stand out more in the x - rays than in the optical light distribution ( @xmath40 ) . under the assumption of hydrostatic equilibrium ,
the intracluster gas temperature , measured through the x - ray spectrum , is a direct probe of the cluster mass : @xmath41 ( where @xmath42 is the proton mass , @xmath43 the gas mean molecular weight , @xmath44 the galaxy 1d velocity dispersion and @xmath45 the cluster virial mass ) .
x - ray luminosity , a more directly observable quantity with current instrumentation , shows a good correlation with temperature , @xmath46 with @xmath47 and a scatter @xmath48 .
the practical implication , even only on a phenomenological basis , is that clusters selected by x - ray luminosity are in practice mass - selected , with an error @xmath49 % ( see e.g. @xcite and references therein for a more critical discussion ) .
last , but not least , the selection function of an x - ray cluster survey can be determined to high accuracy , knowing the properties of the x - ray telescope used , in a similar way to what is usually done with magnitude - limited samples of galaxies @xcite .
this is of fundamental importance if one wants to compute statistical quantities and test cosmological predictions as , e.g. , the mean density or the clustering of clusters above a given mass threshold @xcite .
[ fig : reflex_cone ] plots the large - scale distribution of x - ray clusters from the reflex ( rosat - eso flux limited x - ray ) cluster survey , the largest redshift survey of x - ray clusters with homogeneous selection function to date @xcite .
this data set , completed in 2000 and publicly released at the beginning of 2002 , is based on the x - ray all - sky survey performed by the rosat satellite in the early 1990 s ( see e.g. @xcite for a comprehensive summary ) .
reflex includes 452 clusters over the southern celestial hemisphere and is more than 90% complete to a flux limit of @xmath50 erg s@xmath51 @xmath52 ( in the rosat energy band , 0.1 - 2.4 kev ) .
the volume explored is larger than that of the 2df survey and comparable to the volume that will be filled by the sdss 1-million - galaxy redshift survey @xcite . ] .
the fine structure visible in fig . 1
is obviously lost in the cluster distribution ; however , a number of cluster agglomerates and filamentary structures with large sizes ( @xmath53 ) are evident , showing that inhomogeneities are still strong on such very large scales .
large - scale structure models as cdm are specified in terms of a specific shape for the power spectrum of density fluctuations @xmath54 .
analogously to standard signal theory , the power spectrum describes the squared modulus of the amplitudes @xmath55 ( at different spatial wavelengths @xmath6 ) of the fourier components of the fluctuation field @xmath56 @xcite . studying the power spectrum of the distribution of luminous objects on sufficiently large scales ,
where the growth of clustering is still independent of @xmath57 , we hope to recover a relatively undistorted information to test the models .
the uncertainties in relating the observed @xmath54 of , e.g. , galaxies to that from the theory are due to ( a ) nonlinear effects that modify the linear shape below some scale ; ( b ) the unknown relation between the distribution of the light and that of the mass , that is what the models predict .
the first problem can be circumvented by pushing redshift surveys to larger and larger scales , as to work well into the linear regime ( and/or following nonlinear evolution through numerical simulations ) .
the second one involves knowing the _ bias _ parameter @xmath58 .
the bias parameter can be defined either in integral terms , through the ratio between the variances in galaxy counts and in the mass density @xmath59 , or differentially as @xmath60 . for galaxy surveys
, @xmath58 can only be deduced using additional external constraints on the power spectrum normalization ( e.g. from cmb anisotropies , that directly probe fluctuations in the mass @xcite ) , or studying higher - order moments of the distribution @xcite .
x - ray selected clusters , on the other hand , have a specific advantage in this respect , as their bias factor can be computed directly once the sample selection function is known ( e.g. @xcite ) .
the 2df and reflex surveys have produced the best estimates to date of the power spectrum of galaxies and x - ray clusters , respectively .
[ fig : pk ] ( left panel ) compares these data sets directly , evidencing the remarkable similarity of the shape of @xmath54 for these two classes .
this provides a direct confirmation of the bias scenario , where clusters form at the rare high peaks of the mass density distribution @xcite and for this reason display a stronger clustering amplitude .
[ xi - surveys ] in the same figure i have also plotted the predictions for the mass power spectrum of two models of the cdm family , computed as described in @xcite .
a model very close to what we defined as the ` concordance ' model ( @xmath4 , @xmath61 , @xmath62 ) provides in general an excellent fit to the 2df power spectrum , with a bias parameter ( i.e. normalization ) close to unity , in a pure cdm universe the observed shape depends only on @xmath0 , not on @xmath63 ( which on the other hand influences the normalization ) . in the literature
, this is often parameterized through a _ shape parameter _
@xmath64 , where @xmath65 in case of negligible baryon fraction . the ` concordance ' model , therefore , has @xmath66 . ] .
the upper solid curve , on the other hand , is the same model re - normalized as @xmath67 where the cluster bias parameter @xmath68 has been computed taking into account the specific mass distribution function of the cluster sample , using a relatively straightforward theory @xcite ( see @xcite for more details ) .
it is for these computations that a well - understood mass selection function of our clustering tracers is crucial .
the general result ( an additional step with respect to galaxies ) , is that our fiducial low-@xmath0 cdm model is capable to match very well * both * the shape and amplitude of the cluster @xmath54 @xcite .
the same shape agrees well also with the power spectrum of the distribution of qso s from the 2qz survey , a large redshift survey of colour - selected quasars carried out using the same 2df spectrograph at the anglo - australian telescope @xcite . as can be seen from fig .
[ fig : pk ] , the low-@xmath0 cdm model predicts a maximum for @xmath54 around @xmath69 .
this turnover in the power spectrum is an imprint of the horizon size at the epoch of matter - radiation equality @xcite and marks an `` homogeneity scale '' , above which ( smaller @xmath57 s ) the variance drops below the white - noise behaviour . in a pure fractal universe , for example
, @xmath54 would continue to rise when moving to smaller and smaller @xmath57 s @xcite .
in fact , at least visually the data of fig .
[ fig : pk ] do not really show a convincing indication for a maximum .
in addition , on such extremely large scales ( @xmath70 ) , the effect of the survey geometry on the measured power can be very significant , resulting in an effective survey _ window function _ in fourier space which is convolved with the true underlying spectrum ( e.g. @xcite ) . for highly asymmetric geometries ,
the plane - wave approximation intrinsic in the fourier decomposition fails , and the convolution with the window function easily mimics a turnover in a spectrum with whatever shape ( e.g @xcite ) .
the best solution in such cases is to resort to survey specific eigenfunctions as those provided by the karhunen - loeve ( kl ) transform @xcite .
this technique has been intensively applied to the reflex data @xcite , confirming a shape corresponding to @xmath71 ( corresponding to a turnover @xmath72 ) .
a more recent kl analysis using the joint constraints from both the reflex mass function and power spectrum , provides a much stronger result , @xmath73 , together with a power spectrum normalization @xmath74 ( expressed in the conventional form as the variance within @xmath75-radius spheres ) @xcite .
this normalization agrees very well with the completely independent value obtained from a joint analysis of the 2df survey and cmb data@xcite .
the sdss will be the best data set to study the details of the power spectrum around the peak scale .
preliminary results based on a small subset of the survey @xcite indicate a best - fitting cdm spectral shape @xmath76 , i.e. virtually the same as measured by 2df and reflex .
again , we see an impressive convergence of independent observations towards the same low-@xmath0 cdm model . the sdss luminous red - galaxy sample , in particular , will be unique to investigate the presence of _ baryonic features _ in the power spectrum , produced by oscillations in the baryonic matter component within the last - scattering surface @xcite .
these features are expected to be of very low amplitude unless the baryon density is much higher than currently established and thus require a very high signal - to - noise and frequency resolution in @xmath54 .
potential wiggles seen in the 2df power spectrum , in fact , are shown to be an artifact of the survey window function and the strong bin - to - bin correlation @xcite .
in fact , the simplest statistics one can compute from the data and also that for which the selection function is more directly corrigible , is not the power spectrum , but rather its fourier transform , the _ two - point correlation function _
@xmath77 , which measures the excess probability over random to find a galaxy at a separation @xmath78 from a given one @xcite . fig .
[ xi - surveys ] ( right panel ) shows the correlation function measured in _ redshift space _ , @xmath79 ( see next section ) , from the 2df and sdss current data sets @xcite , compared to the lcrs @xcite .
also shown ( dotted lines ) is the real - space @xmath77 reconstructed from the apm angular survey @xcite .
the shape of @xmath79 is roughly a power law @xmath80 between 0.1 and @xmath81 , with a _ correlation length _ @xmath82 .
the overall difference with the @xmath77 from the apm survey ( which is in real space , being based on a deprojection of angular clustering ) , is due to redshift - space effects , that i will address in detail in the next section .
note how @xmath79 maintains a low - amplitude , positive value out to separations of more than @xmath83 , with the 2df and sdss data possibly implying a zero - crossing scale approaching @xmath84 .
this comparison shows explicitly why large - size galaxy surveys are so important , given the weakness of the clustering signal at such large separations in the power spectrum , corresponding to a spatial wavelength @xmath6 , relates approximately to @xmath85 in @xmath77 . ] .
the separation @xmath86 between two galaxies computed using their observed redshifts is not a true distance : the red - shift observed in the galaxy spectrum is in fact the quantity @xmath87 , where @xmath88 is the component of the galaxy peculiar velocity along the line of sight .
this contribution is typically of the order of @xmath89 for galaxies in the general field , but can rise above @xmath90 within high - density regions as rich clusters of galaxies .
[ xi - surveys ] shows explicitly the consequence of such _ redshift - space distortion _ for the correlation function : @xmath79 ( all points ) is _ flatter _ than its real - space counterpart ( dotted lines ) .
this is the result of two concurrent effects : on small scales ( @xmath91 ) , clustering is suppressed by high velocities in clusters of galaxies , that spread close pairs along the line of sight producing what in redshift maps are sometimes called `` fingers of god ''
. some of these are perhaps recognisable in fig . [
2df_cones ] as thin radial structures .
on the other hand , large - scale coherent streaming flows of galaxies towards high - density structures enhance the apparent contrast of these , when seen perpendicularly to the line of sight . this effect , on the contrary , amplifies @xmath79 above @xmath92 .
[ fig : reflex_csipz ] such peculiar velocity contribution can be disentangled by computing the two - dimensional correlation function @xmath3 , where the separation vector @xmath86 between a pair of galaxies is decomposed into two components , @xmath93 and @xmath94 , parallel and perpendicular to the line of sight respectively ( see @xcite for details ) .
the result is a bidimensional map , whose iso - correlation contours look as in fig .
[ 2dfbutterfly ] , where the 2df @xmath3 is plotted @xcite .
redshift - space distortions contain important information as galaxy motions are a direct dynamical probe of the mass distribution @xcite .
non - linear distortions are a measure of the `` temperature '' of the galaxy soup on small scales , and they are in principle related to @xmath0 through a _ cosmic virial theorem
_ @xcite , which however has been shown to be difficult to apply in practice to real data @xcite .
linear distortions produced by infall provide a way to measure the parameter @xmath95 , i.e. essentially the mass density of the universe modulo the bias parameter .
as thoroughly explained in the excellent review by andrew hamilton @xcite , this can be achieved by measuring the oblate compression of the contours of @xmath3 along @xmath93 .
one way to do this is to expand @xmath3 in spherical harmonics . in linear perturbation theory ,
only the monopole @xmath96 , quadrupole @xmath97 and hexadecapole @xmath98 are non - zero , and it has been shown @xcite that @xmath99 can in principle be derived directly through a combination of these quantities . in practice , linear and non - linear effects
are interlaced out to fairly large scales ( @xmath100 h@xmath51 mpc ) , and require a careful modeling also for a survey as large as the 2df .
this has been done , but with a 1-@xmath101 error of @xmath102 , as discussed by hamilton in his contribuion to this volume . ]
@xcite , showing that the quadrupole - to - monopole ratio of the map of fig .
[ 2dfbutterfly ] requires a universe with @xmath103 .
if 2df galaxies are unbiased ( @xmath1 ) , this leads to @xmath104 .
clusters of galaxies clearly also partake in the overall motion of masses produced by cosmological inhomogeneities .
line - of - sight spurious effects ( as projections in optically - selected cluster catalogues @xcite or large redshift errors ) and limited statistics , prevented so far the detection of true velocity anisotropies in cluster @xmath3 maps . fig .
[ fig : reflex_csipz ] plots @xmath3 for the reflex survey , which shows evidence for the compression of the contours along the line of sight , of the kind expected by the linear infall of clusters towards superstructures @xcite .
we are definitely in a golden age for observational cosmology and in particular for the study of large - scale structure .
we never had such a wealth of diverse data at our disposal , through which we are pinning down the values of cosmological parameters to a high accuracy ( e.g. @xcite ) .
the observational facts we have reviewed here contribute to further reinforce the remarkable convergence among different observables ( cmb , large - scale structure , distant supernovae , cluster evolution , to mention a few ) towards a model with flat geometry ( @xmath105 ) provided by the combination of a dominating cold dark matter component ( @xmath106 , with @xmath107 ) and a dark energy of unknown nature ( the cosmological constant , @xmath108 ) .
still , we can not avoid to note that such wonderful `` standard '' cosmological model is full of `` unseen '' ingredients , as a dark matter nobody has detected so far ( but see contribution by p. belli in this volume ) and a dark energy we have little idea where it could come from . seen from outside , this might look as an almost epicyclic model and i believe understanding its foundations provides one of the major challenges for particle physics and cosmology in the next decade . * acknowledgments . *
i thank the organizers of the dark2002 meeting for inviting me to give this review .
i am grateful to all my collaborators for all our common results discussed here , in particular p. schuecker , c. collins and h. bhringer .
i thank i. zehavi and e. hawkins for providing their clustering results in electronic form .
blumenthal , g.r .
, et al . :
_ nature _ * 311 * , 517 ( 1984 ) de bernardis , p. , et al . :
_ nature _ * 404 * , 955 ( 2000 ) perlmutter , s. , riess , a. : in `` cosmo-98 : ii int .
workshop on particle physics and the early universe '' , d.o .
caldwell ed .
, aip conference proceedings , vol .
478 . , ( woodbury , ny 1999 ) , p. 129
peacock , j.a . : _ cosmological physics _
( cambridge university press , cambridge 1999 ) padmanabhan , t. : _ structure formation in the universe _ ( cambridge university press , cambridge 1993 ) white , s.d.m . , &
rees , m.j . : mnras , * 183 * , 341 ( 1978 ) jenkins , a. , et al .
: mnras * 321 * , 372 ( 2001 ) somerville , r.s .
: mnras * 320 * , 504 ( 2001 ) kauffmann , g. , charlot , s. , haehnelt , m. : phil
. trans . of the royal soc . of london ,
series a , vol . * 358 * , p. 2121
( 2000 ) cole , s. , et al . :
mnras * 319 * , 168 ( 2000 ) silva , l. , et al .
: ap&ss * 276 * , 1073 ( 2001 ) maddox , s.j .
, et al . :
mnras * 242 * , 43p ( 1990 ) borgani s. , guzzo l. : _ nature _ * 409 * , 39 ( 2001 ) peacock , j.a . , et al . ( the 2df team ) : _ nature _ * 410 * , 169 ( 2001 ) rood , h.j .
: ara&a * 26 * , 245 ( 1998 ) colless , m. , et al . ( the 2dfgrs team ) : mnras * 328 * , 1039 ( 2001 ) shectman , s.a . , et al .
: apj * 470 * , 172 ( 1996 ) vettolani , g. , et al . ( the esp team ) : a&as , * 130 * 323 ( 1998 ) percival , w. , et al .
( the 2dfgrs team ) : mnras * 327 * 1297 ( 2001 ) peacock , j.a . & the 2dfgrs team : in _ 20th texas symposium on relativistic astrophysics _ , ( austin , december 2000 ) , in press ( 2002 ) ( astro - ph/0105450 ) york , d.g .
( the sdss collaboration ) : aj , * 120 * 1579 ( 2000 ) fernndez - soto , a. : ap&ss , * 276 * , 965 ( 2001 ) yee , h. : in _ xth rencontres de blois : the birth of galaxies _ , in press ( 1998 ) ( astro - ph/9809347 ) becker , r.h . & the sdss collaboration : aj , in press ( 2002 ) ( astro - ph/0108097 ) eisenstein , d.j .
, et al . ( the sdss collaboration ) : aj , * 122 * 2267 ( 2001 ) guzzo , l. : in _ xixth texas symposium on relativistic astrophysics & cosmology _ , nucl .
b ( proc . suppl . ) vol.80 , p.233 ( 09/06 ) , e. aubourg et al .
( 2000 ) ( ext . version in astro - ph/9911115 ) nichol , r.c .
: in _ tracing cosmic evolution with galaxy clusters _ , asp conf .
series , s. borgani et al .
, in press ( 2002 ) ( astro - ph/0110231 ) bartlett , j.g .
: in _ tracing cosmic evolution with galaxy clusters _ , asp conf . series , s. borgani et al .
, in press ( 2002 ) ( astro - ph/0111211 ) ellis , r.s .
: in _ tracing cosmic evolution with galaxy clusters _ , asp conf .
series , s. borgani et al .
, in press ( 2002 ) ( astro - ph/0112540 ) borgani , s. , et al .
: apj * 561 * , 13 ( 2001 ) rosati , p. , et al .
: apj * 445 * , l11 ( 1995 ) bhringer , h. , et al .
( the reflex team ) : a&a * 369 * , 826 ( 2001 ) henry , j.p . : in `` amiba 2001 : high - z clusters , missing baryons , and cmb polarization '' , in press ( 2002 ) ( astro - ph/0109498 ) rosati , p. , borgani , s. , & norman , c. : araa , in press ( 2002 ) tegmark , m. , hamilton , a.j.s . , xu , y. : mnras , in press ( 2002 ) ( astro - ph/0111575 ) bunn , e.f .
, & white , m. : apj * 480 * , 6 ( 1997 ) lahav , o. , et al .
( the 2dfgrs team ) : mnras , in press ( 2002 ) ( astro - ph/0112162 ) verde , l. , et al .
( the 2dfgrs team ) : mnras , in press ( 2002 ) ( astro - ph/0112161 ) moscardini , l. , et al .
: mnras * 316 * , 283 ( 2000 ) kaiser , n. : apj * 284 * , l9 ( 1984 ) eisenstein , d.j . , hu , p. : apj * 496
* , 605 ( 1998 ) mo , h.j . , &
white s.d.m .
: mnras * 282 * , 347 ( 1996 ) sheth , r.k .
, mo , h.j . , tormen , g. : mnras * 323 * , 1 ( 2001 ) schuecker , p. , et al .
( the reflex team ) : a&a * 368 * , 86 ( 2001 ) hoyle , f. , et al .
: mnras * 329 * , 336 ( 2002 ) guzzo , l. : new astronomy * 2 * , 517 ( 1997 ) carretti , e. , et al . : mnras * 324 * , 1029 ( 2001 ) vogeley , m.s . ,
szalay , a.s . :
apj * 465 * , 34 ( 1996 ) schuecker , p. , guzzo , l. , bhringer , h. , collins , c.a . , : mnras , in press ( 2002a ) schuecker , p. , bhringer , h. , guzzo , l. , collins , c.a .
: in preparation ( 2002b ) szalay , a.s . , et al .
( the sdss collaboration ) : apj , submitted ( 2001 ) ( astro - ph/0107419 ) dodelson , s. , et al .
( the sdss collaboration ) : apj , submitted ( 2001 ) ( astro - ph/0107421 ) miller , c.j . ,
nichol , r.c . ,
batuski , d.j .
: apj 555 , 68 ( 2001 ) peebles , p.j.e . : _ the large - scale structure of the universe _ , ( princeton university press , princeton 1980 ) hawkins , e. , et al . ( the 2dfgrs team ) : in preparation ( 2002 ) zehavi , i. , et al .
( the sdss collaboration ) : apj * 571 * , in press ( 2002 ) ( astro - ph/0106476 ) tucker , d.l . , et al . :
mnras * 285 * , l5 ( 1997 ) baugh , c.m . :
mnras * 280 * , 267 ( 1996 ) collins , c.a . , et al .
( the reflex team ) : mnras * 319 * , 939 ( 2000 ) hamilton , a.j.s . : in _ the evolving universe . selected topics on large - scale structure and on the properties of galaxies _ , ( kluwer , dordrecht 1998 ) , assl series , vol .
231 , p. 185
( also , astro - ph/9708102 ) .
kaiser , n. : mnras * 227 * , 1 ( 1987 ) fisher , k.b .
, et al . :
mnras * 267 * , 927 ( 1994 ) collins , c.a . ,
mnras * 274 * , 1071 ( 1995 ) guzzo , l. et al .
( the reflex team ) : in preparation ( 2002 ) | i review the status of large - scale structure studies based on redshift surveys of galaxies and clusters of galaxies .
in particular , i compare recent results on the power spectrum and two - point correlation correlation function from the 2df and reflex surveys , highlighting the advantage of x - ray clusters in the comparison to cosmological models , given their easy - to - understand mass selection function . unlike for galaxies , this allows the overall normalization of the power spectrum to be measured directly from the data , providing an extra constraint on the models . in the context of cdm models , both the shape and amplitude of the reflex p(k )
require , consistently , a low value for the mean matter density @xmath0 .
this shape is virtually indistinguishable from that of the galaxy power spectrum measured by the 2df survey , simply multiplied by a constant cluster - galaxy bias factor .
this consistency is remarkable for data sets which use different tracers and are very different in terms of selection function and observational biases .
similarly , the knowledge of the power spectrum normalization yields naturally a value @xmath1 for the bias parameter of @xmath2-selected ( as in 2df ) galaxies , also in agreement with independent estimates using higher - order clustering and cmb data . in the final part
, i briefly describe the measurements of the matter density parameter from redshift space distortions in galaxy surveys , and show evidence for similar streaming motions of clusters in the reflex redshift - space correlation function @xmath3 . with no exception ,
this wealth of independent clustering measurements point in a remarkably consistent way towards a low - density cdm universe with @xmath4 . |
we seek to posit a geochemical model for morphogenesis of eukaryotic organisms that includes cellular details .
our aim is to describe the main laws underlying the morphogenetic processes during normal development as well as during possible perturbations resulting in creating structure de novo as in the course of regeneration @xcite . in this inherently multi - scale context
, we must describe the nano - biology of the cell in its environment without unnecessary detail in a manner compatible with macro - biological aspects of the geometrical shape of an organ or entire organism . indeed
, we shall proceed with a series of anstze , one after another , the first group of which are effectively axiomatic mathematical statements about the structure of cells and the organisms which they comprise .
some aspects of this are inspired by biology as we shall discuss in detail , and others are nothing more than rather obvious but new mathematical formalizations of elementary biological structure .
indeed , for the mathematician and biologist reader alike , the depth of certain anstze may range from sublime to ridiculous though likely quite differently for the two fields as we shall comment in specific cases .
the second group of anstze governs the `` morphogenetic field '' , a term used by ren thom @xcite with the concept carefully articulated by the first - named author with misha shubin in @xcite and discussed in some detail here , which determines the time - evolution of cell states .
in fact , our treatment here differs from @xcite insofar as they consider four discrete cell events ( apoptosis , division , movement , growth ) , whereas in our language , the first two are represented by birth / death events ( death alone or death with the birth of two daughters ) , and the latter two by continuous evolution as a path in a certain bundle ( a hilbert space as fiber over a configuration of finitely many distinct points in three - space ) .
one profound contribution of @xcite is what we shall call the `` epigenetic hypothesis '' : _ the distribution of oligo sacchyride residues of glycoconjugates on the cell surface determines the morphogenetic field via an as - yet unknown code .
_ in fact , we are not so absolute as that and relax it here by allowing any other distribution or gradient on the cell surface as well , notably protein or electromagnetic gradients for example .
one nice aspect of the mathematical formalism we have found is that such gradients on the cell surface can be explicitly prescribed in terms of real - valued functions on the two - dimensional sphere .
another notable aspect of the formalism here is the seamless jump in scale from cells to organisms provided by any gromov - hausdorff type metric on measured metric spaces .
indeed , in our description , each cell has its explicit extent in three - space , the union of which forms a metric subspace comprising the organism .
depending upon which features of the organism are to be emphasized , different measures on the cells yield different metrics on the space of organisms .
our precarious multi - scale endeavor is only confounded by the first ansatz : [ ansatz1 ] no statement in biology is always true except this one .
we first describe the geometry of a single cell in mathematical detail ( based on the biologically sound assumption that cells are star - convex ) , then the notion of cell state which captures certain epigenetic information ( and is succinctly described by a collection of real - valued functions on the two - dimensional sphere @xmath0 of unit radius centered at the origin in space @xmath1 owing to star - convexity ) and finally discuss cell trajectory , i.e. , lifetime from birth to division or to final apoptosis ( programmed cell death ) or necrosis ( death from injury or disease ) .
the structure of a live cell includes the _ microtubule organizing center _ ( mtoc ) or _ centrosome _ , from which emanates a network of microtubules forming the essential part of the cytoskeleton of the cell and holding in place the cell surface just as a system of poles holds in place the surface of a circus tent . most animal and some plant mtoc , apart from its main gamma - tubulin content , contain two cylindrical structures called centrioles meeting at a point ; see figure 1 .
let us say that a cell is _ centriolic _ if its centrosome indeed contains such a pair of centrioles , and likewise an organism if its cells are centriolic .
these organisms are especially amenable to our methods and analyses here since each cell @xmath2 has its distinguished _ mtoc point _ @xmath3 given by the point of contact of the two centrioles .
this is a mere convenience though , and one may rather make the choice of an appropriate point within the centrosome or even an arbitrary point of star - convexity in general . even comprehending just the geometry of a single cell within an organism already seems hopeless as suggested by figure 2 .
one has only to peruse for example protist organisms ( which are unicellular eukaryotes ) depicted on the left to see a wild geometric diversity of detailed cell shapes .
there are furthermore anatomical anomalies such as meter - long human neuro skeletal cells as depicted on the right with their dendritic extremities .
exotic examples abound .
nevertheless modulo ansatz [ ansatz1 ] , all cells do share a common geometric feature as follows . recall that in mathematics , one says that a region @xmath4 of space @xmath1 is _ star - convex _ with respect to a point @xmath5 provided that for each point @xmath6
the line segment @xmath7 connecting @xmath8 to @xmath9 also lies in @xmath10 .
= 1.9 in = 2.5 in [ ansatz2 ] a cell @xmath2 is star - convex with respect to some point @xmath11 in its mtoc , and in particular for a centriolic cell , we may take @xmath12 as the point of contact between centrioles .
it is a biological axiom , to the extent that such there be in light of ansatz [ ansatz1 ] , that the cytoskeleton of a cell is star - convex with respect to a point in its mtoc . in our idealization of the geometry
, we may consider the outer surface of the cytoskeleton as a cell surface .
in reality , one would presumably take a small epsilon neighborhood of the star - convex cytoskeleton to describe the cell shape , though many cells are star - convex already with epsilon equal to zero .
indeed , we shall later consider only epsilon neighborhoods of cells anyway , so the geometric idealization of ansatz [ ansatz2 ] from cytoskeleton to cell is unimportant
. fix in space the centrosome of a centriolic cell at some instant in time .
let @xmath13 be the unit vector in the direction along the cylindrical axis for the earlier - replicated centriole and @xmath14 the unit vector in the direction of the later - replicated one , with @xmath15 . in each case
, the direction along the cylindrical axis is determined by the fact that the mtoc point of centriole contact occurs asymmetrically at a preferred starting end of each cylinder .
a centriolic cell thus determines a so - called positively oriented orthonormal 3-frame @xmath16 describing the spatial orientation of each cell .
we shall not in fact require this further extra structure , but it is worth noting .
now , fix a cell @xmath2 with its mtoc point @xmath17 . given a unit vector @xmath18 , where @xmath0 denotes the usual two - dimensional sphere in @xmath1 ( comprised of endpoints of vectors of unit length based at the origin )
, there is a corresponding ray from @xmath8 in the same direction of @xmath19 , and this ray meets the cell surface at a single point at some distance @xmath20 from @xmath8 according to star - convexity . see figure 3 .
this assignment @xmath21 of a positive real - valued function on the two - dimensional sphere @xmath0 completely determines the shape of the cell @xmath2 . to recapitulate this section ,
given a eukaryotic cell @xmath2 , there is an mtoc point @xmath22 , which is well - defined without choice provided @xmath2 is centriolic , with respect to which @xmath2 is star - convex together with a shape function @xmath23 which completely determines the geometrical cell shape @xmath24 where @xmath25 denotes the usual length of a vector @xmath26 in @xmath1 .
there is furthermore a canonical identification of the standard sphere @xmath0 with the cell surface @xmath27 given by the mapping @xmath28 .
the volume @xmath29 of @xmath2 and its surface area @xmath30 are respectively given in terms of a sufficiently regular shape function @xmath31 as @xmath32 ^ 3~ da,\cr a(c)&=\int\int_{{\mathbb s}^2 } \bigl | \nabla \rho_c(\vec\xi)\cdot\vec \xi
\bigr |~da,\cr \endaligned\ ] ] what exactly constitutes a cell state for morphogenesis or any other purpose depends upon the implied internal structure of the cell and must reflect the predictions and interactions of the theory that depend upon it .
one obvious aspect of cell state is the shape function @xmath33 and another is the spatial location of the mtoc point @xmath34 introduced in the previous section . taken together , these determine the natural parametrization @xmath35 of the cell surface @xmath27 .
it follows from the existence of the parametrization that any real - valued function @xmath36 can equivalently be specified by simply pulling back to a real - valued function @xmath37 , i.e. , @xmath38 .
in particular , the density of some substance on the cell surface can be conveniently specified by just such a real - valued function on @xmath0 . to account for area distortion ] cell state
is determined by the shape function @xmath39 together with a collection @xmath40 of cell surface densities @xmath41 , for @xmath42 .
the collection @xmath43 of cell surface densities is called the _ epigenetic spectrum _ of the the cell .
we shall refer to the full @xmath44 tuple @xmath45 including the shape as the _ ( internal ) cell state_. in fact , the cell surface @xmath27 is relatively fluid with sub - regions evolving in time to different shapes within the cell surface .
thus , we should expect the typical distribution of substance on the cell surface to evolve continuously in time . of special interest
are the distributions of oligo conjugates , small oligosaccharide residues of glycoconjugates on the cell surface .
there are up to 12 varieties of monosaccharides and disaccharides in oligosaccharide residues of glycoconjugates ( six of hexose type ) .
there have been numerous indications that these oligosaccharide residues are related to cellular morphogenetic pathways as has been formulated as an epigenetic hypothesis in @xcite and the references therein .
thus , we propose that the cell state @xmath46 of @xmath2 includes both the cell shape @xmath39 and the epigenetic spectrum of cell densities @xmath47 of the corresponding number of coding " oligo conjugates , as per the epigenetic hypothesis , the nature and number of which should be determined experimentally .
here is an important point of contact between one aspect of our theoretical work and experimental determination .
we leave open the possibility of extending this minimal set of functions determining cell state with other molecular compounds or other factors such as protein or electromagnetic gradients defined on the cell surface , where recent works @xcite suggest the latter are central to morphogenesis .
recall @xcite that there is a standard distance between two tuples @xmath48 and @xmath49 of functions @xmath50 given by their
_ @xmath51 distance _
@xmath52 indeed , letting @xmath53 denote the so - called square integrable functions on @xmath0 , the distance @xmath54 gives the collection @xmath55^{n+1}$ ] of all possible cell states @xmath46 the structure of a hilbert space .
hardly worth mentioning mathematically , the next ansatz gives an explicit concept of distance between cell states .
[ a : l^2 ] the natural distance space '' ) more generally taking @xmath56 , for certain specified parameters @xmath57 .
] between cell states @xmath48 and @xmath49 is the @xmath51 distance @xmath58 .
the sperm - fertilized cell oocyte becomes the zygote , the first cell @xmath59 of the organism , which next mitotically divides to ten or more generations in a highly organized paroxysm of cell divisions , no time at all for diffusive or other equilibria to develop , no time for appreciable cell growth or motion and immediately generating all manner of steep gradients across the early embryo .
recall that the centrosome of a cell @xmath2 at interphase replicates to create a pair of centrosomes .
these centrosomes move to opposite ends of the cell and nucleate microtubules to help form the mitotic spindle separating the chromosomes into two sets , one copy for each daughter cell @xmath60 together with one centrosome for each daughter cell .
one centrosome of @xmath61 or @xmath62 is inherited from the centrosome of @xmath2 , which we call the mother centrosome , while the other , which we call the daughter centrosome since it is younger than its mother , has been assembled upon the mother centrosome before mitosis .
insofar as the replicated centrosomes of @xmath2 are distinguishable , so too are its daughters @xmath61 and @xmath62 .
we can thus label each cell in the organism by its phylogeny : a word in the alphabet @xmath63 and @xmath54 , where @xmath63 indicates the daughter @xmath64 that inherits the mother centrosome from @xmath2 and @xmath54 the daughter @xmath65 that inherits the daughter centrosome .
each cell itself is determined by its _ phylogeny _ , a finite word @xmath66 in the two - letter alphabet @xmath67 , including the empty word for the zygote .
this labeling of cells by @xmath63 and @xmath54 is evidently equivalent to the usual structure of embedded phylogenetic tree : a binary tree with gamete for root and with @xmath63 on the right ( say ) and @xmath54 on the left .
[ trajectory ] each cell @xmath2 is born at some instant @xmath68 in time and terminates either through death or cell division at a later instant @xmath69 ; each cell except for the gamete is born with exactly one distinguishable sister . death may be apoptosis , necrosis or it may reflect some crisis for the organism such as amputation or disintegration
. a somatic cell is always born with exactly one sister and terminates either alone ( by death ) or in giving birth to its daughters ( by division ) .
it is worth saying explicitly that ansatz [ trajectory ] is biologically tautological , no more than a self - evident remark but is a necessary aspect of axiomatic mathematical formulation .
indeed , in other important classical mathematical contexts , births as well as deaths can occur only in pairs .
biologically , the ansatz is trivial and can be ignored , for of course it is tautologically so .
vertices of our phylogenetic tree other than the root are thus labeled by the cell event of death or cell division .
edges of the tree correspond to the lifetime of cells , and we will assign as a length of the edge labeled by the cell @xmath2 its lifetime @xmath70 . in this manner , a point in an edge corresponds to the cell @xmath71 at a particular instant @xmath72 in time , where @xmath73 . according to our assumptions , two sister cells @xmath61 and @xmath62 must share their time of birth @xmath74 with the time of death of their mother ( but see section [ sec : gap ] ) . for an extreme example such as a skeletal muscle cell @xmath2 or an eye lens cell @xmath2 which is fully differentiated once and for all time already in the embryo and never divides again , we find @xmath68 at birth and @xmath69 after death of the organism .
at another extreme are dermal cells with their profuse and nearly identical divisions .
the potency of a cell to produce progeny with different types of tissues and organs or simply to differentiate into many possible cell types is called a morphogenetic potency or just a potency . thus in an organism there are totipotent , pluripotent , bipotent and fully differentiated cells .
let us define the _ potency _ @xmath75 of a cell @xmath2 to be zero if it does not divide at all , and if it does divide , say the daughters have respective epigenetic spectra @xmath76^{n}$ ] , then we compare the spectra defining @xmath77 .
thus , the potency of a fully differentiated cell is nearly zero .
in contrast , a bipotent stem cell , for example , can have a large potency if its product cell line differs substantially in epigenetic spectrum from its stem cell line in the @xmath51 distance .
we define an organism to be the collection of its constituent cells lying in space as determined by a language over the two - letter alphabet given by following the centrosome .
the collection of constituent cell states determines a state of the organism at each instant in time .
a convenient mathematical formalism exists for this as a configuration space of distinct points in space which evolve under time either continuously or through specific rules of birth and death at certain event times together with a bundle over this base space with fiber given by an @xmath51 space of functions on the two - dimensional sphere .
all this is quite mathematically routine given what has come before .
at a given moment in time , the cells @xmath2 underlying an organism comprise a metric subspace which moreover can be further imbued with any function @xmath78 chosen to reflect essential morphological or other parameters of @xmath2 . for example , we could take mass density of a cell , density of distribution of ion channels or other cell constituent concentrations .
we may choose particular functions @xmath78 for specific applications a fundamental ansatz is that two organisms each at a given moment in time can be profitably compared using gromov - hausdorff ( gh ) type distances ( cf .
appendix a ) between corresponding measured metric spaces .
this is precisely what allows us to finally pass from the cellular to the capacious scale in our investigations .
it is important to note that the comparison of two organisms " in this framework should be interpreted first of all as the comparison of the morphology of the same organism at different moments in time reflecting the laws of evolution of its shape .
equally , we consider the comparison of the differences between normal and crisis " developmental scenarios ( that is , strong perturbations of proper morphology ) for the same organism at some moment in time . the zygote @xmath79 comes into being at an instant @xmath80 and then undergoes a sequence of binary divisions , the first one occurring at the instant @xmath81 . at each division , there are distinguishable sister cells produced , and the mother cell producing these progeny subsequently ceases to exist .
and so it goes .
the mother of all cells of the organism is the zygote @xmath79 .
for each subsequent offspring produced , i.e. , each cell that did , does or will or ever occur in the organism there is a well - defined word @xmath66 in the two - letter alphabet @xmath67 as discussed before , where the length of the word is the number of generations from the zygote .
not all words occur for a given organism , and the _ phylogenetic language _ of @xmath82 is the collection @xmath83 of words that do in fact occur for the organism @xmath82 .
let @xmath84 be the cell corresponding to the word @xmath85 including the empty word for the zygote .
thus , each cell @xmath84 is born and dies at respective specific times @xmath86 and @xmath87 .
we shall let @xmath88 denote the cell @xmath84 labelled by @xmath66 at time @xmath72 , where @xmath89 with time - dependent cell state @xmath90^{n+1}$ ] , reflecting shape and epigenetic coding , and mtoc point @xmath91 with auxiliary measure @xmath92 .
we have the mathematical definitions of organism and state given by the following biological tautology : an organism @xmath93 is determined by the collection of its constituent cells @xmath94 and its state by their respective states .
let us say that a cell @xmath95 _ exists _ at some time @xmath72 if @xmath89 for some @xmath96 .
it is convenient to suppress the birth / death times of a cell and write simply @xmath97 to signify the cell at a time @xmath72 at which the cell exists and to signify the empty set otherwise . in the same way ,
let @xmath98 denote the cell state and @xmath99 the measure at a time @xmath72 between cell birth and death and to take constant value zero otherwise ; finally , @xmath100 denotes the mtoc point and @xmath101 the shape function if the cell @xmath95 exists , and these are undefined otherwise .
let @xmath102 be the phylogenetic words of the cells of the organism @xmath82 that exist at time @xmath72 .
the organism @xmath103 at a time @xmath72 is the union of these cells , i.e. , the constituent cells of the organism at the instant @xmath72 .
of course , @xmath104 or any epsilon neighborhood of it inherits the structure of a metric subspace of @xmath1 , and each cell comes with its measure @xmath99 which together give the weighted sum @xmath105 of delta functions as a measure on @xmath104 . the _ state _ of the weighted organism @xmath106 at time @xmath72 consists of the cell states @xmath98 of its constituent cells .
the collection of mtoc points @xmath107 of cells that exist at time @xmath72 come in their prescribed lexicographic ordering ( where , say @xmath108 ) which we shall call the _ ( labelled ) configuration _ of the organism at time @xmath72 .
recall that if @xmath109 is a metric space , then a _ ( labelled ) configuration _ in @xmath109 of @xmath110 points is a collection of @xmath111 distinct labelled points in @xmath109 ; see appendix b for the barest thumbnail discussion tailored to our own needs and @xcite for instance for further generalities .
it is biologically clear that in fact the mtoc points @xmath112 do give a configuration in @xmath1 , lying in @xmath104 , i.e. , the mtoc points of distinct cells never coincide . indeed
, we have the stronger : the cells of an organism that exist at each instant in time have disjoint interiors in space @xmath1 . again a biological tautology , this kind of `` steric '' constraint is commonplace in protein theory for instance .
we do not intend to entirely rule out gap or other cell junctions here , rather , let us consider adding that further layer of detail only later .
the number @xmath113 of mtoc points of the organism @xmath104 is piecewise constant and jumps at various times of lone death ( one point disappears ) or birth ( two new points appear and an old one disappears ) of constituent cells according to ansatz [ trajectory ] .
in fact , there may be finitely many coincidentally simultaneous birth or death times , but we make the following finiteness assumption : [ a : wha ] for any fixed @xmath72 , there is a finite sequence of event times @xmath114 so that for each time @xmath115 , for @xmath116 , the configurations @xmath117 are equinumerous and evolve continuously in time .
the discussion of how exactly the cell state before determines the specific cell states after a birth event time is postponed until the more speculative considerations of the next section .
`` the time has come ''
the walrus said , `` to talk of many things : of shoes and ships and sealing - wax of cabbages
and kings and why the sea is boiling hot and whether pigs have wings . '' this is just to say that now we must discuss things a bit more metaphorically in order to absorb the next anstze . as humans
, we can perceive the natural world on various levels .
for instance , in pond water one finds paramecia , which we might probe in the lab _ in vivo _ or _ in vitro _ to discern further structure .
going to an extreme , we might travel to cern and uncover the constituent elementary particles . as human observers
, we do not even want to try to absorb the molecular let alone the atomic or stringy nature of a paramecium .
with what has been called `` willful ignorance '' and also termed `` twice barbarism '' , our vision in morphogenesis had better contain the required cellular and other data even if we shall ultimately and barbarically ignore this level of detail in the large . to handle the need for minute detail in certain biological regards for cells , we have already introduced the notion of cell state .
we must somehow pass from this data as we have formalized it , corpuscular by its very nature , to the context of continuous data of substantial spatial extent .
naturally enough , we might take an epsilon neighborhood of the actual cells at each instant in time , that is , a neighborhood of @xmath104 in @xmath1 .
we furthermore might naturally let epsilon vary over @xmath104 in order to allow for instance for different packing densities of one type of cell compared to another .
an elegant and precise mathematical formalism for this viewpoint is provided by the class of metrics on measured metric spaces inspired by gh distance . for completeness ,
one such distance ( the sturm @xmath51 distance @xcite ) is explicitly defined in appendix a. [ ans : gh ] given two organisms @xmath82 and @xmath118 with their respective measures @xmath119 and @xmath120 , gromov - hausdorff type distances between the measured metric spaces @xmath121 and @xmath122 such as the sturm @xmath51 distance capture the geometry of organism morphology .
just as the sturm @xmath51 distance is one among many sensible choices for comparing measured metric spaces , so too , we have the flexibility to take different measures @xmath78 on cells @xmath2 . as an example
, the mass density can be one special case of measures involved in building the morphology of an organism . presumably as a human observer employing it as we have done here
is like observing the world with those x - ray glasses that are bought by gullible children ; we would detect density differences between cells within organisms when we compare them wearing our mass density glasses .
a tadpole and a frog would be easily distinguished by their bone structures for instance .
the formalism we have developed is robust enough to handle general examples exchanging x - ray for other glasses by changing the measure @xmath123 on the fixed metric space that is the organism @xmath104 . during mitotic division of the cell @xmath2 , the two daughter cells @xmath64 and @xmath65 labelled by centrosomes as
before are separated by a so - called mitotic plane passing through the interior of @xmath2 often asymmetrically .
this separating plane divides the cell surface @xmath27 into two inequivalent regions @xmath124 forming the cell boundary of respective daughters . [ a : mbe ] the epigenetic spectra of daughters @xmath64 and @xmath65 are nearly inherited in the @xmath51 sense on @xmath0 by inclusion of the cell surfaces @xmath125 from the mother . on the complementary regions to @xmath126 and @xmath127 in daughter cell surfaces
, there are as - yet unspecified rules of how to extend to a function on the entire cell surface .
this ansatz partly describes the _ morphogenetic birth effect _ , that is , the function that determines the cell states of @xmath128 and @xmath129 at the moment @xmath130 when mitotis is complete from the cell state of @xmath131 of a mother cell as mitosis begins at @xmath132 .
it is not that the biology breaks down during the interval between times @xmath132 and @xmath130 .
after all , there is a sensible moment of birth when the cell membranes of the daughters finally separate that we could also take to be the moment of death of the mother @xmath2 .
the surface concentrations in @xmath133 of the oligo conjugates or other compounds are everywhere defined until this moment and after become the surface concentrations in @xmath134 and @xmath135 . in this sense , again the ansatz is disappointingly tautological biologically .
however , this scenario does _ not hold exactly _ and in our barbaric and willful ignorance , we posit in ansatz [ a : mbe ] a precise sense in which @xmath51 distance suitably measures deviation from this idealized biology .
to recapitulate , given a eukaryotic organism @xmath82 and a time @xmath72 , we have the constituent cells @xmath136 that exist at time @xmath72 indexed by a finite set @xmath137 of words in the alphabet @xmath67 .
the union of these star - convex regions is by definition the organism @xmath138 at time @xmath72 as a metric subspace of @xmath1 .
this region comes equipped with a measure @xmath123 , and it is the measured metric spaces @xmath106 that reflect the large - scale geometry of the organism .
standard methods , called gh type metrics , can be applied to these spaces to quantify the sense in which two organisms have the same shape .
nevertheless , the organism has a complicated state , namely , we take the state of the organism to be the collection of cell states of each of its constituent cells , i.e. , cell states @xmath139 and mtoc points @xmath140 , for @xmath141 on the cellular scale , we have organized things in appendix b into a vector bundle @xmath142 over the space @xmath143 of configurations of finitely many points in @xmath1 , where the fiber over a configuration with @xmath111 points is an @xmath111-tuple of elements in @xmath144^{n+1}$ ] , namely , the cell states of the constituent cells .
thus , the entire organism in its entire state is described at an instant in time as a point in the total space of this bundle over the finite configuration of mtoc points in the base , and the measure @xmath119 is an auxiliary function defined on the base .
the idea from @xcite is that the `` morphogenetic field '' is a function defined on some bundle of states that governs the time evolution of an organism starting from ovum .
their formalism was different with discrete time cell events as was already mentioned . here
we imagine continuous evolution in some @xmath145 with jumps in @xmath111 at finitely many event times compatible with ansatz [ a : wha ] .
the morphogenetic field of @xcite should be realized here as some sort of functional which governs the time evolution of an organism , namely , as a lagrangian action : [ a : lag ] there is some lagrangian formulation of organism evolution , i.e. , an action functional on paths in @xmath146 whose stationary paths are the time trajectories of organisms .
it is important to emphasize that the lagrangian formulation of ansatz 4.3 is different from that of the underlying physics : our action should be formulated on the space of organism states as we have defined them . using our definition of potency ( see the end of section [ sec : life ] ) and the observation that each organism starts to differentiate from the zygote having a huge potency towards its later less - potent progeny , we posit : the average _ potency _ @xmath75 over the set of all cells @xmath2 in an organism is strictly decreasing in generation during embryo development .
this is of course in keeping with the second law of thermodynamics .
[ a : sis ] nearly identical sisters almost always remain nearly in contact .
this is interesting as it relates the phylogeny to the geometry . of course , sisters are never truly identical , but if they were , then the morphogenetic field could not distinguish them , so of course they would evolve in the same manner .
nature is nearly like this as an essentially deterministic physical system .
this seems to give a nice explanation for organ formation and other aspects as well and really makes good sense . again , with the @xmath51 distance between cell states , we can conveniently quantify differences between sisters or indeed any pair of cells viz ansatz [ a : sis ] the morphogenetic field is locally supported in the sense that for any cell @xmath95 that exists , there is a neighborhood @xmath147 of it in @xmath1 so that the forward time evolution of @xmath95 for some fixed interval of time depends only upon the states of the cells in @xmath148 . in particular
, phylogeny does not matter _ a priori _ , which implies that there is indeed a physical determinism underlying the morphological determinism we are studying . and this determinism is governed by the laws formulated as a morphogenetic field concept .
[ a : stable ] for any sub - organism @xmath149 in any reasonable state there is a stable ideal outcome organism @xmath150 so that the lagrangian action of an arbtrary path from @xmath118 to a stable outcome @xmath151 is given by a gromov - hausdorff type distance between @xmath150 and @xmath151 .
this gh - deviation from ideal outcome is admittedly a bold hypothesis for the dynamics providing thereby the key mechanism driving morphogenesis ranging from embryo development to the creation of proper morphology during the regeneration processes in the event of crisis .
the ansatz is interesting on several levels first of all in its postulate that there are stable ideal outcome organisms at all under admittedly unspecified reasonable conditions .
the comparison of stable outcomes using gh techniques is consistent with the earlier ansatz [ ans : gh ] .
indeed , perhaps another more traditional differential calculus type expression of action in terms of the gh - metric could be possible rather than relying on stability of outcomes .
secondly , it ties the cellular scale of the morphogenetic field with the large - scale geometry of organisms through the action , and it evidently implies the earlier ansatz [ a : lag ] in its specificity .
this appendix gives a precise example of the sort of distance between measured metric spaces that we propose for use in comparing organisms . for completeness
, definitions are given from first principles .
[ def : metric]if @xmath109 is a set , then a function @xmath152 is a _
metric _ if and only if for all @xmath153 , we have : [ symmetric ] @xmath154 ; [ triangle inequality ] @xmath155 ; [ positive ] @xmath156 and @xmath157 if and
only if @xmath158 . given metric spaces @xmath159 and @xmath160 , let @xmath161 denote the set of all possible metrics on the disjoint union @xmath162 of @xmath109 and @xmath163 extending @xmath164 on @xmath109 and @xmath165 on @xmath163 , that is , @xmath54 satisfies definition [ def : metric ] on @xmath162 and @xmath166 for all @xmath167 and @xmath168 , for @xmath169 . [
d : bpm ] a _ borel probability measure _
@xmath119 on a space @xmath109 is the specification of a real number @xmath170 to each set @xmath30 in the sigma algebra is a _
sigma algebra _ provided it is closed under countably many applications of the set operations complement , union and intersection . ]
@xmath171 of subsets generated by the open sets of @xmath109 satisfying the following properties : [ sigma additive ] if @xmath172 are pairwise disjoint subsets for @xmath173 with @xmath174 countable , then @xmath175 ; [ non - negative ] @xmath176 for all @xmath177 ; [ null empty set ] @xmath178 .
[ probability measure ] @xmath179 . given two measured metric spaces @xmath180 and @xmath181 ,
let @xmath182 denote the collection of all borel probability measures @xmath119 on @xmath183 extending @xmath184 on @xmath109 and @xmath185 on @xmath163 , that is , @xmath119 satisfies definition [ d : bpm ] on @xmath186 and @xmath187 for all @xmath177 , @xmath188 .
one collection of distances for measured metric spaces inspired by the gh metric is given by : the _ sturm @xmath189-distance _
@xcite between measured metric spaces @xmath180 and @xmath181 is defined to be @xmath190^p \mu(dx , dy)\biggr ) ^{1\over p},\ ] ] where the infimum is over all @xmath191 and @xmath192 , and @xmath193 is some fixed natural parameter .
we can take this sturm distance , say with @xmath194 for definiteness , as just one of many possibilities .
further gh type metrics on collections of measured metric spaces as well as discussions of their computational implementation in practice for digital shape comparison can be found in the very accessible @xcite .
as in the text , we have : if @xmath109 is a metric space then a _ ( labelled ) configuration _ in @xmath109 of @xmath110 points is a collection of @xmath111 distinct labelled points in @xmath109 .
let @xmath195 where the exponent @xmath196 denotes the @xmath111-fold cartesian product .
let @xmath197 denote the disjoint union .
insofar as @xmath198 is a subset of the cartesian product @xmath196 , it inherits a natural subspace topology .
for example , @xmath199 while a configuration on two points is the complement of the diagonal in the cartesian square @xmath200 .
there is a nice theory of analysis and measure on configuration spaces satisfying appropriate boundedness conditions , cf .
@xcite . for our purposes here , just the underlying topological spaces and vector bundles over arbitrary configurations are enough .
again , as a subset of the cartesian product @xmath203 , the new space @xmath204 inherits its natural metric topology , and the forgetful map @xmath205 has the natural structure of a vector bundle with fiber @xmath206 for each @xmath110 .
we informally refer to @xmath207 itself as a @xmath29-bundle over @xmath208 .
s. albeverio , yu . g. kondratiev , m. rckner .
analysis and geometry on configuration spaces .
154 * ( 1998 ) , 444 - 500 ; analysis and geometry on configuration spaces : the gibbsian case
157 * ( 1998 ) , 242 - 291 . | we introduce a formalism for the geometry of eukaryotic cells and organisms .
cells are taken to be star - convex with good biological reason .
this allows for a convenient description of their extent in space as well as all manner of cell surface gradients .
we assume that a spectrum of such cell surface markers determines an epigenetic code for organism shape .
the union of cells in space at a moment in time is by definition the organism taken as a metric subspace of euclidean space , which can be further equipped with an arbitrary measure .
each cell determines a point in space thus assigning a finite configuration of distinct points in space to an organism , and a bundle over this configuration space is introduced with fiber a hilbert space recording specific epigenetic data . on this bundle , a lagrangian formulation of morphogenetic dynamics is proposed based on gromov - hausdorff distance which at once describes both embryo development and regenerative growth . |
the r - process nucleosynthesis is called for to explain the origin of about half the elements heavier than iron observed in nature .
its astrophysical origin remains a mystery .
the r - process is one of the most complex nucleosynthesis process to explore because of the numerous difficulties still affecting the description of both the explosive astrophysical conditions believed to host the process and the nuclear properties of the exotic neutron - rich nuclei involved . from the nuclear physics point of view
, the major difficulty lies in the determination of nuclear data for the thousands of nuclei far from the @xmath0-stability , for which essentially no experimental data exist nowadays .
these concern mainly nuclear structure properties , @xmath0-decays , neutron captures , photodisintegrations as well as fission processes .
in particular , mass predictions for neutron - rich nuclei play a key role since they affect all the nuclear quantities of relevance in the r - process , the @xmath0-decay , neutron capture and photodisintegration rates , as well as the fission probabilities .
future ri - beam facilities , which are now under construction or planning , place their first priority to measure masses and half - lives of neutron - rich nuclei which have not been observed yet and are relevant to the r - process studies . in the coming experiments , it is clearly meaningful to measure such masses and half - lives of unknown neutron - rich nuclei .
in addition , we emphasize that information on their experimental errors is crucial to promote theoretical studies on mass and @xmath0-decay half - life .
the present paper aims at guiding such future experiment in defining _ how far _ from the stability line and _ how much precisely _ these physical quantities should be measured .
different approaches can be followed to answer such questions . however , one major fact that should be kept in mind is that the r - process astrophysical site remains totally unknown to date .
although the solar system signature clearly shows that the nuclear mechanisms responsible for the production of r - process nuclei concern neutron captures and beta - decay in the exotic neutron - rich region , no astrophysics model can nowadays consistently predict the neutron densities required for a successful r - process .
the `` hot bubble '' scenario or the postexplosion outflows expected from protoneutron stars in the seconds after successful core - collapse supernovae are thought to be a likely candidate site for the r - process ( e.g. , meyer 1992 ; woosley 1994 ) .
however , recent models of spherical `` neutrino - driven winds '' from protoneutron stars ( e.g. takahashi 1994 ; thompson 2001 ) fail to produce robust r - process nucleosynthesis up to and beyond the third ( @xmath5 ) r - process peak for `` canonical '' neutron stars with @xmath6 = 1.4 @xmath7 and @xmath8 = 10 km .
the other proposed sites include such scenarios as `` neutron star mergers '' ( freiburghaus 1999 ) , weak r - process by the shock processing of the helium and/or carbon shells of core - collapse supernovae ( truran & cowan 2000 ) , magnetic protoneutron star winds ( thompson 2003 ) , prompt explosions from collapsing o - ne - mg cores ( wanajo 2004 ) , or even interestingly , some settings with rapid ejection of high - entropy but nearly symmetric matter to produce the r - process nuclei without excess neutrons ( meyer 2002 ) . each of them , however , faces severe problems and can not at the present time explain the production and galactic enrichment of the r - process nuclei observed in the universe .
moreover , recent observational studies ( e.g. sneden 2000 ) of the relative abundance pattern of the r - process elements in very metal poor stars and also analysis ( e.g. , wasserburg , busso & gallino 1996 ) based on the isotopic abundances for the early solar system measured in meteorites have suggested that different r - process sites are responsible for the lighter ( @xmath9 135 - 140 ) and heavier ( @xmath9 135 - 140 ) r - process nuclei .
this makes the determination of the physical characteristics for the r - process environment further complicated . for the above reason
, it remains extremely difficult to estimate the precision required for mass and @xmath0-decay half - life measurements on the basis of r - process abundance calculations . in order to answer the objective questions treated in the present paper , we have therefore chosen to consider criteria independently of any `` realistic '' astrophysics calculations .
even when the future experiments are performed , it is clear that theoretical predictions will still have to fill the experimental gaps for the thousands of nuclear data required in r - process simulations . in the first step
, these future measurements will therefore mainly help in improving the theoretical models by constraining them further on nuclei closer to the one involved by the r - process , or even directly involved .
they might bring new insights on nuclear physics phenomena at large neutron excesses as well as improve the present parametrizations of mass formulas .
although most of the recent mass formulas show fits to experimental masses of similar quality , the mass extrapolations far from the valley of @xmath0-stability can differ from each other quite significantly ( for a recent review , see lunney 2003 ) .
we have therefore chosen to estimate the nuclei to be involved and the required precision of future measurements by considering arguments on simple astrophysics considerations and existing nuclear model predictions as explained below .
when the future experiments supply with information on new masses and half - lives with a reasonable precision , model predictions will tend to converge if their parameters are updated to fit the newly measured masses .
in this regard , mass formula studies would not benefit if the experimental errors do not resolve the differences between the model predictions for the most exotic neutron - rich nuclei accessible ( or ideally directly involved in the r - process nuclear flow ) . accordingly , as a first rough guide for the required precision in detector developments , we put a rather simple requirement to mass and half - life measurements as follows : experimental errors subsidiary to the r - process nuclei need to be less than _ half _ the difference between the masses ( or half - lives ) predicted by the different nuclear models .
we stress that the total length of the error bar obtained in such a procedure corresponds to @xmath10 . in the following ,
we discuss such a criterion on neutron - rich nuclei at the @xmath3=50 and @xmath3=82 shell closures .
these regions are expected to become accessible in near - future experiments and are known to be of first importance in the development of nuclear structure studies , as well as in our understanding of the r - process nucleosynthesis .
we consider here three mass formulas , known as hfb-2 ( goriely 2002 ; samyn 2001 ) , frdm ( @xcite ) , and kuty ( @xcite ) , available for a wide - range use in the nuclear chart and hence at this moment appropriate for r - process abundance calculations .
the three mass formulas predict the 2135 measured masses with a root - mean - square deviation of about @xmath11680 kev ( see @xcite ) , although they were derived from quite different leading principles .
the hfb-2 model is taken as representative of the microscopic mass formulas recently derived within the hartree - fock - bogoliubov framework based on an effective nuclear force of the skyrme type .
on the other hand , the kuty mass formula corresponds to a semi - empirical approach making use of an empirical gross term for the macroscopic properties of spherical nuclei and spherically - based shell terms for the microscopic corrections . here
the deviation from the gross properties is explained microscopically as shell and deformation effects .
the use of a large number of parameters to describe the single - particle potential and nuclear gross properties enables the kuty model to reproduce relatively well all experimentally known masses as well as the single - particle energy levels .
the frdm model is also of the semi - empirical type and was derived from the finite - range droplet model for the macroscopic part , and from a deformed single - particle potential for the microscopic part .
-stability valley of its neutron - rich side in the n=82 plane .
the values of mass excess from three theoretical mass formulas are shown against proton number , z. the open circles connected with the dashed line depict the mass excess from hfb-2 , the triangles connected with the solid line from kuty , and the crosses from frdm .
the origin ( z=54 , n=82 ) is the stable @xmath12xe while the edge of the abscissa corresponds to the hfb-2 neutron drip line .
( b ) predicted mass differences are plotted against z for the n=82 isotones .
the filled circles show the mass difference between hfb-2 and kuty , and the crosses between frdm and kuty .
the scale of the abscissa is the same as in ( a ) .
, title="fig : " ] -stability valley of its neutron - rich side in the n=82 plane .
the values of mass excess from three theoretical mass formulas are shown against proton number , z. the open circles connected with the dashed line depict the mass excess from hfb-2 , the triangles connected with the solid line from kuty , and the crosses from frdm .
the origin ( z=54 , n=82 ) is the stable @xmath12xe while the edge of the abscissa corresponds to the hfb-2 neutron drip line .
( b ) predicted mass differences are plotted against z for the n=82 isotones .
the filled circles show the mass difference between hfb-2 and kuty , and the crosses between frdm and kuty .
the scale of the abscissa is the same as in ( a ) .
, title="fig : " ] in fig .
1(a ) , we illustrate a global feature of the @xmath0-stability valley given by the three above - mentioned models . in the figure
a sectional diagram of the neutron - rich side of the valley is shown for the n=82 isotones .
the microscopic mass formula is seen to give a steeper slope of the @xmath0-stability valley and hence predicts larger masses compared with those predicted from semi - empirical formulas .
this can be seen in fig .
1(a ) especially for low z , at large neutron excesses . figure 1(b ) depicts the mass differences between hfb-2 and kuty and between frdm and kuty for the n=82 isotones .
in particular , it is seen that the mass difference between hfb-2 and kuty is prominent and increases with decreasing proton numbers , when approaching the neutron - drip line .
namely , both models predict significantly different masses for the neutron - rich nuclei far from the stability line .
when applied to r - process calculations , such mass differences inevitably lead to different r - abundance patterns ( motizuki 2004 ; wanajo 2004 ) . in the following ,
we will focus on the two mass formulas , hfb-2 and kuty , for which the mass differences are seen to be the most significant ones .
the r - process is believed to reach the neutron richness of @xmath13 in dynamical simulations ( e.g. motizuki 2004 ) as well as in the simple parametrized site - independent model ( e.g. goriely & arnould 1996 ) .
for example , within the canonical model prediction making use of the kuty masses and the half - lives calculated by the second version of the gross theory ( see below ) , the r - process path determines @xmath14nb as the polestar neutron - richest nuclide among the n=82 isotones .
this result is obtained assuming the so - called waiting - point approximation at a temperature of @xmath15 k and a neutron density of @xmath16 in order to reproduce the location and width of the @xmath17 130 peak observed in the solar r - process abundance distribution .
the nuclide @xmath14nb is characterized by a ratio of @xmath18 .
we accordingly analyze the mass differences between the kuty and hfb-2 models for the @xmath3=82 isotones down to @xmath14nb . as seen in fig .
2(a ) , the mass difference at @xmath14nb is about 3 mev . as mentioned in sect . 1 ,
the required total error bar @xmath10 is set to _ half _ the mass difference , so that the detectors for mass measurements must have a precision of 1@xmath1 250 kev at the neutron richness of @xmath2 = 3.0 .
= 82 isotones is plotted against @xmath19 .
the vertical dashed line indicates @xmath20 , down to which the r - process is believed to be extended .
( b ) beta - decay half - life difference between the values calculated by the second version of the gross theory ( gt-2 ) with hfb-2 masses and with kuty for the odd - z n=82 nuclei .
both in ( a ) and ( b ) , the total length of the error bars are indicated so as to satisfy the requirement explained in sect .
1.,title="fig : " ] = 82 isotones is plotted against @xmath19 .
the vertical dashed line indicates @xmath20 , down to which the r - process is believed to be extended .
( b ) beta - decay half - life difference between the values calculated by the second version of the gross theory ( gt-2 ) with hfb-2 masses and with kuty for the odd - z n=82 nuclei .
both in ( a ) and ( b ) , the total length of the error bars are indicated so as to satisfy the requirement explained in sect .
1.,title="fig : " ] similarly , we calculate the @xmath0-decay half - life of the n=82 isotones , and more particularly of @xmath14nb , within different approaches to estimate the precision required in half - life measurements .
one of the widely used models used for astrophysics applications is the second version of the gross theory , known as gt-2 ( @xcite ) .
we apply this model using both the hfb-2 and kuty @xmath21 predictions .
figure 2(b ) shows the half - life difference between both models for the odd - z n=82 nuclei .
we observe that the half - lives calculated from the microscopic mass formula are shorter than those from the semi - empirical mass formula .
this can be understood by the steeper slope of the @xmath0-stability valley for the microscopic mass formula : the hfb-2 model leads to essentially larger @xmath21-values than the kuty model .
in particular , for @xmath14nb , the half - life is 3 ms for gt-2 with hfb-2 masses and 5 ms with kuty .
however , the uncertainties in @xmath0-decay predictions stem not only from mass predictions , but also from the theoretical model used to describe the weak interaction .
mean field and shell models have been applied in recent years to the calculation of the @xmath0-decay rates of nuclei of astrophysics interest . in the particular case of @xmath14nb
, these models predict a half - life of about 4 ms for the df3 density functional plus continuum qrpa approximation of borzov ( 2003 ) including only the allowed transitions and about 3 ms if the first forbidden transitions are also included .
a shorter half - life of about 2 ms is obtained by the shell model of martinez - pinedo & langanke ( 1999 ) .
considering such half - life differences for @xmath14nb , we find that 1@xmath1 0.15 ms at @xmath18 is required for the half - life measurements .
a similar procedure can be followed in the @xmath3=50 region .
we find that similar precisions ( 1@xmath1 250 kev for masses and 1@xmath1 0.15 ms for half - lives ) are required from gt-2 calculations at @xmath2 = 2.9 on the @xmath3=50 shell closure , , for @xmath22fe .
however , if we consider the doubly magic nuclide @xmath4ni which has been observed but for which the mass and the half - life remains experimentally unknown , the same criterion leads us to a precision of 1@xmath1 300 kev for mass and of 1@xmath1 5 ms for half - life measurements .
we have derived the required precision of 1@xmath1 250 kev and 1@xmath1 0.15 ms , respectively , for mass and half - life measurements at the neutron richness of @xmath2 = 3.0 at the @xmath3=82 shell closure and at the @xmath2 = 2.9 at the @xmath3=50 shell closure . for the doubly magic nuclide @xmath4ni
, we have found that the detectors must have a precision of 1@xmath1 300 kev for mass and of 1@xmath1 5 ms for half - life measurements . note that not only statistical but also systematic errors should be included in the above discussion .
it should also be kept in mind that the precision estimate presented here is based on simple arguments due to our ignorance of the astrophysical site for the r - process .
future development in nucleosynthesis models ( takahashi , this volume ) will hopefully bring new insight on the nuclear flow followed by the r - process and consequently on the nuclei involved and the major nuclear quantities of relevance .
experiments at riken ri - beam factory will start in 2007 . here
ri - beams are planned to be produced by fragmentation and uranium fission methods .
the intensity of the ri - beams will be strong enough to reach @xmath4ni ( @xmath2=2.8 ) and @xmath22fe ( @xmath2=2.9 ) at the @xmath23 shell closure to measure these masses and half - lives with the suggested precisions .
however , the expectations of the ri - beam intensity created with the _ fragmentation _ method at present come down to one particle per @xmath24 sec at the @xmath2=3.0 ( @xmath14nb ) region at the @xmath25 shell closure .
this means that the measurements with the required precisions might be difficult for the present technology : it is indispensable to contrive new type of detectors to overcome this difficulty .
future measurements with better precision are strongly encouraged in order to develop theories of nuclear masses and half - lives .
progress in these theories and above all in microscopic approaches , as well as further developments of astrophysics models will help us to solve the long - standing mystery that the r - process nucleosynthesis still represents .
we would like to thank k. takahashi for useful comments .
y.m . would like to acknowledge y. ishida , t. suda , and y. yano for information on experimental status at the ri - beam factory .
s.g . is fnrs research associate . | in order to understand the r - process nucleosynthesis , we suggest precision required for mass and @xmath0-decay half - life measurements planned at future ri - beam facilities . to satisfy a simple requirement that we put on nuclear model predictions ,
it is concluded that the detectors for the mass measurements must have a precision of 1@xmath1 250 kev , and that the detectors for the half - life measurements demand a precision of 1@xmath1 0.15 ms .
both the above precisions are required at the neutron richness of @xmath2 = 3.0 at the @xmath3=82 shell closure and @xmath2 = 2.9 at the @xmath3=50 shell closure . for the doubly magic nuclide @xmath4ni ,
a precision of 1@xmath1 300 kev and 1@xmath1 5 ms are required , respectively , for mass and half - life measurements .
this analysis aims to provide a first rough guide for ongoing detector developments . |
cataclysmic variables ( cv ) are short period binary systems consisting of a white dwarf that accretes matter via roche - lobe overflow from a low - mass companion star . these objects exhibit a wide range of phenomenology including optical flickering in nova - like systems , dwarf nova eruptions which are thought to be caused by thermal instabilities in the accretion disks , and classical nova explosions which are thermonuclear runaways of the accreted matter on the white dwarf ( see , e.g. , warner 1995 ) .
the range of observed phenomena depends on the mass transfer rate , the mass ratio of the stellar components , and the magnetic field strength of the accreting white dwarf .
the orbital periods of the majority of cvs range from 8 hours down to about 78 minutes , but both longer and shorter period systems are known . in the former case ,
the donor stars are typically somewhat evolved , while in the latter case , the donor stars are hydrogen exhausted . in this paper
we focus on the gap that exists in the orbital period distribution of cvs in the range of @xmath27 hr ( see , e.g. , warner 1976 ; rappaport , verbunt , & joss 1983 , hereafter rvj ; spruit & ritter 1983 ; hameury , et al . 1988a ; warner 1995 ) .
the overall evolution of cv binaries is thought to be fairly well understood .
the widely accepted explanation for the period gap rests on a mechanism for extracting angular momentum from the binary orbit ( e.g. , via magnetic braking of the secondary ) for periods down to @xmath7 hours , followed by a relatively substantial decrease in the angular momentum loss rate .
the donor star , which had been thermally `` bloated '' in response to the mass loss driven by the systemic angular momentum losses , is then able to relax inside of its roche lobe and mass transfer ceases . the donor star is then thought to reestablish roche - lobe contact by the time the orbital period has decreased to about 2 hr , after which mass transfer resumes . in this paper
we critically examine this paradigm for the creation of the period gap . while most workers believe in the existence of the so - called 2 - 3 hr period gap " , a few ( e.g. , wickramasinghe & wu 1994 ; verbunt 1997 , but see also warner 1995 ; wheatley 1995 ) have questioned its reality , especially when all types of cv are considered ; however , we adopt the view that the period gap is a real feature of the cv population as a whole and , as such , requires a theoretical explanation ( with observational tests ) within the context of their binary evolution . finally in this
regard we note a suggestion by clemens et al .
( 1998 ) that the period gap results from a `` kink '' in the radius - mass relation for main - sequence stars at a mass of about @xmath28 ( but see the rebuttal by kolb , king , & ritter 1998 ) .
in 2 we describe the conventional picture of the evolution of a typical cv , including the period gap , and show some illustrative examples of binary evolution calculations for individual systems . in
3 we explore how the binary evolution alters the relations among mass , radius , and orbital period of the secondary star .
specifically we discuss how the main - sequence radius - mass relationship must be modified to include the addition of a `` bloating factor '' that accounts for the changes caused by departures from thermal equilibrium of the mass - losing secondary star .
we derive semi - analytic mass - period and radius - period relationships for cv secondaries . in
4 we describe our population synthesis and binary evolution codes , while in 5 we present results from our population synthesis study of cvs in which the binary parameters of the cvs at all phases of their evolution are explored . in 6
we show how assumptions that the donor star has a main - sequence radius - mass relation can lead to large errors in the assignment of the constituent stellar masses , most notably within the orbital period range of 3 - 5 hr .
this period range should encompass the maximum bloating exhibited by a cv secondary compared to a main - sequence star of the same mass .
also in 6 we discuss some specific observational implications resulting from our theoretical work .
in particular , a specific test for cvs just above the period gap which will enable us , in principle , to distinguish unambiguously among different possible explanations for the period gap is presented .
finally , we present our summary and conclusions in 7 .
in the conventional picture of cv evolution ( see , e.g. , faulkner 1971 ; paczyski & sienkiewicz 1981 ; rappaport , joss , & webbink 1982 , hereafter rjw ; rvj ; spruit & ritter 1983 ; hameury et al . 1988a ; kolb 1993 ) , the early phases are expected to be dominated by angular momentum losses due to magnetic braking via a magnetically constrained stellar wind from the donor star ( see , e.g. , verbunt & zwaan 1981 ; rvj ) . in these early phases , mass transfer rates
are typically @xmath29 to @xmath30 , and orbital periods range from @xmath31 8 hr to @xmath31 3 hr , just at the upper edge of the period gap . at some point in the evolution , the secondary becomes completely convective ( at @xmath32 ) and , in the currently accepted view , magnetic braking is assumed to be greatly reduced .
the near cessation of magnetic braking reduces the mass transfer rate , and allows the secondary to shrink toward its thermal equilibrium radius .
this causes a period of detachment , during which @xmath17 drops to essentially zero , and which lasts until the roche lobe shrinks sufficiently to bring the secondary back into contact with it , at an orbital period of @xmath33 hr .
this is the commonly accepted explanation for the observed period gap between 2 - 3 hrs in cvs ( rvj ; spruit & ritter 1983 ) . when mass transfer recommences at @xmath1 @xmath31 2 hrs , it is then driven largely by gravitational radiation losses at rates of @xmath34 . as the orbit shrinks and the mass of the donor star decreases , the mass - loss timescale increases , but the thermal timescale , @xmath35 , increases much faster , due to its approximate @xmath36 dependence
therefore , at some point the thermal timescale grows larger than the mass - transfer timescale .
when this occurs , the donor star is unable to adjust to the mass loss on its thermal timescale , and it therefore starts to expand upon further mass loss , in accordance with its adiabatic response ; i.e. , @xmath37_{ad } < 0 $ ] . the orbital period at this point is typically @xmath38 min and the mass of the donor star is @xmath39 . from this point on
, the mass of the donor star will continue to decrease ( but with longer and longer @xmath17 timescales ) , the orbital period will increase back toward periods approaching @xmath312 hr ( within a hubble time ) , and the interior of the donor star will become increasingly electron degenerate . a discussion of this later stage of cv evolution is presented by howell , rappaport , & politano 1997 ; hereafter hrp ) . to make these evolutionary descriptions somewhat more quantitative ,
we show in figure 1 the secular evolution of several model cvs under the influence of magnetic braking and gravitational radiation . the evolution code used to generate these results
is a descendant of the one used by rvj , and is described in 4.2 along with recent improvements to the code .
the two panels on the left side of fig .
1 show the evolution with time of a cv binary with initial constituent masses of @xmath40 and @xmath41 , where @xmath19 and @xmath42 are the masses of the donor star and white dwarf , respectively .
other parameters used in the calculation are for our standard model " ( see table 1 for definitions ) .
the top and bottom panels show the evolution with time of the mass transfer rate and orbital period , respectively , for an assumed donor star with solar composition .
the calculations have been carried out to approximately the age of the galaxy .
the evolutionary phases and features discussed above are present in fig . 1 , including the interval where mass transfer is driven by magnetic braking ( @xmath43 yr ) , the period gap ( @xmath44 yr ) , the interval where @xmath17 is driven by gravitational radiation losses ( @xmath45 yr ) , the period minimum at @xmath46 yr , the subsequent increase in @xmath1 back up to @xmath33 hrs , and the sharp falloff in @xmath17 after orbital period minimum . on the right side of fig .
1 the temporal evolution of four other illustrative model cv binaries are shown .
the following discussion contains descriptions of the period gap which develops in these systems ; these are easier to visualize by looking also at fig .
2 . the initial masses ( @xmath47 ) of these systems are ( 0.2 , 0.4 ) , ( 0.35 , 0.35 ) , ( 0.3 , 0.6 ) , and ( 0.65 , 0.7 ) , all in units of @xmath48 . for the system with initial masses ( 0.2 , 0.4 ) the binary comes into roche - lobe contact for the first time at an orbital period below the gap , i.e. , at @xmath49 hr ( solid curve in figs
. 1 c and d ) .
note the enhanced mass transfer rate at @xmath31 30 myr after roche - lobe contact is made .
the subsequent evolution is not dissimilar to the one shown in the left panels . for the system with initial masses ( 0.35 , 0.35 )
, the donor star commences mass transfer at a period of 3.3 hr , with magnetic braking still operative ( dotted curve in figs
. 1 c and d ) . because the two masses are the same when the donor star first fills its roche lobe , the mass transfer is only marginally stable ( see the discussion in 4.2 below ) .
therefore , @xmath17 is initially very high and the system is quickly driven out of thermal equilibrium , causing the orbit to expand .
this system comes out of contact ( i.e. , enters the period gap ) at an orbital period of 3.5 hr .
the system with initial masses ( 0.3 , 0.6 ) is an example of one that commences mass transfer _ in _ the period gap .
lastly , the system with initial masses ( 0.65 , 0.7 ) is another example of a system which exhibits the
usual " 2 - 3 hr period gap , but commences mass transfer at @xmath50 hr . in figure 2 the same evolutions shown in fig .
1 are again presented , but this time the binary parameters are displayed as a function of the evolving orbital period . as in fig .
1 , the left panels are for initial masses ( @xmath51 ) of ( 0.9 , 1.1 ) , while the right panels are for initial masses of ( 0.2 , 0.4 ) , ( 0.35 , 0.35 ) , ( 0.3 , 0.6 ) , and ( 0.65 , 0.7 ) .
the top , middle , and bottom panels show the evolution of @xmath17 , @xmath52 , and @xmath53 , respectively .
as mentioned above , the period gap is more evident in fig .
2 than it is in fig .
1 . we note here several unique features associated with the evolution of individual cvs ; an understanding of these features will aid our interpretation of the results obtained for an entire population of evolving cv systems ( see 5 ) . for example , @xmath17 typically exhibits a sharp spike at the onset of mass transfer , ( see also rvj and hameury et al .
1988b ) ; this behavior will appear in all of the two - dimensional images " we produce from the population synthesis calculations in 5 .
the mass of the donor stays constant during its evolution through the period gap since there is no mass transfer taking place at that time - this is indicated by the horizontal lines in the middle panels .
the abrupt shift in location between the @xmath13 track above the period gap and below the gap will be dramatically apparent in the population synthesis results , and will have important consequences that are discussed below .
finally , the radius of the donor star decreases sharply after the system enters the period gap ; in fact , it is the shrinking of the donor inside of its roche lobe when the magnetic braking ceases that is the putative cause of the period gap .
again , the abrupt shift between the @xmath12 track above and below the period gap will be very pronounced in the population synthesis results .
a noteworthy feature of figures 1a , 1c , 2a , and 2d mentioned above is the sharp rise in @xmath17 whenever mass transfer has just commenced , including the first time that the donor star fills its roche lobe , and after the resumption of mass transfer below the period gap .
this results from the fact that when a low - mass star ( i.e. , @xmath54 ) is in thermal equilibrium ( i.e. , the nuclear luminosity , @xmath55 , equals the bolometric luminosity , @xmath56 ) , the sudden onset of mass transfer forces the star to expand because its adiabatic index is negative ( discussed above ) .
this expansion can cause a temporarily anomalously high rate of mass transfer , viz , the episodes of high @xmath17 seen in figures 1a , 1c , 2a , and 2d .
however , as soon as the donor star expands , its core temperature drops slightly , and @xmath55 , which is a highly sensitive function of temperature , drops dramatically .
this leads to a luminosity deficit wherein @xmath57 .
the star can then lose a net amount of energy , shrink , and approach its new thermal equilibrium radius ( appropriate to its lower mass ) on a kelvin - helmholtz ( i.e. , thermal ) timescale . during the mass loss process , true thermal equilibrium
is never reached , and the luminosity deficit attains a value which is adequate to allow the star to shrink continuously .
the above discussion explains the transient episodes of higher transfer rates at the start of mass transfer epochs , and the outlying " lower probability cv states we shall encounter in 5 .
it also explains the thermal bloating " of the donor star which is discussed in 3 , and which will play a key role in the observational test we propose in 6 .
( for earlier discussions of some of these basic effects , see rjw and hameury et al . 1988a . )
the five individual evolutions shown in figs . 1 and 2 serve to illustrate the range of interesting possibilities for cvs which commence mass transfer with different mass ratios .
the population synthesis study described in 5 , explores these various possibilities in a more systematic and complete way .
we start with the assumption that during mass transfer in a cv the roche lobe of the donor star is located within its atmosphere , i.e. , the donor star is filling " its roche lobe ( see howell et al . 2000 ) .
we then take the roche - lobe radius of the secondary star to be given by the simple expression of kopal ( 1959 ) : @xmath58 this can be combined with kepler s 3rd law to yield the well - known relationship among the mass , radius , and orbital period of the donor star : @xmath59 where @xmath60 and @xmath1 are expressed in units of @xmath48 , @xmath61 , and hours , respectively .
if we now assume that the radius of the donor star is some factor @xmath62 times the radius it would have if it were a normal main - sequence star , we can write @xmath63 where we approximate the radius - mass relation for stars on the lower main sequence ( i.e. , g to m stars ) by @xmath64 where @xmath65 and @xmath66 are constants , and we refer to @xmath62 as the bloating factor " . this bloating factor @xmath62 is simply a measure of how much larger the radius of a cv secondary is than that of a single , main - sequence star of the same mass due to the departure from thermal equilibrium .
we can now combine equations ( 2 ) and ( 3 ) to derive relations for the mass and radius of cv secondaries as a function of the binary orbital period .
@xmath68 for the purposes of this exercise , we take @xmath65 = 0.85 and @xmath66 = 0.85 , which we find by fitting a power law to the main - sequence models of dorman , nelson , & chau ( 1989 ; hereafter dnc ) . with these values for the constants @xmath65 and @xmath66 , the above equations simplify to : @xmath69 @xmath70 where , again , @xmath19 and @xmath21 are in solar units and @xmath1 is in hours .
the conclusions drawn from these expressions are somewhat counterintuitive in that , for a cv at a given orbital period , if the donor star is _ bloated _ , the proper radius and mass that should be inferred from the orbital period are _ smaller _ than the values that would be inferred if the star were on the main sequence ( see also beuermann et al
. 1998 ) . in 6 we derive polynomial fits for @xmath71 and @xmath72 from our population synthesis study ; the analytic expressions given by equations ( 6 ) and ( 7 ) serve mainly to demonstrate how these quantities scale with the bloating factor @xmath62 .
the individual binary evolution runs shown in figs .
1 and 2 for several different combinations of initial constituent masses are instructive , but do not ( _ i _ ) adequately sample the full range of possible initial masses , nor ( _ ii _ ) provide us with the distributions of cv binary properties at the current epoch .
we have therefore undertaken a population synthesis study of cvs which consists of two parts . in the first part
, we utilize a monte carlo approach to select a large number ( @xmath9 ) of primordial binaries , and follow the evolution of these systems to see which ones undergo a common envelope phase . in such events ,
the envelope of the giant star engulfs the secondary , leading to a spiral - in episode which leaves the secondary in a close orbit with a white dwarf ( i.e. , the core of the primary star ; see , e.g. , ( paczyski 1976 ; webbink 1979 ; see also 4.1 for details ) .
primordial binaries which are too wide will not undergo any significant mass transfer and will not lead to the formation of cv systems - the evolution of such wide binaries is not followed in the present study .
successful systems which emerge from the first part of our population synthesis caculations are those which do undergo a common envelope phase and yield a close binary consisting of a white dwarf and low mass ( @xmath73 ) companion .
the second part of the population synthesis considers those white - dwarf main - sequence binaries for which systemic angular momentum losses , or a modest amount of evolution by the normal companion star , can initiate roche - lobe contact within a hubble time .
each of these systems is then evolved in detail through the mass - transfer phase ( cv phase ) until the donor star has been reduced to a negligible mass ( typically 0.03 @xmath48 ) .
a number of prior population synthesis studies of cataclysmic variables have been carried out . these include work by politano ( 1988 ; 1996 ) ; de kool ( 1992 ) ; kolb ( 1993 ) ; di stefano & rappaport ( 1994 , for cvs in globular clusters ) ; and hrp ( emphasizing systems which evolve beyond the orbital period minimum ) .
the current study has several new features and advantages over the previous studies .
first , we compute probability density functions in two parameters , e.g. , @xmath74 and @xmath75 ( see 4.3 ) .
this way of studying and evaluating the results of population synthesis calculations has a distinct advantage over producing distributions of a single parameter .
for example , we are able to quantitatively evaluate phases of the evolution that are short lived or represent unusual evolutionary states , ( e.g. , whenever roche - lobe contact has just been established or when the initial binary mass ratio is near unity ) .
other examples include the ability to discern the spread in @xmath17 at a given orbital period , the distinction between systems with he and co white dwarfs , and the pronounced depression in secondary mass at a given orbital period ( for systems just above the period gap ) .
a second advantage is that our code for evolving the donor stars was originally developed to evolve brown dwarfs of very low mass and to very old ages .
the code has been well calibrated " against other more sophisticated ones that have been used for the purpose of evolving brown dwarfs ( see 4.1 ) .
finally , our population synthesis code which is used to generate the zero - age cvs that are input to the binary evolution code provides an independent check on previous work , and tests the sensitivity of our conclusions to various uncertainties in the physics , initial conditions , and other input parameters .
finally , we mention a possibly important limitation on the study we present here , which also applies to most other prior work in this area . we have considered only donor stars with @xmath76 , and do not allow for the chemical ( nuclear ) evolution the donor .
the latter approximation is realistic if the donor star commences mass transfer within @xmath313@xmath7710@xmath78 years of the common envelope event , or if the donor has a mass of @xmath79 .
these conditions apply to most of the systems that successfully evolve through the cv phase in our calculations .
furthermore , we find that only @xmath31 5% of all the stable mass - transferring , zero - age cvs in our population synthesis study have secondaries which are older than 1/3 of their main - sequence lifetime prior to the start of mass transfer .
theoretically , there should indeed be some cvs which evolve from donor stars that are initially more massive than @xmath80 , and they should be followed in future population synthesis studies . for the present study
, we simply assume that such systems , with donors whose initial mass exceeds 1 @xmath81 , do not contribute substantially to the cv population and , above all , would not affect our conclusions concerning systems near the period gap . in this
regard , recent work by beuermann et al .
( 1998 ) examines the properties of the secondary star in cvs in an effort to determine if they are indeed similar to normal main - sequence stars .
they show that , in the spectral type-@xmath1 plane , the @xmath82 cvs with measured spectral types lie below the expected relation for main - sequence stars ( i.e. , they are cooler at a fixed value of @xmath1 ) . beuermann et al .
conclude that for systems with @xmath83 hr this effect could result from mass loss ( see 5 ) , but that for longer orbital periods this effect suggests chemical evolution of the donor star .
this is a potentially important finding for systems with orbital periods longer than we consider here , and could also possibly impact the shorter period systems as well .
> from our population synthesis results , we find that @xmath84 of cvs could potentially form with progenitors whose mass is initially sufficiently high ( i.e. , @xmath85 ) that chemical evolution of the donor star might indeed be significant .
systems with such donor stars are not followed in the present study .
if , for some as yet unknown reason , the more massive donor stars have a greater efficiency for producing cvs than their lower - mass counterparts , then chemical evolution may indeed prove influential in the evolution of cvs .
these possibilities should be examined in future population synthesis studies .
the properties of the primordial binary systems are chosen via monte carlo techniques as follows .
the primary mass is picked from eggleton s ( 2000 ) monte carlo representation of the miller & scalo ( 1979 ) imf , @xmath86^{-1},\end{aligned}\ ] ] where @xmath87 is a uniformly distributed random number .
this distribution flattens out toward lower masses , in contrast with a salpeter - type power - law imf ( 1955 ) .
we considered primary stars whose mass is in the range of @xmath88 .
next , the mass of the secondary , @xmath19 , is chosen from the probability distribution , @xmath89 , where @xmath90 ( abt & levy 1978 ; but also see halbwachs 1987 ) .
this mass ratio distribution is , at best , poorly known empirically .
our adopted distribution has the property that the mass of the secondary is correlated with the mass of the primary , but is not strongly peaked toward @xmath91 .
we find that our results are not very sensitive to the choice of @xmath92 , unless an extreme is adopted such as the assumption that the two masses are to be chosen completely independently of one another ( see , e.g. , rappaport , di stefano , & smith 1994 ; hereafter rds - their table 2 and fig .
secondary masses as small as @xmath93 are chosen ( we wanted to ensure that only stars with masses clearly above the minimum main - sequence mass are included ) . to choose an initial orbital period , a distribution that is uniform in log@xmath94 over the period range of @xmath95 day to @xmath96 years
is used ( see , e.g. , abt & levy 1978 ; duquennoy & mayor 1991 ) .
since we consider only circular orbits the adopted orbital period distribution more properly pertains to the tidally circularized orbits than to the initial orbits of the primordial binaries .
after the masses and orbital period are chosen , the orbital separation is calculated using kepler s 3rd law .
we utilize an analytic expression for the relation among the core mass , the radius , and the total mass of the primary to estimate the mass of the degenerate core , @xmath42 , when the primary fills its roche lobe .
the expression we used for this purpose ( see rds ) was designed to reproduce the features of figure iii.2 of politano ( 1988 ) and figure 1 of de kool ( 1992 ) , except that the core - mass radius relation for stars with mass @xmath97 was renormalized to match the fitting formula of eggleton ( 2000 ; see eq .
[ 4 ] of joss , rappaport , & lewis 1987 ) .
mass loss via a stellar wind prior to the start of the first mass - transfer phase was computed via an analytic expression derived by m. politano ( 1999 , private communication ) . in practice
, the inclusion of this wind mass loss does not significantly affect the results . in order to select only systems which undergo a common envelope phase
we require that the radius of the roche lobe of the primary be larger than the radius of a star of mass @xmath20 at the base of the giant branch ( see , e.g. , paczyski 1965 ; webbink 1979 , 1985 , 1992 , de kool 1992 , and references therein ) .
this ensures that unstable mass transfer will occur on a timescale that is substantially shorter than a thermal time , and should lead to a common envelope phase .
once mass transfer from the primary to the secondary commences , we assume that a common envelope phase occurs and compute the final spiral - in separation based on simple energetic considerations ( see , e.g. , taam , bodenheimer , & ostriker 1978 ; meyer & meyer - hofmeister 1979 ; livio & soker 1988 ; webbink 1992 ; rds ; taam & sandquist 1998 ) .
the expression we use for determining @xmath98 , the final orbital separation after spiral - in , is given by : @xmath99 where @xmath100 and @xmath101 are the core and envelope masses of the primary , @xmath102 is the radius of the primary , @xmath103 is the initial orbital separation , and @xmath104 is the energy efficiency factor for ejecting the envelope .
we take @xmath104 to have a value of 1.0 in our standard model .
the two terms in parentheses on the right hand side of equation ( 9 ) represent the binding energy of the envelope of the primary to itself and to its core .
the dimensionless coefficients multiplying each term were computed for an assumed polytropic envelope structure with polytropic index @xmath105 ( rds ) .
for other similar values of @xmath106 the ratio of @xmath107 between the two coefficients is roughly the same .
we assume that the duration of the spiral - in is sufficiently short ( @xmath108 yr ; see above references ) that the mass of the secondary does not change significantly during the common envelope phase .
after the spiral - in episode , the separation , the white dwarf mass , the secondary mass , and the corresponding roche - lobe radius of the secondary are known .
if at the end of the common envelope phase the secondary would already be overfilling its roche lobe , then we eliminate the system .
( in most cases , this circumstance would be expected to lead to a merger of the secondary star with the degenerate core of the primary , which presumably would result in the formation of a giant star . ) in practice , if the roche lobe is larger than @xmath109 , then neither magnetic braking nor gravitational radiation would bring the system into roche - lobe contact before the secondary would evolve past the base of the giant branch .
we can also eliminate these systems since the ensuing mass transfer would either be dynamically unstable ( see , e.g. , paczyski 1967 ; kippenhahn , kohl , & weigert 1967 ; webbink 1979 , 1992 ) or lead to an even wider orbit ; either case would not produce a cv of the ordinary kind .
we typically start with @xmath110 primordial binaries and end up with @xmath111 pre - cvs to evolve through the mass - transfer phase with the bipolytrope evolution code described in the next section .
the computational time for this first portion of the calculations is negligibly short .
as mentioned earlier , the evolutionary tracks of cv systems are calculated using a version of the code that was first developed by rvj ( see also rjw ) to explore the effects of the parameterized verbunt & zwaan ( 1981 ) magnetic braking law on the evolutionary properties of cataclysmic variables . according to their algorithm , the mass losing donor is approximated by a bipolytrope wherein the convective envelope is represented by an n=3/2 polytrope and the radiative core by an n=3 polytrope .
one of the advantages of this code is that it allows for the rapid computation of a large number of evolutionary tracks and provides a more physically intuitive interpretation of the results .
the original version of the code has been modified substantially to allow for improvements to the input physics , and to ensure that the conditions near the surface ( atmosphere ) are more physically realistic .
a number of these changes have been discussed in previous papers .
the most significant of these modifications and updates are described by nelson , rappaport , & joss ( 1986a , 1986b ; 1993 ) who used a single polytrope model to follow the evolution of fully convective low - mass stars and brown dwarfs .
the results of the brown dwarf cooling evolutions and the calculation of zams models of low - mass stars are in excellent agreement with those calculated using more sophisticated techniques ( see , e.g. , dnc ; burrows et al . 1993 ; burrows et al . 1997 ; baraffe et al .
1998 , and references therein ) .
specifically , coulombic corrections to the pressure equation of state were incorporated and an updated version of the alexander , johnson , & rypma ( 1983 ; alexander [ 1989 ] ) low - temperature , radiative ( surface ) opacities was used .
the molecular hydrogen partition function was also calculated more accurately .
most importantly , the specific entropy at the surface was matched directly to the specific entropy in the interior ; i.e. , at the interface between the radiative core and the convective envelope . model ( @xmath112 of the specific entropy ) , decreasing to zero for fully convective stars . ]
in addition to these changes , the atmospheric pressure boundary condition was modified so as to approximate more closely the scaled solar t-@xmath113 ( krishna - swamy 1972 ) relation .
the radiative surface opacities did not include the effects of grain formation .
since grains can only form in the atmospheres of very low - temperature stars ( @xmath114 1500 k ) , this should affect mostly the evolution of those cvs that have evolved beyond the orbital period minimum .
however , we have found that the evolution of cvs through and beyond the period minimum , is not particularly sensitive to this omission .
the overall result of all of these changes is that the theoretical radius - mass relation for our zams models with masses @xmath115 is now in substantial agreement with other theoretical calculations as well as with observational studies of low - mass stars ( see dnc ) . for similar abundances of hydrogen and for stars of approximately solar metallicity
, we find that the radii of our new models compared with other theoretical models ( and the dnc results ) typically agree to within an rms error of @xmath116 ( @xmath117 ) .
deviations among the theoretical models are greatest for the higher mass stars due to uncertainties in the mixing length parameter and the treatment of inefficient ( superadiabatic ) convection . when observations of double stars are considered , we believe that our zams radii are accurate to within @xmath118 .
our zams models become fully convective at a mass of @xmath119 .
this is considerably smaller than the value given in rvj but agrees well with the dnc results ( as well as with newer generations of models ) .
mass transfer in cvs is driven by angular momentum losses due to gravitational radiation ( landau & lifshitz , 1962 ) and other systemic angular momentum losses such as magnetic braking " .
the magnetic braking law that we utilize is that of verbunt & zwaan ( 1981 ) and parameterized by rvj .
the magnetic braking parameters were chosen so as to best reproduce the observed period gap . according to the parameterization described in rvj
, we took @xmath120 and did _ not _ adjust the multiplicative constant ( defined here as @xmath121 ) used in the rvj prescription .
we also shut off " magnetic braking when the radiative core had been reduced to less than 15% of the mass of the donor .
magnetic braking is assumed to be greatly reduced as a result of the restructuring of the magnetic field of the donor star when it becomes nearly fully convective .
this reduction in the angular momentum loss rate gives the donor an opportunity to shrink inside of its roche lobe on a thermal timescale .
further angular momentum losses due to gravitational radiation cause mass transfer to recommence once the binary system is brought back into a state of semi - detachment ( see , e.g. , rvj ; spruit & ritter 1983 ; hameury et al .
1988a for a more detailed explanation ) . as pointed out in several places in this work , the actual mechanism that produces the bloating of the donor , and the means by which mass transfer is interrupted , are not central to the conclusions drawn in this paper .
what is important in this regard is that the bloating be sufficiently large as to produce the observed width of the period gap . for our standard evolutionary model
the period gap covers the range of 2.1 @xmath122 2.85 hr . according to warner ( 1995 ) , this synthetic gap approximates the observed one very well .
we assume that mass and orbital angular momentum _ are _ lost as a result of nova explosions on the surface of the white dwarf accretor . for our standard model we assumed that all of the mass that is accreted by the white dwarf is lost with the same specific angular momentum as the white dwarf itself ( see schenker , kolb & ritter 1992 ) .
given the relatively low mass transfer rates , it is likely that the nova events are extremely hydrodynamic , and thus it is unlikely that any of the accreted mass actually contributes to increasing the mass of the white dwarf ( see , e.g. , prialnik & kovetz 1995 , starrfield 1998 , and references therein ) .
after a potential cataclysmic variable system has been generated with the population synthesis code , the two detached components are given the opportunity to come into contact , via magnetic braking , within the age of the galaxy ( minus the cv formation time ) .
however , the initial mass transfer may actually be unstable , thereby leading to a common envelope phase ( and the ultimate demise of the binary system ) .
as derived by rjw , the expression for the long - term mean mass transfer rate in a cv is given by @xmath123 = @xmath124 , where the numerator , @xmath125 , contains the drivers of mass transfer , e.g. , systemic angular momentum losses , and the thermal expansion / contraction of the donor star ( see eq . [ 33 ] in rvj ) .
the denominator is given by @xmath126\end{aligned}\ ] ] where @xmath127 ( note that this is the inverse of the definition used in rvj ) , @xmath128 is the fraction of the mass lost by the donor star that is ultimately retained by the white dwarf , @xmath129 is the specific angular momentum carried away by matter ejected from the binary system in units of the binary angular momentum per unit reduced mass , and @xmath130 is the adiabatic index of the donor star , i.e. , @xmath37_{ad}$ ] . for our standard model
( see table 1 ) , we take @xmath131 ( i.e. , all the mass accreted by the white dwarf is eventually ejected in nova explosions . ] ) , and @xmath132 . with these definitions
, the above equation reduces to @xmath133 as discussed by rjw , stable mass transfer requires @xmath134 and @xmath135 . as an example , consider donor stars with @xmath136 and @xmath130 = -1/3 . in this case , stability ( based on eq .
[ 11 ] ) requires that @xmath137 .
this allows for considerably larger values of @xmath19 than the more conventional limit for conservative transfer where @xmath138 is required for stable mass transfer ( with low - mass unevolved donors ) .
thus , the mass ratios that appear in our population synthesis can often approach unity or exceed it .
we define a birth rate function , @xmath139 , for the progenitor primordial binaries , where @xmath140 is the elapsed time between the formation of the galaxy and the birth of the primordial binary .
if a binary is born at time @xmath140 , then an additional time @xmath141 must elapse before the primary evolves to the point where a common envelope phase may occur ( see 4.1 ) .
we define this time ( @xmath142 ) to be the birth time of the incipient cv .
the resultant zero - age cv is then evolved in the binary evolution code for a total time @xmath143 = ( @xmath144 - @xmath141 ) yr , which is the maximum time any cv that is descended from a similar primordial binary could evolve before the current epoch .
( the binary evolution code starts with the white dwarf and companion star as they emerge from the common envelope , so the elapsed time , @xmath145 , includes the interval before the donor star fills its roche lobe . ) at each step in the evolution code , specified by time @xmath145 ( with respect to the first time step in the code ) , we sum in discrete binned arrays for various combinations of @xmath1 , @xmath52 , @xmath146 , @xmath18 , @xmath17 , @xmath22 , and @xmath23 , the following quantity , @xmath147 : @xmath148 in this expression , the argument of @xmath149 is the time that the primordial binary was born with respect to the formation of the galaxy , @xmath150 is the time interval for that particular step in the evolution run , and @xmath125 is the total number of systems that are selected to start the population synthesis run . for all of
the population synthesis runs in this study , the @xmath149 was taken to be constant in time .
even though we have adopted a constant stellar birth rate per unit time , the method we use for generating the cv population at the current epoch is completely general ( see also kolb 1993 ) .
the net result of this procedure is that the sum of the @xmath147s at the end of the population synthesis run , in any particular bin , represents the number of cvs at the current epoch with that particular parameter value .
the computed population of current - epoch cvs as generated by the above techniques is displayed as a sequence of color images in figures 3 through 7 . in figure 3
we show the model cv population in the @xmath11 plane for our standard model ( cf . fig .
the image is generated in such a way that the color reflects the logarithm of the number of current - epoch cvs at a particular location in the @xmath11 plane . in each of the images the color scale is located on the right side .
the image in fig . 3
is comprised of 100 pixels per hour interval in @xmath1 , and 100 pixels per decade in @xmath17 .
the most noteworthy features in figure 3 include the distinct groups of systems located above and below the period gap .
note the substantial difference in @xmath17 for systems above and below the period gap ; for the latter systems only gravitational radiation losses drive mass transfer . the minimum orbital period ( @xmath15165 min )
is also clearly evident , as are systems that have evolved well past the minimum period back up to values of @xmath152 hr .
it has been proposed that these latter systems may be related to the so - called toads ( tremendous outburst amplitude dwarf novae " ; see , e.g. , howell et al .
1995 , hrp ) . in the systems above the gap
, there is a central band of evolutionary tracks ( blue and green ) where a typical cv is most likely to be found at a particular point in time during its evolution .
one also notices very short lived episodes ( red and yellow ) of high mass transfer rates .
these occur for individual systems as the donor star first fills its roche lobe and commences mass transfer , but before it can come into a quasi - steady state of mass transfer ( see discussion in 2 ) . the same type of behavior is seen ( green structure ) for systems that have come into contact for the first time below the period gap , i.e. , with initially very low - mass donor stars .
the two main tracks ( purple ) evident in the systems below the period gap are for he ( lower ) and co ( upper ) white dwarfs , respectively .
we also call attention to the small vertical ( blue ) feature at @xmath152 hr .
this may be related to the statistically significant larger number of cvs with periods in the range of @xmath153 minutes ( first pointed out by hameury et al .
finally , we note that there are systems found within the period gap , though fewer per period interval than for systems below the gap .
systems found within the period gap are typically ones which had initial donor masses of @xmath310.22 - 0.34 m@xmath154 and commenced roche - lobe overflow at orbital periods in the range of @xmath312 - 3 hr ( see also fig . 2 and its associated discussion ) .
further , in regard to the @xmath11 plane shown in figure 3 , we point out that for systems above the period gap , the width of the distribution in @xmath17 at any fixed value of @xmath1 is only about a factor of @xmath312 ( we define the width " as containing @xmath155 of the systems ) .
this is in contrast with the _ observed _ spread in @xmath17 for cvs which is closer to an order of magnitude ( see , e.g. , patterson 1984 ; warner 1995 ) .
one cause of this spread may be the inherent uncertainty in translating observed parameters into accurate estimates of @xmath156 .
additionally , some of this discrepancy might be resolved by the inclusion of the effects of nova explosions in cvs which , on a quasi - regular basis , slightly increase ( or perhaps even decrease ) the orbital separation ( e.g. , by @xmath157 ) which is sufficient to change @xmath17 appreciably for some interval of time ( see , shara et al .
1986 ; schenker et al . 1998 ; kolb et al .
however , we note that schenker et al .
( 1998 ) showed that , except for extreme model parameters , the occurrence of the nova explosions generally does not substantially affect the overall secular evolution of the cvs .
therefore the main results and conclusions presented in this work should be robust even without the inclusion of _ orbital _ perturbations due to nova explosions ( we do , in fact , take into account the mass and angular momentum lost in such events ) .
the population of current - epoch cvs in the @xmath158 plane for our standard model is shown in figure 4 .
the shape traced out in this figure represents a statistical ensemble of the type of evolutions graphed in figs . 2c and 2f .
the usual features of the upper branch " of systems above the period gap , systems in the lower branch " below the gap , and the minimum orbital period are all represented in this figure . again , as in fig .
3 , we see that some systems are formed within the period gap .
it is difficult from this image to judge quantitatively how many systems are in the gap , versus the density of points just below the gap .
this is quantified later in this section ( see fig .
note that an extrapolation of the upper branch " to shorter orbital periods would undershoot the lower branch " , on which the stars are close to thermal equilibrium . as discussed in 2 , this undershooting actually ( counterintuitively ) results from the thermal _ bloating _ of the donor star when it has a higher mass loss rate which is driven by magnetic braking .
in particular , see equation ( 7 ) where we show that @xmath21 scales as @xmath159 , where @xmath62 is the bloating factor .
the low - density features ( yellow ) just above the main tracks through the upper branch " are systems that have just come into roche - lobe contact for the first time and have not yet established a quasi - steady state of mass transfer .
the blue - green thumb " feature just below the main track of the upper branch " near the top edge of the period gap , represents systems with he white dwarfs and donor stars of comparable mass that have just come into contact .
their mass transfer rates are higher than normal for these orbital periods ; thus the bloating factors for these donor stars are significantly larger than for systems on the main track ( cf . fig .
perhaps the most dramatic demonstration of the effects of thermal bloating of the donor star can be seen in figure 5 which shows the population of current - epoch cvs in the @xmath160 plane for our standard model .
the shape traced out in this figure represents a statistical ensemble of the type of evolutions graphed in figs .
2b and 2e . all of the features that appear in the @xmath158 image ( fig .
4 ) , also appear in this @xmath160 image , except in a more exaggerated form .
this is a direct result of the simple scaling argument summarized in equation ( 6 ) in 2 , which indicates that the bloating effect on the masses just above the period gap scales as @xmath161 .
a casual inspection of figure 5 shows that the masses of the donor stars in cvs with periods just above the period gap are fully @xmath162 lower than would be expected if their radius - mass relation followed that of main - sequence stars .
it is this effect that we propose be used to discriminate between the currently held explanation for the period gap and alternate scenarios .
we return to a quantitative discussion of this issue in the next section .
lastly in regard to the color images of the @xmath12 and @xmath13 planes ( figs . 4 and 5 ) , we comment on the relatively large spreads in @xmath21 and @xmath19 for systems above the period gap in contrast with those below the gap . as we showed in equations ( 6 ) and ( 7 ) , for a fixed value of the bloating parameter @xmath62 , both @xmath19 and @xmath21 are unique functions of the orbital period ( which would imply narrow tracks ) . for systems well above the period gap ,
the kelvin timescale , @xmath35 , is shorter than the mass - loss timescale , @xmath163 , but , as the orbital period decreases and approaches the period gap , the two timescales become more comparable . thus , as discussed in 2 , the donor star must become ever more bloated so as to establish a luminosity deficit , which in turn enables the donor to contract inside its ever shrinking roche lobe .
additionally , the adiabatic stellar index is changing from positive to negative , and this tends to make the star expand even further as it loses mass ( see also beuermann et al .
these two effects lead to the bloating behavior that is seen in figs . 4 and 5 .
the actual amount of bloating depends upon the absolute values of the two constituent masses as well as on the thermal history of the donor ; therefore , it is to be expected that @xmath62 may vary from one donor star to another . as a result
, we not only see enhanced bloating as systems approach the period gap , but a relatively wider and wider spread in the values of @xmath19 and @xmath21 for these systems , especially for @xmath1 in the range of 3 - 5 hr . by contrast , for systems just below the period gap , @xmath164 increases abruptly - by about an order of magnitude - because the mass transfer is then driven only by gravitational radiation losses ( at least according to our model ) , and therefore the donor stars can remain much closer to thermal equilibrium .
this allows the systems right below the gap to establish a nearly main - sequence radius - mass relation ( i.e. , @xmath165 ) , thereby leading to a relatively narrow set of evolution tracks .
however , as the secondary s mass approaches the minimum main - sequence mass ( before the orbital period minimum ) , @xmath35 becomes very long ( due to a sharp decrease in the secondary s nuclear luminosity ) , thereby causing @xmath35 and @xmath164 to again become approximately equal .
thus , the width of the tracks broadens somewhat near the orbital period minimum . for systems beyond the orbital period minimum ,
the interiors become increasingly electron degenerate .
this leads to a nearly unique mass - radius relationship @xmath166 which , in turn , leads to entirely different @xmath167 and @xmath168 relations than are given by equations ( 6 ) and ( 7 ) .
nonetheless , they are unique relations ( easily derivable from eqs . [ 4 ] and [ 5 ] ) which also lead to a very narrow set of tracks in figs . 4 and 5 .
the distribution of expected mass ratios , @xmath18 , in cvs at the current epoch is shown as a function of orbital period in figure 6 . at any given orbital period
the range of @xmath18 values is considerably broader than the distribution of values of @xmath21 or @xmath19 as can be seen by comparison with figs . 4 and 5 .
the reason for this is straightforward .
equations ( 6 ) and ( 7 ) indicate that , as long as the bloating factor @xmath62 depends largely on the orbital period of a cv , then both the radius and mass are nearly unique functions of the orbital period .
thus , the much broader distribution of @xmath18 in fig .
6 is due largely to the substantial range of masses that the white dwarf may have , which is much less constrained by the orbital period than is @xmath19 . for both the systems above and below the period gap , the upper set of tracks corresponds to he white dwarfs , while the lower tracks are for co white dwarfs .
the period gap is especially conspicuous in this figure , especially for systems with co white dwarfs .
note that some of the mass ratios extend up to values of unity and , in some cases , above unity .
the stability of mass transfer in these systems was discussed in 4.2 [ see equation ( 11 ) ] .
the evolution of our model population of cvs in the @xmath15 and luminosity@xmath169 planes is shown in figure 7 .
the left panel displays the effective temperature , @xmath22 of the donor star , while the right panel has a superposition of the optical ( bolometric ) luminosity , @xmath56 and nuclear luminosity @xmath55 .
where the two sets of luminosity tracks cross ( e.g. , near @xmath1 = 1 hr ) or overlap ( e.g. , to a minor extent between 3 and 6 hrs ) , the default is to display @xmath56 . with regard to the @xmath22 curves
, we first note that the absolute temperature scale for our main sequence stars ( based on our bipolytrope code ) is somewhat shifted from that produced with more sophisticated codes , e.g. , our bipolytrope main - sequence models are @xmath31200 k higher than the dnc models over the mass range of @xmath170 .
however , aside from this small quantitative difference we are confident that the overall qualitative trends and shapes of these tracks are highly indicative of the behavior and properties of the donor through its evolutionary history .
note that for values of @xmath1 below the gap as well as above @xmath315 hr , the @xmath22 tracks are quite narrow , in analogy with the tracks in the @xmath13 and @xmath12 planes , since the donor stars are typically quite close to thermal equilibrium .
by contrast , within the period range of 3 - 5 hr , @xmath22 of the donors is systematically lowered by up to 250 k compared with @xmath22 of main sequence stars at the same @xmath1 .
this lower temperature amounts to a change to a later spectral type ( at a given @xmath1 ) of @xmath171 in decimal subclass .
additionally , we can see from fig .
7 that , over this same period range , the use of temperature ( or spectral type ) to determine the mass of the secondary star would require very precise measurements , since the expected @xmath15 distribution is relatively flat .
we draw two conclusions from this figure : ( 1 ) an observationally produced @xmath15 or spectral type-@xmath1 relation for cvs should indeed yield a fairly simple shape ( especially when smoothed out by uncertainties in the measurements ) , and ( ii ) the use of @xmath22 or spectral type in the 3 - 5 hr period range will not yield reliable indications of the mass of the donor star .
recently , smith & dhillon ( 1988 ) 90 min and @xmath172 brighter than @xmath3117th magnitude , thus smith & dhillon s conclusions about finding no evidence for post - period minimum systems or very low - mass brown dwarf - like secondaries can not be drawn from the sample they used . ]
published results which took a critical look at the relation between orbital period , spectral type , and secondary mass based on the best observational data available in the literature .
they presented a relatively smooth spectral type - orbital period relation but concluded that one can not reliably estimate @xmath19 in any given cv based solely on its spectral type .
our theoretical results are quite consistent with their conclusion .
the image in the right panel of figure 7 displays a superposition of the evolutionary tracks for @xmath56 and @xmath55 as functions of @xmath1 . for systems above the gap , the highest luminosity track corresponds to @xmath56 , while the two prominent lower ( green ) tracks are for @xmath55 and are related to the corresponding features in the @xmath13 image .
these two lower luminosity tracks are for systems with co ( upper ) and he ( lower ) white dwarf accretors ( see the discussion of fig .
the large luminosity deficit in the @xmath1 range of 3 - 5 hr , already discussed in 2 , shows up quite dramatically in this fig .
the group of systems with the highest luminosity deficit ( with he white dwarf accretors ) has the largest values of @xmath17 and the donors are the most out of thermal equilibrium ( largest bloating factor ) . in spite of the relatively low values of @xmath55 in this period range ,
the bolometric luminosity @xmath56 is depressed only modestly ( e.g. , by factors of @xmath173 2 ) over main - sequence stars at the same orbital period . for systems below the period gap ,
both luminosities fall off dramatically , especially for donor masses below @xmath174 , where the donors are already below the hydrogen - burning main sequence , and are cooling toward their ultimate degenerate state . the higher track for all points below the period gap corresponds to @xmath56 , the lower one to @xmath55 . while it is formally true that the @xmath56 and @xmath55 tracks cross " at @xmath175 , the two luminosities are never equal in this part of the diagram ; they reach the crossing point at very different times .
finally , with regard to the color image representations of cv populations in parameter space , we note that in figs . 3 through 7 , the color represents the logarithm of the numbers of systems expected at the current epoch . as can be seen in any of these figures ( but especially figs . 3 and 6 ) , the number of systems below the period gap outweighs the number above the period gap by a large margin ( by about 100:1 ; see the more quantitative discussion below ; see also de kool 1992 , kolb 1993 ) . however , due to observational selection effects , the systems with the shorter orbital periods , lower values of @xmath17 , and generally longer intervals between dwarf - nova outbursts , are more difficult to discover .
the exact factors that go into the observational selection effects are complex , especially since some cvs are discovered via their dwarf - nova outbursts , others ( e.g. , longer period cvs ) by their blue colors or flickering behavior , and still others by their nova outbursts . some of these issues are discussed in rjw and kolb ( 1993 ) . for purposes of the present work we will indicate only qualitatively how the numbers of observationally known cvs
might be expected to be distributed by orbital period .
we adopt two crude detectability factors which scale simply as @xmath176 and as @xmath17 .
the first of these is appropriate to steady - state accretion luminosities , that are proportional to @xmath17 in the optical bandpass , which give rise to a bolometrically flux - limited detectability proportional to @xmath176 ( analogous to the 3/2 slope of a log(n)-log(s ) curve for isotropically distributed sources ) . the other
scaling , appropriate to the case where the flux in the optical bandpass is proportional to @xmath177 ( see lynden - bell & pringle 1974 ; rjw ; webbink et al .
1987 ) , leads to a flux - limited detectability proportional to @xmath17 .
other factors leading to the discovery of cvs , beyond the simple consideration of flux limited samples , in particular the detection of dwarf nova outbursts , would undoubtedly substantially modify the rudimentary dependences on @xmath17 that we use here for purposes of illustration .
in figure 8 we redisplay fig . 3 , but this time rescaled by a factor of @xmath156 .
it is clear from a casual inspection of fig . 8
that the number of detectable " systems above the period gap is now at least as great as for those below the gap .
the actual quantitative values for this simple scaling are presented below .
we again caution , however , that either an @xmath176 or @xmath17 scaling is oversimplified .
the color images of parameter space shown in figures 3 7 can be displayed in a somewhat more quantitative fashion by projecting the numbers of systems onto the various axes and plotting the results as simple histograms .
for example , the data used to produce any of the images , can be projected onto the @xmath1 axis to yield the orbital period distribution .
the results are shown in figure 9 .
the solid histogram in fig .
9a is the distribution of cvs at the current epoch in the entire galaxy for our standard model ( see table 1 )
. the stellar birthrate function and imf in our standard model [ eq .
( 8) ] are normalized in such a way that there are @xmath178 stars born in the galaxy per year with a mass @xmath179 , just above the threshold for producing a remnant white dwarf by the current epoch .
thus , the absolute values " of the numbers plotted in fig .
9 can be appropriately scaled up or down for either lower or higher assumed birthrates . if the numbers of cv systems are scaled by the types of observability factors " discussed above , before the histogram is produced , the results are the dashed and dotted histograms superposed in figure 9a . as discussed above , in conjunction with fig .
8 , this qualitatively helps to explain the relative numbers of cvs observed above the period gap compared with the number observed below ( see especially the dotted histogram ) .
inspection of the compilation of cvs with known orbital periods given in warner ( 1995 ) reveals that our histogram shown in fig .
9a with the @xmath17 scaling provides qualitative agreement with current observational results , especially considering the many observational selection effects that exist ( e.g. , magnitude - limited color surveys , large - amplitude but infrequent outbursts compared with semi - periodic lower amplitude outbursts , discovery in x - ray surveys , etc . ) .
the distributions of orbital period shown in figure 9b are for systems that have not yet evolved to the minimum orbital period ( solid curve ) , and systems that have evolved beyond the period minimum ( dashed curve ) - no scaling in @xmath17 has been applied here .
the distributions of white dwarf masses and donor masses at the current epoch are shown in figure 10 ( left and right panels respectively ) for four different ranges of orbital period .
the he and co white dwarfs are easy to distinguish by mass . note that for systems with @xmath180 hr , which typically have donor stars with masses @xmath181 , there are few he white dwarfs , since the mass transfer would tend to be unstable .
the distributions of donor star masses show a steady trend toward higher masses at the longer periods , as expected .
this results qualitatively from the fact that the larger orbital periods require less dense , and therefore usually more massive , stars .
note that the distributions shown in this figure are not produced with sufficient resolution in @xmath1 to allow one to make quantitative predictions as to what mass donors are needed to validate the basic paradigm for the period gap .
such information may be found , however , in figs . 5 and 12 , and table 2 .
the distribution of mass ratios @xmath18 ( @xmath182/@xmath42 ) is shown in figure 11 for two different orbital period ranges . attempts to determine @xmath18 observationally
can be made from , for example , superhump period analysis or spectroscopic analysis .
the observational distribution for @xmath18 in short period ( @xmath183 hr ) cvs has recently been compiled ( mennickent et al .
1999 ) , and is seen to show an approximate gaussian distribution with @xmath184 = 0.14 .
however , observational selection effects allow few cvs with small @xmath18 to be discovered due to their intrinsic faintness .
our results ( fig . 11 - top panel ) show that the actual distribution should not drop off at @xmath18 values lower then 0.14 , but rather should peak at values of @xmath18 = 0.05 - 0.1 , with an overall distribution that is clearly non - gaussian .
discovery and observation of additional faint ( short period ) cvs are needed in order to confirm this theoretical prediction .
the rapid rate of mass loss for donor stars in cvs just above the period gap should lead to significant thermal bloating of the donor .
thus , in the conventional paradigm for the formation of the period gap , this mass loss rate is abruptly decreased at orbital periods near 3 hr and the donor star shrinks inside its roche lobe ( see 2 ) , leading to the cessation of mass transfer .
specific choices of the parameters utilized in any such evolutionary model change the bloating factor quantitatively , but do not change the overall evolution qualitatively . to demonstrate this ,
we show in figure 12 cvs at the current epoch in the @xmath160 plane for four different sets of model parameters ( see table 1 ) .
panel ( a ) is for our standard model , while the other panels are for models where ( b ) the proportionality constant in the magnetic braking formula was reduced by a factor of 2 ( @xmath185 ) , ( c ) the specific angular momentum carried away by mass lost from the system in nova explosions is twice that of the white dwarf ( @xmath186 ) , and ( d ) all mass transferred to the white dwarf is ultimately retained by the white dwarf ( i.e. , @xmath187 ; in this somewhat artificial model , white dwarfs are allowed to exceed the chandrasekhar limit ) .
we see from a study of figure 12 that the effects of thermal bloating on the mass of the donor stars in cvs for orbital periods just above the gap are qualitatively similar for all four models .
the actual factors by which the masses are lower than would be inferred by making the assumption that the donor has a main - sequence mass - radius relation range from @xmath188 ; the exact range depends on which model parameters are chosen and whether one includes the cvs with he white dwarfs where the mass transfer can be only marginally stable . to quantify the effect of thermal bloating on mass determinations , we have carried out weighted least squares fits of polynomials to each of the upper branches " shown in fig
the results are given in table 2 which also includes the evaluation of the polynomial fit at @xmath189 hr .
as we can see from table 2 , the effect of thermal bloating on the inferred donor mass of cvs is quite significant , and potentially testable for any of these models .
we note that , in general , the _ spread _ in values of @xmath19 , at a given @xmath1 , around the best - fit @xmath168 curve is substantially smaller than the mean deviation from an @xmath168 curve based on the assumption of a main - sequence mass - radius relation , especially in the crucial period range of 3 - 5 hr .
a large part of this spread is due to the different values of the mass of the accretor , @xmath20 , with lower values of @xmath19 corresponding to the lower values of @xmath20 ( see also the discussion in 5 ) .
however , we do not attempt here to produce fits of the more general form @xmath190 .
such fits are not straightforward to construct since , among other things , there is the added complication of the existence of a minimum value for @xmath191 at any @xmath1 ( due to issues of mass - transfer stability ; see 4.2 ) . in any case , the main effect to be confirmed observationally concerns the substantially reduced values of @xmath19 just above the period gap ( 3 - 5 hr ) , compared to what would be expected if the donor stars followed a main - sequence mass - radius relation .
if sufficient numbers of high quality mass determinations of the secondary stars can be made , and this basic effect is confirmed , then a secondary goal would be to look for a weak , but positive correlation between @xmath19 and @xmath20 . in figure 13
we plot the polynomial fits that we made to the upper branches in the @xmath13 plane for the four different models . for comparison
we show the @xmath13 relation that would be obtained if the donor star followed a main - sequence radius - mass relation ( the one derived from our bipolytrope code ) .
this set of curves shows quantitatively how mass determinations based solely on @xmath1 are affected by the thermal bloating effect .
note how the effect should go from a maximum at @xmath192 hr to quite small at @xmath193 hr . finally , we point out that if , in fact , the period gap is in any way related to a relaxation from thermal bloating , then the inferred effect on mass determinations based on the orbital period must be approximately in the range of @xmath188 . to demonstrate this , we note that in the basic paradigm for producing the period gap , the system masses do not change from the upper boundary of the gap ( @xmath194 ) to the lower boundary ( @xmath195 ) , while the radius shrinks from its bloated state , characterized by a bloating factor , @xmath62 , to nearly its main - sequence radius at the lower edge of the gap .
a simple application of kepler s 3rd law for the case of a roche - lobe filling star ( which is true at both the upper and lower edges of the gap ) shows that @xmath196 where @xmath62 must range from @xmath197 , depending on whether the period gap is taken to be 3/4 of an hour in width or 1 hour , respectively ( we have assumed that the gap is centered at 2.5 hr ) . from equation ( 6 ) we see that this value of @xmath62 should reduce the inferred mass , at the top edge of the period gap , by amounts ranging from @xmath198 , in basic agreement with our more detailed population synthesis study .
( for a related discussion see beuermann et al . 1998 . )
this type of discrepancy between the mass inferred for a secondary star , based on the cv orbital period and the assumption that its radius is that of a main - sequence star has probably already led to a number of incorrect mass determinations reported in the literature , particularly for systems with @xmath1 between 3 - 5.5 hr . for example , at an orbital period of 3.2 hours , the mass assigned to a cv secondary would be 0.35 m@xmath154 while our calculations show that it would actually be only 0.26 @xmath199 , although bloated in size . for a known or inferred mass ratio of say @xmath200 , we would then calculate a white dwarf mass of 0.89 m@xmath154 , when the true white dwarf mass is only 0.65 m@xmath154 .
thus , ignoring the bloating effect in the secondary stars in cvs with orbital periods of 3 - 5 hr , can lead to a significant overestimation of both component masses .
it is interesting to note here that the secondary stars which are farthest from thermal equilibrium are those in systems with orbital periods just above the period gap ( see figs 4 , 5 & 13 ) .
observationally , this orbital period region ( 3 - 4 hr ) essentially contains only high mass transfer rate , novalike ( nl ) types of cvs .
the inferred high mass transfer rates for these systems would then be expected , on theoretical grounds , to lead to a large bloating of the secondary stars , and hence lower masses than might otherwise be anticipated .
precise observational determinations of the secondary star masses in nls would allow a confirmation of this basic effect which is , in fact , required if the period gap is to be explained by the interrupted magnetic braking scenario .
figure 13 provides our theoretical predictions for the most likely mass of the secondary star at any given orbital period ( see table 2 ) .
observational determinations accurate to a few percent would be needed in order to differentiate between the four models presented ; but , accuracies of only @xmath201 will allow a test of the bloating model in general , and the predicted deviation of the donor star from the main sequence .
this is a challenging observational project , however , since the systems with orbital periods in the 3 - 5 hr range are ones in which the secondary star is rarely directly observed .
ir spectral studies ( e.g. , howell , et al .
2000 ; mason , et al .
2000 ; dhillon , et al .
1997 ) have looked in detail for the secondary star in a number of cvs with only marginally successful results . in these cvs , spectral identification of absorption features due to
the secondary star is difficult since the lines are rotationally broadened and filled in by radiation from the accretion disk . for the critical 3 - 5 hr period range ,
a signal - to - noise ratio of @xmath202100 in the continuum will be needed to allow the atomic and molecular features of the secondary to be observed against the high background accretion - disk dominated continuum .
we therefore advocate high signal - to - noise , orbital phase - resolved , near- and mid - ir spectroscopic observations with large ground - based telescopes ( e.g. , gemini , keck ) , and eventually with sirtf , of sources such as the brightest nls and other cvs which have @xmath1 = 3 - 5 hr .
in this paper we briefly reviewed our current understanding of the secular evolution of cvs through their mass transfer phase , including the currently accepted model for the 2 - 3 hr period gap " in the orbital period distribution .
the results of evolution calculations for a representative sample of individual systems are presented , both as functions of time and of orbital period .
a population synthesis code , that starts with some @xmath203 primordial binaries , was then used to generate @xmath204 systems which evolve successfully through the cv phase of mass transfer .
this allows for a more complete exploration of parameter space .
the results are displayed as probability densities in the @xmath11 , @xmath13 , @xmath12 @xmath14 , and @xmath205 planes , for cvs at the current epoch .
this method of displaying the results can lead to considerable insight into the relationships among the various system parameters .
we find that for cvs with orbital periods above 5.5 hr and below the period gap ( but above the period minimum ) the secondary stars closely follow the main - sequence r - m relation ( cf .
beuermann et al .
however , for those with @xmath1 between 3 - 5.5
hr , the effect of bloating causes them to deviate substantially from this same relation . among our more interesting results
, we have shown that the donor star masses in cvs with orbital periods just above the period gap should be as much as @xmath206 lower than would be inferred on the assumption that the donor stars obey a main - sequence radius - mass relation .
this conclusion is only valid if the basic underlying cause of the period gap is thermal bloating of the donor star for systems above the period gap ( see 1 - 6 ) .
on the basis of our results , we have proposed a direct observational test of , in particular , the basic paradigm of the period gap and , more generally , our overall understanding of the evolution of cvs .
this test involves the challenging , but realistic , task of making relatively accurate ( e.g. , @xmath207 ) determinations of the secondary masses in about a half dozen cvs in the period range of 3 - 4 hr . if the masses are consistent with the assumption of a main - sequence radius - mass relation for the donor stars , then the currently accepted explanation of the period gap can not be correct , and the very existence of the gap would pose a major conundrum . if , on the other hand , the masses are mostly consistent with the lower values predicted in this work , then a substantial part of our basic understanding of the secular evolution of cvs will be validated .
previously , much observational attention in cv studies has been focused on determinations of the white dwarf masses . while this is clearly of great interest ,
we hope with this work to stimulate more interest in the important issue of determining the secondary masses .
this research was supported in part by nasa under atp grants gsfc-070 and nag5 - 8500 ( to s.b.h . ) , and nag5 - 7479 and nag5 - 4057 ( to s.a.r . ) .
l.a.n . acknowledges the financial support of nserc ( canada ) and thanks cita and the university of toronto for a reinhardt fellowship and for their hospitality .
we thank m. politano for a number of useful discussions relating to this work .
we are grateful to an anonymous referee who made numerous helpful and insightful comments that led to significant improvements in the paper .
we also thank d. maccannell and g. esquerdo for their technical assistance .
lccccc a & 0.005863 & 0.001251 & 0.02353 & 0.0 & 0.214 + b & 0.4323 & 0.3294 & 0.04942 & 0.005028 & 0.247 + c & 0.1829 & 0.1031 & 0.01041 & 0.0 & 0.220 + d & 0.5280 & 0.3856 & 0.06261 & 0.006076 & 0.230 + main - sequence donor & & & & & 0.35 + .
1 evolution with time of the mass transfer rate , @xmath17 , and orbital period , @xmath1 , for several model cataclysmic variable systems .
left panel - the evolution of a single cv with initial masses ( @xmath208 ; @xmath209 ) .
this system first comes into roche - lobe contact at @xmath210 hr , evolves through the period gap , to the minimum in @xmath1 , and back up to longer periods by @xmath144 yr .
right panel - the evolutions of a selection of four other illustrative initial binary constituent masses , @xmath19,@xmath42 = 0.2,0.4 ( solid ) , 0.35,0.35 ( dotted ) , 0.3,0.6 ( dashed ) , and 0.65,0.7 ( long dashed ) , all in units of @xmath48 .
fig . 2 evolution with orbital period , @xmath1 , of the mass transfer rate , @xmath17 , secondary mass , @xmath19 , and secondary radius , @xmath21 , for several illustrative model cataclysmic variable systems .
the initial masses for the systems whose evolutions are displayed in the left and right sets of panels are the same as described in fig .
1 . fig .
3 computed population of cataclysmic variables at the current epoch in the @xmath11 plane for our standard model ( see table 1 ) . here
@xmath17 is the mass transfer rate , and @xmath1 is the orbital period .
the color represents the logarithm of the number of systems in a particular @xmath11 cell , of which there are 100 per hour interval in @xmath1 and 100 per decade in @xmath17 .
the color scale is given on the right side of the figure .
we note that the scattered , isolated ( red ) points in the image _ below _ the main tracks are minor numerical artifacts of the evolution code that occasionally appear when the roche lobe makes initial contact with the atmosphere of the donor star .
one of these dots corresponds to only @xmath211 cvs in the entire galaxy at the current epoch , and so is of no significance .
4 computed population of cataclysmic variables at the current epoch in the @xmath12 plane for our standard model ( see table 1 ) .
here @xmath21 is the radius of the donor star .
the color represents the logarithm of the number of systems in a particular @xmath12 cell , of which there are 100 per 0.1 @xmath61 and 100 per hour interval in @xmath1 .
the color scale is given on the right side of the figure .
figs . 5 computed population of cataclysmic variables at the current epoch in the @xmath13 plane for our standard model ( see table 1 ) . here
@xmath19 is the mass of the donor star .
the color represents the logarithm of the number of systems in a particular @xmath13 cell , of which there are 100 per 0.1 @xmath48 and 100 per hour interval in @xmath1 .
the color scale is given on the right side of the figure .
computed population of cataclysmic variables at the current epoch in the @xmath14 plane for our standard model ( see table 1 ) ; @xmath212 .
the color represents the logarithm of the number of systems in a particular @xmath14 cell , of which there are 100 per @xmath213 and 100 per hour interval in @xmath1 .
the color scale is given on the right side of the figure .
fig . 7 computed population of cataclysmic variables at the current epoch in the @xmath22 @xmath1 plane ( left panel ) , and the @xmath214 plane ( right panel ) for our standard model ( see table 1 ) .
we show both the stellar luminosity ( top curve ) and the core nuclear luminosity ( lower distributions ) .
the color represents the logarithm of the number of systems in a particular @xmath215 or @xmath216 cell of which there are 100 per decade in @xmath217 , 100 per 500k in @xmath22 , and 100 per hour interval in @xmath1 .
the color scale for both plots is given on the right .
9 computed orbital period distributions for cataclysmic variables at the current epoch . left panel - solid curve is the distribution for all systems that appear in figure 3 ; the dashed curve was produced by scaling the contributions of each system evolved by @xmath176 while the dotted curve is for an @xmath218 scaling ( see text ) .
the @xmath176- and @xmath218-scaled curves have been shifted vertically by arbitrary amounts for ease in comparison .
right panel - solid curve is for all systems in fig .
3 which have not yet reached orbital period minimum ; dashed curve is for systems that have evolved past the orbital period minimum .
fig . 10 computed distributions of the secondary ( right panels ) and
white dwarf masses ( left panels ) in cataclysmic variables at the current epoch .
the mass distributions are ordered according to the range of orbital period .
the dotted histogram ( upper right ) is for post - period minimum cvs and has been arbitrarily divided by 1.5 for presentation purposes .
11 computed distribution of mass ratios in cataclysmic variables at the current epoch .
the top panel is for systems with orbital periods in the range of 1 - 3 hr ( which includes all post period - gap systems ) , while the bottom panel is for systems above the period gap .
12 same as figure ( 5 ) , except that in addition to the standard model ( a ) , the results for three other models are shown ( see tables 1 and 2 ) : ( b ) reduced magnetic braking constant ; ( c ) specific angular momentum lost with the ejected matter is twice that of the white dwarf ; and ( d ) conservative mass transfer and retention by the white dwarf .
13 secondary ( donor ) mass , @xmath19 as a function of orbital period .
the solid curve is based on the assumption that the donor star fills its roche lobe and has a radius - mass relation appropriate to stars on the main sequence ( i.e. , eq . [ 2 ] ) the main - sequence models were generated with the same bipolytrope code that was used to carry out the binary stellar evolution calculations and are discussed in the text .
the dashed curves are polynomial fits to the @xmath13 relations derived from the population synthesis study shown in figure 12 . the labels , a through d , correspond to the four different panels in figure 12 .
starrfield , s. 1998 , in wild stars in the old west : proceedings of the 13th north american workshop on cataclysmic variables and related objects " , asp conference series , vol .
s. howell , e. kuulkers , and c. woodward , p.352 . | we critically examine the basic paradigm for the origin of the 2 - 3 hr period gap in cataclysmic variables ( cvs ) , i.e. , binary systems in which a white dwarf accretes from a relatively unevolved , low - mass donor star . the observed orbital period distribution for @xmath0
cvs shows that these systems typically have orbital periods , @xmath1 , in the range of @xmath2 min to @xmath3 hr , but a distinct dearth of systems with 2 @xmath4(hr ) @xmath5 .
this latter feature of the period distribution is often referred to as the period gap " .
the conventional explanation for the period gap involves a thermal bloating of the donor star for @xmath6 hr due to mass transfer rates which are enhanced over those which could be driven by gravitational radiation ( gr ) losses alone ( e.g. , magnetic braking ) .
if for some reason the supplemental angular momentum losses become substantially reduced when @xmath1 decreases below @xmath7 hr , the donor star will relax thermally and shrink inside of its roche lobe .
this leads to a cessation of mass transfer until gr losses can bring the system into roche - lobe contact again at @xmath8 hr .
we carry out an extensive population synthesis study of cvs starting from @xmath9 primordial binaries , and evolving some @xmath10 surviving systems through their cv phase .
in particular we study current - epoch distributions of cvs in the @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , and @xmath16 planes , where @xmath17 is the mass transfer rate , @xmath18 is the mass ratio @xmath19/@xmath20 , and @xmath19 , @xmath21 , @xmath22 , and @xmath23 are the donor star mass , radius , effective temperature , and luminosity , respectively .
this work presents a new perspective on theoretical studies of the long - term evolution of cvs .
in particular , we show that if the current paradigm is correct , the secondary masses in cvs just above the period gap should be as much as @xmath24 lower than would be inferred if one assumes a main - sequence radius - mass relation for the donor star .
we quantify the @xmath13 relations expected from models wherein the donor stars are thermally bloated .
finally , we propose specific observations , involving the determination of secondary masses in cvs , that would allow for a definitive test of the currently accepted model ( i.e. , interrupted thermal bloating ) for the period gap in cvs . _ the astrophysical journal , in press _
@xmath25present address : physics dept .
, bishop s university , lennoxville , qc , + canada j1 m 1z7 _ subject headings : _ cataclysmic variables @xmath26 stars : binaries :
close @xmath26 stars : evolution @xmath26 stars : mass loss @xmath26 stars : low mass |
laboratory experiments @xcite and numerical simulations @xcite can now systematically explore sustained homogeneous nonequilibrium systems of quite large aspect ratios ( @xmath2 ) which possibly approximate a thermodynamic limit of infinite system size .
these advances raise the theoretical question of identifying order parameters for analyzing and classifying spatiotemporal chaotic states so that a quantitative comparison can be made between theory and experiment @xcite .
the most appropriate order parameter for a given nonequilibrium system is presently not well understood although numerous possibilities have been studied .
some order parameters , such as spatial correlation lengths obtained from exponentially decaying correlation functions , emphasize the average spatial disorder and have been widely used in condensed matter physics @xcite .
others such as the metric entropy and the lyapunov fractal dimension @xcite are familiar from nonlinear dynamics and emphasize the average temporal disorder or dynamical complexity arising from the geometric structure of an attractor in phase space .
we would then like to know whether these different kinds of order parameters are related and whether there is a need for new order parameters .
as an example , does knowledge of an easily measured correlation length give information about the fractal dimension , which is difficult to estimate from experimental time series @xcite ? that a relation between spatial disorder and dynamical complexity may exist is suggested by the prominence in many nonequilibrium states of defects @xcite whose dynamics often determine the average spatial disorder @xcite .
examples of defects are amplitude holes in one - dimensional complex ginzburg - landau equation ( abbreviated below as cgle ) @xcite , domain walls and droplets of opposite spin in coupled map lattices ( cmls ) with an ising - like transition @xcite , vortices in the two - dimensional cgle @xcite , and spirals and centers in the recently discovered spiral - defect chaos state in thermal convection @xcite .
the nucleation , motion , and annihilation of defects are important features of the chaotic dynamics and so dynamical quantities such as the fractal dimension may be related to their spatial statistics @xcite .
complicating this simple picture is the fact that not all fluctuations are associated with defect motion , e.g. , phase fluctuations in the cgle @xcite .
the fractal dimension may then be larger than that suggested by defect statistics . in recent work @xcite , egolf and greenside
have explored the relation between temporal and spatial disorder for the cgle on a large periodic interval @xcite . observing that sufficiently large chaotic systems become extensive so that a fractal dimension @xmath3 grows linearly with volume size @xmath4 ( where @xmath5 is the system size and @xmath6 is the spatial dimensionality ) @xcite ,
they calculated the dimension correlation length @xmath0 which is defined @xcite in terms of the intensive dimension density @xmath7 by the equation @xmath8 the length @xmath0 can be interpreted crudely as a characteristic size of dynamically independent subsystems or a characteristic range of chaotic fluctuations .
near a transition from phase- to defect - turbulent states @xcite , egolf and greenside found that the length @xmath0 was approximately equal to , and had the same parametric dependence as , the spatial correlation length given by the magnitude fluctuations of the ginzburg - landau field @xcite .
over the same parameter range , the correlation length @xmath9 arising from phase fluctuations ( and also of the ginzburg - landau field itself ) was found to increase to quite large values , suggesting that the chaos fluctuations measured by @xmath0 were short - ranged and decoupled from the phase . in this paper ,
the one - dimensional investigations of egolf and greenside @xcite are extended by examining similar questions of how spatial disorder and dynamical complexity are related for spatial dimensionality @xmath10 and @xmath11 .
we study a class of dissipative coupled map lattices @xcite that undergoes an ising - like ferromagnetic - ordering transition as a lattice coupling constant @xmath1 ( defined below in eq .
( [ eq : miller - huse - map ] ) ) is increased through a range of finite positive values , corresponding to a transition from a high - temperature paramagnetic phase to a low - temperature ferromagnetic phase . instead of the point - like space - time defects of the one - dimensional cgle @xcite
, long - lived defects occur in the form of domain walls and droplets involving regions of opposite sign . at a critical transition point
@xmath12 , the usual two - point correlation length @xmath13 diverges to infinity while the magnetization ( the average of the signs of all lattice variables ) bifurcates from a zero to nonzero value .
the main attraction of the miller - huse model @xcite is that a transition with a diverging correlation length occurs from one chaotic state to another .
this permits a careful comparison of different length scales near the transition point . by calculating the lyapunov spectrum and related dynamical quantities such as the dimension correlation length @xmath0
, we show that correlations in chaotic fluctuations near the ordering transition have a short range and are decoupled from the diverging long - range order measured by the length @xmath13 .
as was the case for the 1d periodic cgle @xcite , the length @xmath0 and the correlation length @xmath14 arising from magnitude fluctuations of the cml variables both turn out to be short here about one lattice spacing but a quantitative relation can not be determined since these lengths do not vary substantially with parameters . in one - space dimension , a more substantial variation in these quantities
is obtained by increasing the radius @xmath15 of the coupling from nearest to @xmath15th - nearest lattice neighbors .
with increasing radius @xmath15 , the lengths @xmath0 and @xmath16 then increase approximately linearly and with slopes related by a factor of order one .
this suggests that the two lengths may be related and that @xmath15 is important in determining the length scale of dynamical fluctuations .
several dynamical quantities such as the largest lyapunov exponent @xmath17 , the metric entropy density @xmath18 , and the lyapunov fractal dimension density @xmath19 attain minimum values near but distinctly not at the critical transition point @xmath12 . at first glance , minima in quantities such as @xmath20 or
@xmath19 seem counterintuitive since the onset of ferromagnetic ordering should correspond to increased correlations , i.e. , decreased dynamical complexity and a decreased dimension density or entropy density .
such a monotonic decrease in the metric entropy is , in fact , observed for a non - dissipative equilibrium cml that undergoes an ising - like transition @xcite .
but because the dimension correlation length @xmath0 is quite short in the miller - huse model , we argue below that dynamical quantities are only sensitive to nearest neighbor dynamics .
as the coupling constant @xmath1 increases , the discrete laplacian eventually becomes antidiffusive , magnifying rather than reducing short - wavelength structure , and the dimension density and entropy density start to increase .
the role of coupling lattice neighbors is demonstrated with calculations on periodic 2d hexagonal and 3d cubic lattices .
for these cases , @xmath0 is again about one lattice spacing in size and the extrema in dynamical quantities occur at a value @xmath21 determined by the number @xmath22 of nearest neighbors ( @xmath23 for the square lattice , @xmath24 for the hexagonal and cubic lattices ) .
the positions of the extrema do not coincide with the bifurcation of the magnetization and do not seem to depend on the lattice symmetry and dimensionality . the origin of these minima remains to be explained .
these results , together with previous work on the 1d cgle and with some unpublished work on cmls with algebraic decay of spatial correlations @xcite , suggest that the dimension correlation length will typically be short so that chaotic fluctuations are decoupled from long - range spatial order as measured by correlation functions .
this leads to the qualitative conclusion that defects , although a striking visual feature of spatiotemporal chaos , are not the source of complexity that leads to large fractal dimensions and to small dimension correlation lengths .
the physical meaning and utility of the length @xmath0 remains to be understood and further studies on different kinds of systems will be useful . the rest of this paper is organized as follows . in section [ methods ] , we define the coupled map lattice and discuss some details of how its lyapunov spectrum is calculated using a cm-5 parallel computer @xcite . in section [ results ] , we discuss various results of our simulations , especially the dependence of order parameters on the lattice coupling constant and on the symmetry and dimensionality of the lattice .
finally , in section [ conclusions ] , we summarize our results and relate them to other recent research .
in this section , we define the mathematical models used in our simulations and discuss some details about how the lyapunov spectrum and spatial correlation lengths were calculated numerically . since a cm-5 parallel computer played an important role in our being able to explore large space - time regions for many parameter values , we also discuss some details of how the algorithms were adapted for parallel computation .
as easily simulated models of spatiotemporal chaos , we consider homogeneous coupled map lattices ( cmls ) in which the same chaotic map @xmath25 is associated with each point of a finite periodic lattice and for which nearest neighbor maps are coupled linearly by diffusion .
cmls have a significant advantage over partial differential equations of being analytically amenable @xcite and easier to simulate on a computer .
cmls have the drawback that their solutions can not generally be related quantitatively to experiment and they may not have universal critical properties near transitions @xcite .
we study cmls suggested by recent work of miller and huse @xcite , who analyzed the long - wavelength properties of a two - dimensional cml that orders ferromagnetically , in analogy to the equilibrium ising model @xcite . following these authors
, we use a lattice map @xmath26 with odd symmetry so that domains of opposite `` spin '' or sign arise .
the odd symmetry is a necessary but not sufficient condition @xcite for obtaining an ising - like transition in which the magnetization ( defined below in eq .
( [ eq : magnetization - defn ] ) ) bifurcates to a nonzero value as a coupling constant @xmath1 is varied . following miller and huse
, we choose @xmath25 to be a piecewise - linear map with slope of constant magnitude greater than one , for which only chaotic states of constant measure exist in the absence of lattice coupling .
if @xmath27 denotes the variable at spatial site @xmath28 at integer time @xmath29 ( with @xmath30 , @xmath31 , @xmath32 ) , then the rule for updating each lattice variable to time @xmath33 is given by @xcite : @xmath34 where the parameter @xmath1 is the spatial coupling constant .
the sum goes over indices @xmath35 that denote nearest neighbors sites of site @xmath28 , e.g. , the four nearest neighbors on a square two - dimensional lattice or the six nearest neighbors in a 2d hexagonal lattice or 3d cubic lattice . for most of our calculations
, we used the same local map as in ref .
@xcite : @xmath36 with a slope of constant magnitude equal to 3 . to understand the competition between local chaos and diffusion ( which decrease and increase spatial correlations respectively )
, we also used a more weakly chaotic map with a slope of smaller constant magnitude equal to 1.1 : @xmath37 once a local map @xmath25 has been chosen , a numerical simulation is specified by the dimensionality of the lattice @xmath6 , the symmetry of the lattice ( e.g. , square , hexagonal , or cubic ) , the size of the lattice @xmath5 ( number of sites along an edge ) , the coupling constant @xmath1 , the initial condition @xmath38 , and the total integration time @xmath39 . in this paper , we used an integer lattice in 1d , square and hexagonal lattices in 2d and a cubic lattice in 3d .
initial conditions consisted of assigning a random uniformly - distributed number in the interval @xmath40 $ ] to each site ; results were not dependent on the choice of initial conditions provided the integration time was sufficiently long .
typical integration times were @xmath41 iterations for calculations of lyapunov exponents and @xmath42 iterations for calculations of correlation functions .
we made a few runs with longer integration times of @xmath43 and @xmath44 to check the convergence of the lyapunov exponents and the correlation functions , respectively .
different lattice sizes were used depending on which order parameters were being studied , typically @xmath45 for calculating lyapunov exponents ( which are quite costly to compute ) and @xmath46 for estimating correlation lengths .
many runs were repeated using several lattice sizes to verify the absence of significant finite - size effects . for calculations of statistical averages such as the magnetization and correlation functions ,
the statistics could often be improved substantially by averaging the results of an ensemble of @xmath47 runs each of duration @xmath39 , with each run differing only in the choice of initial conditions .
calculations indicate that this ensemble average is ergodically equivalent to a single integration of duration @xmath48 @xcite . for most of the results reported below , we used an ensemble average over @xmath49 runs which could be executed simultaneously and in parallel on the vector units of a 16-node partition of a cm-5 computer .
some dynamical order parameters can be constructed by combining in different ways the lyapunov exponents @xmath50 associated with a given attractor @xcite . for the cml given by eq .
( [ eq : the - cml ] ) with a total of @xmath51 lattice sites , there are @xmath47 real - valued lyapunov exponents @xmath50 ( labeled in decreasing order @xmath52 ) that characterize the long - time average - rate - of - separation of nearby orbits in phase space . from the @xmath50 , we can calculate a lyapunov dimension @xmath3 given by the kaplan - yorke formula @xcite : @xmath53 and calculate an entropy defined by the sum of the positive exponents @xcite : @xmath54 the number @xmath55 in eq .
( [ eq : dim - defn ] ) is the largest integer such that the sum @xmath56 of the first @xmath55 exponents is nonnegative ; this sum is positive for @xmath57 if an orbit is chaotic ( @xmath58 ) and is negative for @xmath59 if the dynamics is dissipative and so the sum typically crosses zero at an intermediate index @xmath60 for a chaotic dissipative system . the exponents @xmath50 were calculated numerically by a now - standard numerical method @xcite , in which @xmath61 linearizations of the equations of motion , eq . ( [ eq : the - cml ] ) , are evolved in time .
this allows one to follow @xmath55 lyapunov vectors in a tangent space from which local stretching information and the lyapunov exponents can be extracted .
together with a particular nonlinear orbit defined by the equations of motion , a total of @xmath62 cmls was evolved to calculate @xmath55 lyapunov exponents .
repeated orthonormalizations of lyapunov vectors at time intervals @xmath63 are needed to prevent floating - point overflow from the exponentially growing values and to prevent inaccuracies arising from the loss of linear independence as they fold up along the direction of the fasting growing exponent @xcite .
for the maps eqs .
( [ eq : miller - huse - map ] ) and ( [ eq : slope-1.1-map ] ) , we found empirically that values @xmath64 gave reasonable results for all lattices studied with the largest value of @xmath63 depending on the parameters @xmath1 and @xmath5 .
smaller renormalization times did not change the values of the lyapunov spectrum ( although the code was more costly to run ) while larger values led to serious errors due to linear dependence of the lyapunov vectors .
the orthonormalizations of tangent vectors consumed most of the computing time on a thinking machines cm-5 parallel computer .
the orthonormalizations require substantial communication between different processors since each processor evolves independently only a few of the @xmath55 linearized equations in its own local memory ; this communication decreases the efficiency of the code .
although the communication inherent in the orthonormalization procedure could not be avoided , in all other portions of the code communication between nodes was eliminated by iterating redundantly an identical copy of the nonlinear cml eq .
( [ eq : the - cml ] ) with identical initial conditions on each processor . in this way
, information about the underlying orbit ( needed when iterating the linearized cmls ) did not have to be communicated from one processor to all others at each time step . by monitoring the lyapunov exponents and the lyapunov dimension as a function of time @xmath29 , we found empirically that an integration time @xmath65 time steps gave an acceptable relative accuracy of better than one percent for calculating the dimension @xmath3 and entropy @xmath66 for all lattices studied ( @xmath67 ) .
[ fig : dim - vs - time ] shows how the dimension @xmath3 converges over time for lattice size @xmath68 and for @xmath69 .
the dimension curve is noisy with fluctuations that diminish slowly over time ( the @xmath50 , not shown , have substantially noiser time series ) .
goldhirsch et al @xcite have argued that the amplitude of the exponent fluctuations should decay as @xmath70 where @xmath39 is the total integration time and so one could fit and extrapolate to get a better estimate @xcite . extrapolation was not needed in plots like fig .
[ fig : dim - vs - time ] which already give an adequate relative accuracy of better than one percent . by repeating plots such as fig .
[ fig : dim - vs - time ] for different system sizes @xmath5 with all other parameters held fixed , we found extensive scaling of the dimension @xmath3 with the volume of the system @xmath71 for a wide range of parameter values @xmath1 , an example of which is given in fig .
[ fig : ext - chaos ] . fig .
[ fig : ext - chaos](a ) shows that the dimension @xmath3 increases linearly and extensively with @xmath47 beyond a system size of about @xmath72 . the lyapunov dimension density @xmath19
could then be obtained from the slope of a least - squares fitted line in the extensive region .
the intercept of a least - squares - fitted line through the four right - most points was @xmath73 which is quite small ( and is also approximately zero for extensively chaotic solutions of the 1d cgle @xcite ) .
there is then the possibility that a single dimension calculation for a sufficiently large system may suffice to estimate its dimension density . by comparing the dimension per volume @xmath74 with the dimension density @xmath19 , fig .
[ fig : ext - chaos](b ) shows how the extensive regime is approached rapidly and achieved for fairly small system sizes @xmath75 .
we finish this subsection with two comments about the meaning of the lyapunov dimension @xmath3 and about why we chose to calculate @xmath3 from lyapunov vectors rather than from time series measurements . the reader should recall that there is an infinity of fractal dimensions @xmath76 ( often called the renyi dimensions ) associated with a strange attractor with the parameter @xmath77 varying over the real numbers @xcite . since each dimension @xmath76 will be extensive for a homogeneous extensively chaotic system ,
the particular values only reflect the system size and are not interesting themselves .
instead , one should define a continuum of corresponding intensive dimension densities @xmath78 to provide a partial characterization of such systems
. for large fractal dimensions ( @xmath79 ) , present computers and algorithms only allow the calculation of the lyapunov fractal dimension eq .
( [ eq : dim - defn ] ) and its corresponding density .
the lyapunov dimension is conjectured to be the same as the renyi dimension with @xmath80 ( i.e. , the information dimension @xmath81 @xcite ) and it is not known to what extent the corresponding density @xmath82 characterizes the unknown function of densities @xmath83 , e.g. , whether it is close to the mean value of the @xmath83 .
there are one - dimensional maps for which the ratio of @xmath81 to @xmath84 ( the two most commonly calculated fractal dimensions ) can be arbitrarily large @xcite , and so the variation of the function @xmath83 around its mean value can be large .
a perhaps even more important question is whether the different dimension densities @xmath83 each have a similar dependence on model parameters , e.g. , all increasing or decreasing together .
until this issue is resolved , the dimension density @xmath82 and the corresponding length @xmath85 need to be interpreted with caution .
the dynamical quantities @xmath3 and @xmath66 are calculated in terms of the lyapunov spectrum @xmath50 , and not in terms of time series @xmath86 at a given lattice site @xmath28 , because of the impractical computational demands of time series algorithms @xcite .
while the computational complexity of the method based on the lyapunov spectrum scales algebraically with system volume @xmath87 or dimension @xmath3 @xcite , it is well known that the computational complexity of classical time series algorithms such as that proposed by procaccia and grassberger grows _ exponentially _ with @xmath3 , imposing severe restrictions on the largest dimension that can be estimated from experimental data .
although it remains controversial what is the practical upper bound for clean time series of less than a million points ( there are claims from 6 to 20 @xcite ) , existing time series methods can not treat extensively chaotic systems whose fractal dimensions may be in the hundreds ( see fig . [
fig : ext - chaos ] ) .
in addition to the dynamical quantities described in the previous section , we quantify the evolution of the cmls by a `` magnetization '' @xmath88 and by length scales measured from two - point and mutual information correlation functions .
we describe these briefly to indicate the method and errors involved . following miller and huse @xcite , the average magnetization @xmath88 of the cml
is defined by a space - time and ensemble average of the signs @xmath89 of the lattice values @xmath27 , @xmath90 where the index @xmath91 labels a particular cml running on processor @xmath91 .
an average over 64 independent cmls running on a 16-node partition of a cm-5 was typically used . for the local maps
( [ eq : miller - huse - map ] ) and eq .
( [ eq : slope-1.1-map ] ) , @xmath88 undergoes a pitchfork bifurcation from a zero to finite value as the lattice coupling @xmath1 is increased from small values .
the critical value @xmath92 at which @xmath88 bifurcates to a nonzero value is unchanged if the values of the site variables @xmath86 are used instead of their signs in eq .
( [ eq : magnetization - defn ] ) .
the bifurcation of @xmath88 to a nonzero value defines the onset of ferromagnetic order at @xmath12 @xcite as illustrated in fig .
[ fig : m - for - square - lattice ] . to characterize the average spatial disorder , we examined two of many possible definitions of spatial correlation lengths , one from an exponentially decaying two - point correlation function , another from an exponentially decaying mutual information function @xcite .
( some other correlation lengths are discussed on pages 945 - 947 of ref .
@xcite . ) the two - point correlation function was defined in the usual way : @xmath93 where @xmath94 denotes the position of lattice point @xmath28 and where the brackets @xmath95 denote the averaging process of eq .
( [ eq : magnetization - defn ] ) . given the periodicity of the lattice and the availability of efficient parallel fast fourier transforms ( ffts ) on the cm-5 , we calculated eq .
( [ eq:2-pnt - correlation - fn ] ) via the wiener - khintchin theorem @xcite , first obtaining the time - averaged magnitude squared of the fourier coefficients , from which eq .
( [ eq:2-pnt - correlation - fn ] ) was obtained by an inverse fft . in most cases
, there was a substantial region of exponential decay from which the two - point correlation length @xmath13 was obtained by a least - squares fit of the form @xmath96 ; a representative plot is given in fig .
[ fig:2-point - corr - fn ] . the correlation functions and corresponding values of @xmath13 do not change if eq .
( [ eq:2-pnt - correlation - fn ] ) is defined in terms of the sign of the variables , @xmath97 .
the two - point correlation functions decay more rapidly as shown in fig .
[ fig:2-point - corr - mag - fn ] if the signs of the field values @xmath98 in eq .
( [ eq:2-pnt - correlation - fn ] ) are replaced by their magnitudes @xmath99 .
the correlation length @xmath100 obtained from the initial rapid exponential decay is approximately one lattice spacing and changes little when the coupling constant @xmath1 is varied over a large range , including near the bifurcation point @xmath12 .
for dimensionality @xmath101 , we show below that this short length scale @xmath16 is related to the dimension correlation length @xmath0 and that both vary linearly with the radius @xmath15 of neighboring lattice sites that are coupled together spatially . as a possible alternative for characterizing the spatial disorder of a nonlinear system , we also calculated a correlation length @xmath102 based on the exponential decay of the mutual information function @xmath103 @xcite of the variables @xmath86 .
[ fig : mutual - info - decay ] shows the exponential decay of a mutual information function @xmath104 for two - dimensional square lattice of size @xmath105 and for the parameter @xmath106 . again
if the magnitudes of the field variables are used when calculating @xmath103 , the exponential decay is much more rapid , with @xmath107 being approximately one lattice spacing for a wide range in @xmath1 .
although there is not yet a compelling theoretical reason to prefer @xmath102 over other correlation lengths such as @xmath13 @xcite , the former has the distinction of depending nonlinearly on the dynamics and so may depend on details that are missed by @xmath13 .
for this reason , an increasing number of scientists have reported correlation lengths in terms of @xmath102 @xcite . as shown in fig .
[ fig : mi - vs-2-point ] for the two - dimensional miller - huse cml on a square lattice , the length scales @xmath13 and @xmath102 are linearly related over a substantial dynamical range near the ferromagnetic transition . at least for the present models , these lengths are equivalent measures of spatial disorder and we report values only for @xmath13 below .
in this section , we discuss our calculations of the lyapunov spectrum and of correlation lengths .
our goal is to explore how spatial disorder ( as characterized by the two - point correlation length or by the mutual information correlation length ) is related to dynamical complexity ( as measured by the intensive dimension density and by the dimension correlation length eq .
( [ xi - delta - defn ] ) ) and to investigate how these order parameters vary near the nonequilibrium transition point @xmath12 at which the magnetization bifurcates to nonzero values ( fig . [ fig : m - for - square - lattice ] ) .
results for the 2d square lattice are given first , followed by results for lattices with different symmetries and dimensionalities .
we do not address issues related to critical exponents of these different models which have been discussed by miller and huse @xcite and more recently by marcq and chat @xcite .
related interesting results on similar cmls have also recently been reported by boldrighini et al @xcite .
for the two - dimensional periodic square lattice with map eq .
( [ eq : miller - huse - map ] ) , fig .
[ fig : m - for - square - lattice ] shows that there is a bifurcation at @xmath108 .
this bifurcation corresponds to the onset of long - range order of the lattice variables @xmath86 as demonstrated by the divergence of the two - point correlation length @xmath13 as @xmath1 approaches @xmath92 ( fig . [ fig : divergence - of - xi_2](a ) ) . over this same parameter range ,
the dimension correlation length @xmath0 varies smoothly ( fig .
[ fig : divergence - of - xi_2](b ) ) , deviating by less than four percent from a value of one lattice spacing and attaining a maximum value close to where the correlation length diverges .
the lyapunov spectrum of exponents also varies smoothly from one side of the transition to the other as shown in fig .
[ fig : lyapunov - spectra ] .
we conclude that chaotic fluctuations have a short range , are decoupled from the onset of long - range order measured by @xmath13 , and that the spectrum of exponents is at most weakly dependent on the onset of long - range spatial order . to understand further how various dynamical quantities change near the transition point , we have plotted in fig .
[ fig : d - h - l1-vs - g-2d - square ] the variation of the lyapunov fractal dimension density @xmath19 , of the metric entropy density @xmath20 , and of the largest lyapunov exponent @xmath17 across the ferromagnetic transition for a lattice of size @xmath68 , which is already extensively chaotic according to fig .
[ fig : ext - chaos ] . as was the case for the length @xmath0 in fig .
[ fig : divergence - of - xi_2](b ) , these quantities change by only a small amount through the transition ( at most by 20% ) and all go through a minimum close to , but distinct from , the ferromagnetic transition at @xmath109 .
this result was surprising to us since one consequence of coupling neighboring maps more strongly ( increasing the parameter @xmath1 ) would intuitively be to increase correlations between their dynamics , which should decrease both @xmath19 and @xmath20 .
[ fig : d - h - l1-vs - g-2d - square](a ) indicates that roughly one quarter of the maximum number of degrees of freedom disappear when the lattice attains its minimum dimension density of @xmath110 .
( an upper bound of @xmath111 is set by the integer lattice spacing . )
it is not clear why the fractal dimension density and other dynamical quantities have extrema near @xmath112 . for an equilibrium non - dissipative cml of ising dynamics on a two - dimensional square lattice ,
sakaguchi @xcite did not find a local minimum in the entropy @xmath66 but instead found a monotonic decrease consistent with the analytical solution of the spin-@xmath113 ising model on a square lattice @xcite .
one explanation for the extrema may be that the dissipative linear coupling in eq .
( [ eq : the - cml ] ) becomes antidiffusive for @xmath114 , enhancing rather than damping short - wavelength fluctuations and so decorrelating nearby lattice variables .
the issue is somewhat more subtle than this because the existence of the minimum depends also on details of the local map @xmath25 in eq .
( [ eq : the - cml ] ) . for the less chaotic lattice map eq .
( [ eq : slope-1.1-map ] ) with slope of constant magnitude 1.1 , an ising - like transition still occurs as shown by the bifurcation of the magnetization near @xmath115 in fig .
[ fig : m - for - square - lattice - map2 ] . fig .
[ fig : d - h - l1-for - square - lattice - map2 ] now shows that the lyapunov dimension density @xmath19 and entropy density @xmath20 decrease monotonically as the parameter @xmath1 is increased , with the largest exponent @xmath17 remaining constant .
figures [ fig : m - for - square - lattice ] , [ fig : divergence - of - xi_2](a ) , and [ fig : d - h - l1-vs - g-2d - square ] suggest that the extrema of dynamical quantities may be related to the ferromagnetic transition . on the other hand , the short dimension correlation length in fig .
[ fig : divergence - of - xi_2](b ) contradicts this by implying that chaotic fluctuations occur over a length scale that is short compared to the ferromagnetic ordering . to understand this further
, we have explored cmls of different symmetry and dimensionality and found that the near - proximity of the extrema with the transition is a coincidence for the two - dimensional lattice with square symmetry .
more generally , the positions of extrema seem to be determined simply by the number of nearest neighbors @xmath22 , and not by the symmetry or dimensionality of the cml or by the position of the magnetization bifurcation point .
[ fig : m - for - hexagonal - lattice - map1 ] summarizes calculations for a two - dimensional periodic hexagonal lattice by plotting the dependence of magnetization @xmath88 and of dimension correlation length @xmath0 on the coupling constant @xmath1 .
the magnetization bifurcates to a nonzero value at @xmath116 which is a smaller value than that for the square lattice since the larger number of nearest neighbors ( six versus four ) increases the effective strength of the diffusive coupling . the relative difference between the transition at @xmath12 and the positions of the extrema in @xmath0 and in related dynamical quantities is substantially larger than was the case for the 2d square lattice . for the hexagonal lattice , extrema in quantities like
the length @xmath0 occur at a value close to @xmath117 where @xmath24 is the number of nearest neighbors .
a similar result is found for the same cml on a 3d cubic lattice , as shown in fig .
[ fig : m - for - cubic - lattice - map1 ] .
the transition at @xmath118 occurs at a value close to but smaller than the value on the hexagonal lattice .
extrema in the dynamical quantities like @xmath0 again occur at a value close to @xmath119 with @xmath24 . that the positions of the extrema of dynamical quantities is dependent primarily on the number of nearest neighbors is a consequence of the nearest - neighbor diffusive coupling in eq .
( [ eq : the - cml ] ) and of the fact that , for the local map eq .
( [ eq : miller - huse - map ] ) , the chaos is sufficiently strong to make the dimension correlation length @xmath0 quite small , about one lattice length .
that the positions of the extrema seem to be given quantitatively by the specific formula @xmath120 is more delicate to understand but may be related to the fact that the discrete laplacian operator changes from diffusive to antidiffusive behavior at this value .
the value @xmath120 is the value for which the weight of each of the neighbors is equal to the weight of the central lattice site to be updated .
for all cmls that we studied , the dimension correlation length @xmath0 was about one lattice spacing in size and this was also the `` radius '' of the diffusive coupling in eq .
( [ eq : the - cml ] ) .
this suggests that the length @xmath0 may be determined by the spatial extent of the diffusive coupling , becoming larger as more sites are coupled to a given site .
this conjecture was tested in one - space dimension by coupling together , with equal weight @xmath1 , all lattice variables within a radius @xmath15 of a given site @xmath28 : @xmath121 for this one - dimensional periodic cml with the lattice map eq .
( [ eq : miller - huse - map ] ) , the dynamics varies in a complicated way with radius @xmath15 . for most initial conditions ,
chaos was found for smaller radii ( @xmath122 ) . for larger radii @xmath123 , the transients lasted much longer and the asymptotic dynamics was periodic .
as an example , for @xmath124 the dimension as a function of time initially reached a value @xmath125 even after @xmath126 transient iterations were skipped ; however , the dimension then decreased steadily to zero over the next 30,000 iterations .
we believe that this asymptotic periodic behavior is a finite - size effect . for a sufficiently large system size @xmath5 , with the crossover length increasing with the radius @xmath15
, the asymptotic state should be chaotic . for dimensionality @xmath101
, the dimension correlation length @xmath0 varies more strongly with increasing radius @xmath15 than with coupling constant @xmath1 , which allows several different length scales to be compared .
[ fig:1d - cml - results ] shows that the two - point correlation length @xmath100 obtained using the magnitudes @xmath127 of the field values has approximately the same linear dependence on the coupling radius @xmath15 as the dimension correlation length @xmath0 .
in addition , these two length scales are the same order of magnitude .
the two - point correlation length @xmath13 obtained using the actual field values is larger than @xmath0 and does not have the same simple linear dependence on @xmath15 .
the situation is then similar to results found for spatiotemporal chaotic solutions of the 1d periodic cgle @xcite in that the spatial correlation length of fluctuations in the magnitude of a field provides a way of estimating the length @xmath0 .
in this paper , we have extended recent calculations @xcite concerning the relation between spatial disorder and dynamical complexity of a sustained homogeneous nonequilibrium system from dimensionality @xmath101 to dimensionalities @xmath10 and @xmath11 .
this was accomplished by choosing a coupled map lattice , eq .
( [ eq : the - cml ] ) , that underwent an ising - like transition with diverging two - point correlation length as a parameter @xmath1 was varied @xcite . by comparing various length scales such as the two - point correlation length @xmath13 , the dimension correlation length @xmath0 , and the two - point correlation length of magnitude fluctuations @xmath14 near the transition point ,
we were able to show that the lengths @xmath0 and @xmath14 were short , of order one lattice spacing , even as the length @xmath13 diverged to infinity . in agreement with previous work @xcite ,
the chaotic fluctuations are decoupled from the average long - range spatial order .
the correlation length of magnitude fluctuations seems to provide an effective way to estimate the size of the length @xmath0 .
our calculations of the lyapunov spectrum and related quantities such as the lyapunov dimension density @xmath19 , metric entropy density @xmath20 , and the largest lyapunov exponent @xmath17 show that the onset of long - range spatial order ( diverging @xmath13 ) does not affect dynamical invariants , which vary smoothly and weakly through the transition point @xmath12 .
thus the average spatial disorder ( measured by @xmath13 ) does not determine dynamical complexity ( measured by @xmath0 ) .
rather surprisingly , the intensive densities @xmath19 and @xmath20 go through a minimum near the transition point so that the onset of long - range order does not correspond to a decrease in complexity . by examining cmls of different symmetry and of different dimensionality @xmath6
, we showed that the positions @xmath1 of the extrema were determined by the number @xmath22 of neighbors nearest to a given lattice site ( with @xmath128 ) but not by the symmetry or by @xmath6 .
this result can be understood as a consequence of the extremely short dimension correlation length @xmath0 , about one lattice size , so that lattice variables are independent except when they are nearest neighbors .
we believe that the minima in @xmath19 and in @xmath20 occur approximately when the discrete laplacian in eq .
( [ eq : the - cml ] ) becomes antidiffusive with increasing parameter @xmath1 .
short - wavelength fluctuations are then magnified instead of damped , decreasing correlations between neighboring sites .
some of our results concerning extrema in dynamical quantities have been independently obtained by boldrighini et al @xcite although these authors worked with extensive , rather than intensive , quantities and they did not determine whether their calculations corresponded to extensively chaotic regimes .
boldrighini et al investigated cmls of the form eq .
( [ eq : the - cml ] ) for dimensionalities @xmath101 and @xmath10 but with some new local maps @xmath25 .
besides also finding extrema in the lyapunov fractal dimension , boldrighini et al showed that the odd symmetry of the map @xmath25 was not a sufficient condition for the magnetization to bifurcate to a nonzero value . using a strongly chaotic local map with slope of constant magnitude equal to 5
, they also showed that the magnetization @xmath88 did not bifurcate to a nonzero value if the local map were made sufficiently chaotic compared to the ordering caused by diffusion . in section [ other - lattices ] , we used the less - chaotic local map eq .
( [ eq : slope-1.1-map ] ) with slope of constant magnitude 1.1 to show that the dimension density @xmath19 can decrease monotonically without a minimum even when the magnetization @xmath88 bifurcates to a nonzero value .
the dependence of these minima on details of the local map is not yet understood and should be pursued with further studies .
the small values of @xmath0 in the present cmls , in the 1d cgle @xcite , and in a nonequilibrium cml with algebraic decay of spatial correlations @xcite have several interesting implications .
one is that many previous laboratory experiments @xcite and numerical simulations @xcite concerning spatiotemporal chaos are likely already extensive so that it is meaningful to talk about dimension and entropy densities ( see also the earlier paper by bohr @xcite . ) a second implication is that the dimension correlation length @xmath0 may not be a useful order parameter for future studies of spatiotemporal chaos since it depends only weakly on parameters .
a third implication is that the short value of @xmath0 suggests the nonexistence of macroscopic chaotic states for dynamics with local interactions , a point already made by several researchers @xcite . finally , we speculate that @xmath0 is the length scale below which one can replace chaotic fluctuations by a white noise source when trying to develop a hydrodynamic ( long wavelength ) description of spatiotemporal chaos ( see the discussion on pages 953 - 954 in ref .
@xcite ) .
the short values of @xmath0 raise the question of what determines this length scale .
our calculations on the 1d cml eq .
( [ eq : radius - r - cml ] ) with a variable radius of coupling @xmath15 suggest that the length @xmath0 is determined partly by the length @xmath14 characterizing magnitude fluctuations although the reason for and the generality of this correspondence is not understood @xcite .
the length @xmath0 is also related to the radius @xmath15 over which nearby lattice sites are coupled together ( fig .
[ fig:1d - cml - results ] ) .
further calculations with different kinds local maps and of diffusive operators and for different values of @xmath15 should provide further insight .
it is appropriate to finish with a discussion about the relevance of these results for laboratory experiments . as discussed at the end of section [ exponents ] , it does not seem possible in the near future to calculate the lyapunov spectrum , the fractal dimension , or the fractal dimension density of a high - dimensional extensively chaotic experimental system for which only time series measurements are available @xcite .
our success in calculating these quantities was a result of having explicit knowledge of the dynamical equations which could then be integrated numerically on a powerful parallel computer using algorithms whose complexity only grew algebraically with the dimension @xmath3 @xcite . for many laboratory experiments ,
a quantitative mathematical description is either lacking ( e.g. , for chemical reactions ) or , if known , is too difficult to work with numerically ( e.g. , the five three - dimensional boussinesq equations describing buoyancy - induced convection in a large - aspect - ratio container @xcite ) . our calculations in section [ results ] suggest that one possible way to estimate the dimension correlation length @xmath0 may be to calculate the correlation length of some function of the physical fields , e.g. , the field magnitude
. another possibility will be to discover and to validate algorithms that can calculate the intensive dimension density eq .
( [ delta - defn ] ) directly from time - series measurements that are localized in space @xcite , in lieu of calculating a large extensive fractal dimension @xmath3 and then dividing by the extensive system volume @xmath87 .
several steps have been taken in this direction @xcite , but a theoretical foundation has not yet been established nor have the numerical algorithms been adequately tested .
we thank l. bunimovich , h. chat , p. hohenberg , d. huse , h. riecke , and j. socolar for useful discussions .
this work was supported by grants nsf - cda-91 - 23483 and nsf - dms-93 - 07893 of the national science foundation , by grant doe - de - fg05 - 94er25214 of the department of energy and by an allotment of cm-5 computer time at the national center for supercomputing applications . the second author ( d.a.e . )
would like to thank the office of naval research for fellowship support .
much of the research reported here was carried out by the first author as part of his undergraduate honors thesis in physics at duke university .
j. p. gollub and r. ramshankar , in _
new perspectives in turbulence _ , edited by s. orszag and l. sirovich ( springer - verlag , berlin , 1990 ) , pp.165194 ; s. w. morris , e. bodenschatz , d. s. cannell , and g. ahlers , phys .
* 71 * , 2026 ( 1993 ) ; m. assenheimer and v. steinberg , nature * 367 * , 345 ( 1994 ) ; f. t. arecchi _
et al . _ , physica d * 61 * , 25 ( 1992 ) ; q. ouyang and h. l.swinney , chaos * 1 * , 411 ( 1991 ) . b. i. shraiman
_ et al . _ , physica d * 57 * , 241 ( 1992 ) ; h .- w .
xi , j. d. gunton , and j. vials , phys .
lett . * 71 * , 2030 ( 1993 ) ; j. e. pearson , science * 261 * , 189 ( 1993 ) ; d. a. egolf and h. s. greenside , nature * 369 * , 129 ( 1994 ) ; w. decker , w. pesch , and a. weber , phys .
lett . , 648 ( 1994 ) ; m. cross , d. meiron , and y. tu , chaos * 4 * , ( 1994 ) .
s. w. morris , e. bodenschatz , d. s. cannell , and g. ahlers , phys .
* 71 * , 2026 ( 1993 ) ; m. assenheimer and v. steinberg , phys .
rev . lett . * 70 * , 3888 ( 1993 ) ; w. decker , w. pesch , and a. weber , phys .
lett . * 73 * , 648 ( 1994 ) .
h. chat , in _ spatio - temporal patterns in nonequilibrium complex systems _ , vol .
xxi of _ santa fe institute studies in the science of complexity _ , santa fe institute , edited by p. e. cladis and p. palffy - muhoray ( addison - wesley , reading , massachusetts , 1995 ) , pp .
3350 . for a numerical dynamical model with @xmath47 independent variables
, an optimal integration scheme will advance the variables in one time step with a computational effort that grows at best as slowly as @xmath129 .
each linearized equation also involves @xmath47 variables and so the effort needed to calculate @xmath55 lyapunov exponents by the algorithm discussed in section [ methods ] over a fixed time time @xmath39 will grow at best as slowly as @xmath130 .
since the number of exponents needed to estimate the lyapunov dimension grows linearly with system volume , as does the number @xmath47 itself , we conclude that the computational effort to calculate the lyapunov dimension for an extensively chaotic system , for an observation time @xmath39 , will grow at best as slowly as @xmath131 , i.e. , algebraically with @xmath47 .
it is presently not understood how the total integration time @xmath39 should itself scale with system volume or with @xmath47 and so the exponent of the algebraic scaling with dimension @xmath3 is not yet known @xcite .
y. pomeau , acad .
paris * 300 * , ser .
ii , 239 ( 1985 ) ; p. grassberger phys .
scripta * 40 * , 346 ( 1989 ) ; a. torcini , a. politi , g. p. puccioni , and g. dalessendro , physica d * 53 * , 85 ( 1991 ) ; l. s.tsimring , phys .
e * 48 * , 3421 ( 1993 ) ; m. bauer , h. heng , and w. martienssen , phys .
rev.lett . * 71 * , 521 ( 1993 ) . | by simulating a nonequilibrium coupled map lattice that undergoes an ising - like phase transition , we show that the lyapunov spectrum and related dynamical quantities such as the dimension correlation length @xmath0 are insensitive to the onset of long - range ferromagnetic order . as a function of lattice coupling constant @xmath1 and for certain lattice maps , the lyapunov dimension density and other dynamical order parameters go through a minimum .
the occurrence of this minimum as a function of @xmath1 depends on the number of nearest neighbors of a lattice point but not on the lattice symmetry , on the lattice dimensionality or on the position of the ising - like transition . in one - space dimension , the spatial correlation length associated with magnitude fluctuations and the length @xmath0 are approximately equal , with both varying linearly with the radius of the lattice coupling . |
this supplementary information gives more details about realization of fault - tolerant quantum random number generator through topological manipulation of majorana fermions . in sec .
i , we give the detailed proof on how to certify genuine randomness through observation of violation of the mabk inequality . in sec .
ii , we summarize the topological properties of majorana fermions and show the implementation of the necessary topological quantum gates on the logic qubits encoded with these majorana fermions . [ [ randomness - certified - by - observation - of - violation - of - the - mabk - inequality ] ] randomness certified by observation of violation of the mabk inequality ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section , we establish a link between randomness of the measurement outputs of a quantum system and violation of the mabk inequality . a link between randomness and violation of the bell - chsh inequality has been established in ref .
@xcite . here
, we generalize the result from the two - qubit chsh inequality to the three - qubit mabkinequality .
consider a quantum nonlocality test on three qubits .
each qubit has two settings of two - outcome measurements , denoted by @xmath82 , respectively for the three qubits .
the measurement outputs @xmath83 of this quantum system are characterized by the joint probability distribution @xmath84 .
randomness of the outputs @xmath83 are quantified by the min - entropy , defined as @xmath85 $ ] . with an experimental observation of violation @xmath29 of the mabk inequality ,
our aim is to find a lower bound on the min - entropy @xmath86this is equivalent to solving of the following optimization problem 2010pironio : @xmath87where @xmath88 is defined in eq.(2 ) of the main text and @xmath89 constitutes a quantum realization of the bell scenario @xcite .
thus , the minimal value of @xmath90 compatible with the mabk violation @xmath29 and quantum theory is given by @xmath91 $ ] .
consequently , to obtain @xmath40 we only need to solve ( [ min - entropy - single ] ) for all possible input and output triplets @xmath92 and @xmath75 . this can be effectively done by casting it to a _ semi - definite program _ ( sdp )
an infinite hierarchy of conditions that need to be satisfied by all quantum correlations are introduced in ref .
all these conditions can be transformed to a sdp problem and the hierarchy is complete in the asymptotic limit , i.e. , it guarantees existence of a quantum realization if all the conditions in the hierarchy are satisfied .
generally , conditions higher in the hierarchy are more constraining and thus better reflect the constraints in ( [ min - entropy - single ] ) and give a tighter lower bound . to obtain a lower bound of the min - entropy for a given mabk violation @xmath29
, we use the matlab toolbox sedumi @xcite and solve the sdp corresponding to the certificates between order @xmath65 and order @xmath93 @xcite .
the result is plotted in fig.1 in the main text . from the figure
, @xmath40 equals zero at the classical point @xmath94 and increases monotonously as the mabk violation @xmath29 increases . for the maximal violation @xmath79 , @xmath95 , corresponding to @xmath96 bits .
equation ( 4 ) in the main text can be derived using arguments similar to those in ref .
the difference is that the bell scenario in refs .
@xcite is based on the two - qubit chsh inequality , which needs to be extended in our scheme with the three - qubit mabk inequality .
suppose we run the experiments @xmath17 times and denote the input and output string as @xmath19 and @xmath20 , respectively .
as in the main text , let @xmath25 be a series of mabk violation thresholds , and denote @xmath28 the probability that the observed kcbs violation @xmath29 lies in the interval @xmath30 . denote by @xmath32 the possible classical side information an adversary may have . to derive eq . ( 4 ) in the main text ,
let us first introduce the following theorem : * theorem 1*. suppose the experiments are carried out @xmath17 times and each triplet of inputs @xmath97 is generated independently with probability @xmath18 .
let @xmath98 , @xmath99 be two arbitrary parameters and @xmath100 , then the distribution @xmath101 characterizing @xmath17 successive use of the devices is @xmath38-close to a distribution @xmath102 such that , either @xmath103 or @xmath104where @xmath105 . equation ( [ theorem1bound ] ) is equivalent to eq .
( 4 ) in the main text .
theorem 1 tells us that the distribution @xmath106 , which characterizes the output @xmath24 of the device and its correlation with the input @xmath71 and the adversary s classical side information @xmath32 , is basically indistinguishable from a distribution @xmath102 that will be defined below @xcite .
if we find that the observed mabk violation @xmath29 lies in @xmath30 with a non - negligible probability , i.e. , @xmath34 , the entropy of the outputs @xmath24 is guaranteed to have a positive lower bound @xmath107 , that is , the randomness of the outputs is guaranteed to be larger than @xmath108 up to epsilonic correction . _
proof_. we use a procedure similar to those in ref . 2012pironio to prove the above theorem .
let us define a function @xmath109 , which is concave and monotonically decreasing given by the solution of the optimization problem in eq .
( [ min - entropy - single ] ) ( shown in fig . 1 of the main text ) .
denote by @xmath110 ( @xmath111 ) the string of outputs before the @xmath112th round of experiment ( similarly , @xmath113 denotes the string of inputs ) .
we introduce an indicator function @xmath114 as : @xmath115 if the event @xmath116 happens and @xmath117 otherwise .
consider the following random variable @xmath118where @xmath119 and @xmath12 are defined in the main text , and @xmath120 if @xmath15 is @xmath121 and @xmath122 if @xmath15 is @xmath123 .
it is easy to check that eq.([randomvar ] ) reduces to the mabk expression ( 2 ) in the main text and the expectation value of @xmath124 conditional on the past @xmath125 is equal to @xmath126 , i.e. , @xmath127 .
we use @xmath128 to denote all the events before the @xmath129th round of experiment and the possible adversary s classical side information .
the estimator of the mabk violation can be defined as : @xmath130 . with these notations ,
first we introduce two lemmas for proof of the main theorem . from the solution to the optimization problem in eq .
( min - entropy - single ) , the probability @xmath138 is bounded by a function of the mabk violation @xmath126 : @xmath139 .
thus , we have : @xmath140here , to obtain the second inequality , we have used the equality @xmath127 and the fact that @xmath141 is logarithmically concave and monotonically decreasing .
the third inequality is obtained from the definition of @xmath142 and the fact that @xmath141 is decreasing . to get eq .
( [ prt1 ] ) , we can define another random variable @xmath143 .
then it is easy to verify that ( i ) @xmath144 , ( ii ) @xmath145 , and ( iii ) @xmath146 .
thus , the sequence @xmath147 is a martingale process @xcite . applying the azuma - hoeffding inequality @xmath148 1967azuma,1960hoeffding,2001grimmett - book , we have @xmath149where @xmath105 .
equation ( margeq1 ) combined with the definition of @xmath142 gives eq .
lemma 1 is thus proved . in the above proof , we only considered the case that the random variable sequence @xmath24 takes values in the output space @xmath150 .
as in ref .
@xcite , we can extend the range of @xmath24 to include abort - output " @xmath151 , and view @xmath24 as an element of @xmath152 with @xmath153 if @xmath154 . the physical meaning of @xmath151 is that when @xmath151 is produced by the device , then no mabk violation has been obtained and no randomness is certified .
* lemma 2*. there exists a probability distribution @xmath155 , which is @xmath156-close to @xmath157 , i.e. , @xmath158 , and satisfies the following condition @xmath159for all @xmath160 such that @xmath161 . _
proof_. we show how to construct a probability distribution satisfying the above two conditions . to this end
, we introduce @xmath162 .
@xmath163 is defined as : @xmath164then by lemma 1 , it is straightforward to get that the distribution @xmath102 satisfies eq .
( [ dcondition2 ] ) for all @xmath160 with @xmath165 .
the distance between @xmath106 and @xmath102 can be calculated as : @xmath166 \\ & \leq & \epsilon ^{\prime } .
\notag\end{aligned}\]]this proves lemma 2 .
with lemma 2 , now the proof of theorem 1 becomes straightforward .
define a subset of the outputs as @xmath167 and let @xmath168 denote the distribution of @xmath169 conditioned on a particular value of @xmath170 , then we have : @xmath171\nonumber\\ & = & -\mathtt{log}_{2}\sum_{\mathcal{i},\mathcal{e}}\mathcal{d}(\mathcal{i},\mathcal{e}|m)[\max_{\mathcal{o}}\mathcal{d}(\mathcal{o}|\mathcal{i},\mathcal{e},m ) ] \notag \\ & = & -\mathtt{log}_{2}\sum_{\mathcal{i},\mathcal{e}}\mathcal{d}(\mathcal{i},\mathcal{e}|m)\frac{1}{\mathcal{d}(m|\mathcal{i},\mathcal{e})}\max_{\mathcal{o}\in \mathcal{x}_{m}}\mathcal{d}(\mathcal{o}|\mathcal{i},\mathcal{e } ) \notag \\ & \geq & -\mathtt{log}_{2}\sum_{\mathcal{i},\mathcal{e}}\mathcal{d}(\mathcal{i},\mathcal{e}|m)\frac{\mathcal{g}^{k}(\mathcal{l}_{m}-\epsilon ) } { \mathcal{d}(m|\mathcal{i},\mathcal{e } ) } \\ & = & -\mathtt{log}_{2}\sum_{\mathcal{i},\mathcal{e}}\frac{\mathcal{d}(\mathcal{i},\mathcal{e})}{\mathcal{d}(m)}\mathcal{g}^{k}(\mathcal{l}_{m}-\epsilon ) \notag \\ & = & kf(\mathcal{l}_{m}-\epsilon ) -\log _ { 2}\frac{1}{\mathcal{d}(m)}. \notag\end{aligned}\]]here we have used the bayes rule in the first , second and the fourth equalities and eq .
( [ dcondition2 ] ) from lemma 2 in the third inequality ; for the last equality , the equation @xmath172 is used .
the last equality immediately leads to the claim in theorem 1 .
this concludes the proof .
it is worthwhile to clarify that in deriving eq .
( [ theorem1bound ] ) we have made the following four assumptions @xcite : ( i ) the system can be described by quantum theory ; ( ii ) the inputs at the @xmath173th trial @xmath174 are chosen randomly and their values are revealed to the systems only at step @xmath173 ; ( iii ) the three qubits are separated and non - interacting during each measurement step .
( iv ) the possible adversary has only classical side information .
there are no constraints on the states , measurements , or the hilbert space .
moreover , there is even no requirement that the system behaves identically and independently for each trial .
in particular , the system could have an internal memory ( classical or quantum ) so that the results of the @xmath173th trial depend on the previous @xmath175 trials .
we also note that there is a significant difference between the two - qubit scenario in ref . @xcite and our three - qubit scenario here . in the two - qubit case
, the randomness can be certified by the no - signalling conditions as well without the assumption of quantum mechanics . however ,
in our three - qubit scenario , the no - signalling conditions are not sufficient to certify randomness .
actually , we have numerically checked that even for the maximal possible mabk violation @xmath176 , @xmath177 can be equal to the unity for certain @xmath75 and @xmath16 if only the no - signalling conditions are imposed , which can not certify any randomness .
a possible reason for this difference is that the mabk inequality only contains four out of eight possible correlations . in other words ,
the input choices @xmath119 is only a subset of @xmath178 . as a result
, the no - signalling constraints become less effective . [
[ encoding - and - operation - of - qubits - by - topological - manipulation - of - majorana - fermions ] ] encoding and operation of qubits by topological manipulation of majorana fermions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section , we discuss in detail how to control the logical qubits encoded with majorana fermions .
the fusion rule of majorana fermions is of the ising type : @xmath179 , where @xmath180 , @xmath62 , and @xmath63 stand for a majorana fermion , the vacuum state , and a normal fermion , respectively .
generally , there are two encoding schemes .
the first scheme encodes each logical qubit into a pair of majorana fermions ( two - quasiparticle encoding ) .
when the pair fuse to a vacuum state @xmath62 , we say that the qubit is in state @xmath181 ; and when they fuse to @xmath63 , the state is @xmath182 .
there is also an ancillary pair , which soak up the extra @xmath63 if necessary to maintain the constraint that the total topological charge must be @xmath58 for the entire system 2006georgiev,2009ahlbrecht . in this encoding scheme ,
braiding operations of majorana fermions exhaust the entire two - qubit clifford group .
however , for three or more qubits , not all clifford gates could be implemented by braiding .
the embedding of the two - qubit swap gate into a @xmath183-qubit system can not be implemented by braiding @xcite .
in the two - quasiparticle encoding scheme , no violation of the mabk inequality can be obtained as we can not change the measurement basis through local braiding of majorana fermions within each logic qubit .
as we mentioned in the main text , we use the four - quasiparticle encoding scheme where the qubit basis - states are represented by @xmath184 and @xmath185 .
let us first consider braiding operations of majorana fermions within each logic qubit .
consider four majorana operators @xmath186 @xmath187 in one logic qubit , which satisfy @xmath188 , @xmath189 and the anti - commutation relation @xmath190 .
the pauli operators in the computational basis can be expressed as @xcite : @xmath191unitary operations can be implemented by counterclockwise exchange of two majorana fermions @xmath192 : @xmath193specifically , we can write down the three basic braiding operators in the computational basis : @xmath194where @xmath195 means that we ignore an unimportant overall phase .
using these basic braiding operators , a single - qubit hadamard gate can be implemented as @xmath196 .
the corresponding braidings are shown in fig.2 of the main text .
note that the set of operations implemented through composition of @xmath197 and @xmath80 are still very limited , however , it is fortunate that @xmath80 and @xmath198 give all the gates that we need for change of the measurement bases in test of the mabk inequality .
as shown in the main text , we actually get maximum quantum violation of the mabk inequality by randomly choosing either a @xmath80 or an @xmath198 gate on each logic qubit before measurement of the anyon fusion . with only braiding operations of majorana fermions
, no entangling gate can be achieved for logic qubits in the four - quasiparticle encoding scheme due to the _ no - entanglement rule _ proved in ref .
@xcite . in order to overcome this problem
, we need assistance from another kind of topological manipulation : nondestructive measurement of the anyon fusion , which can be implemented through the anyon interferometry as proposed in ref .
suppose that we have eight majorana modes @xmath199 , where the first ( last ) four modes encode the control ( target ) qubit , respectively . as shown in ref .
2002bravyi,2005bravyi , a two - qubit controlled phase flip gate @xmath200 can be implemented through the following identity : @xmath201 note that the first two operations in eq .
( [ conz ] ) can be directly implemented by braiding operations .
the key step is to implement the operation @xmath202 . to this end
, we use another ancillary pair of majorana fermions @xmath203 and @xmath204 .
we measure fusion of the four majorana modes @xmath205 .
the outcome is @xmath206 , corresponding to either a vacuum state ( @xmath207 ) or a normal fermion ( @xmath208 ) .
the corresponding projector is given by @xmath209 .
then , we similarly measure fusion of the majorana modes ( operator ) @xmath210 , with the project denoted by @xmath211 corresponding to the measurement outcomes @xmath206 .
we have the following relation @xcite : where @xmath213 , @xmath214 , @xmath215 , and @xmath216 .
all the gates @xmath217 can be implemented through one or several braiding operations of majorana fermions .
so this identity shows that an effective controlled phase flip gate can be implemented on logic qubits through a combination of anyon braiding and measurement of anyon fusion .
depending on the measurement outcomes @xmath218 of @xmath205 and @xmath210 , one can always apply a suitable correction operator @xmath217 to obtain the desired operation @xmath202 . with controlled phase flip gates , one can easily realize quantum controlled - not ( cnot ) gate with assistance from the hadamard operations that can be implemented through the anyon braiding .
with cnot and hadamard gates , we can then prepare the maximally entangled three - qubit state as required for test of quantum violation of the mabk inequality .
f. hassler , a. r. akhmerov , c. y. hou , and c. w. j. beenakker , new j. of phys .
* 12 * 125002 ( 2010 ) .
the theoretical results in ref .
@xcite were improperly formulated , and the inaccuracies in formulation were corrected in the recent refs .
@xcite and @xcite . | braiding of majorana fermions gives accurate topological quantum operations that are intrinsically robust to noise and imperfection , providing a natural method to realize fault - tolerant quantum information processing .
unfortunately , it is known that braiding of majorana fermions is not sufficient for implementation of universal quantum computation . here
we show that topological manipulation of majorana fermions provides the full set of operations required to generate random numbers by way of quantum mechanics and to certify its genuine randomness through violation of a multipartite bell inequality .
the result opens a new perspective to apply majorana fermions for robust generation of certified random numbers , which has important applications in cryptography and other related areas .
the complex - valued solutions to the dirac equation predict that every elementary particle should have a complex conjugate counterpart , namely an antiparticle .
for example , an electron has a positron as its antiparticle .
however , in @xmath0 ettore majorana @xcite showed that the complex dirac equation can be modified to permit real wave - functions , leading to the possible existence of the so called _ majorana fermions _ which are their own antiparticles @xcite . in condensed matter physics
, majorana fermions may appear as elementary qusi - particle excitations . to search for majorana fermions ,
several proposals have been made in recent years , including @xmath1 fractional quantum hall system @xcite , topological insulator ( ti)superconductor ( sc ) interface @xcite , interacting quantum spins @xcite , chiral p - wave superconductors @xcite , spin - orbit coupled semiconductor thin film @xcite or quantum nanowire @xcite in the proximity of an external s - wave superconductor .
based on these proposals , experimentalists have made great progress recently .
for instance , ref .
@xcite reported an experimental observation of coexistence of the superconducting gap and the topological surface state in the @xmath2 thin film as a step towards realization of majorana fermions .
more recently , signature of majorana fermions in hybrid superconductor - semiconductor nanowire device has been reported @xcite , which has raised strong interest in the community .
majorana fermions are exotic particles classified as non - abelian anyons with fractional statistics , and braiding between them gives nontrivial quantum operations that are topological in nature .
these topological quantum operations are intrinsically robust to noise and experimental imperfection , so they provide a natural solution to realization of fault - tolerant quantum gates .
application of majorana fermions in implementation of fault - tolerant quantum computation has raised great interest @xcite .
unfortunately , braiding of majorana fermions are not sufficient yet for realization of universal quantum computation @xcite , and we need assistance from additional non - topological quantum gates which are prone to influence of noise . in this letter , we show that topological manipulation of majorana fermions alone can be used to realize a quantum random number generator in a fault tolerant fashion and to certifies its genuine randomness through violation of the mermin - ardehali - belinskii - klyshko ( mabk ) inequality 1993belinskii,1992ardehali,1990mermin .
random numbers have tremendous applications in science and engineering 2001ackermann,1991hultquist,2003tu . however , generation of genuine random numbers is a challenging task @xcite .
any classical device does not generate genuine randomness as it allows a deterministic description in principle .
quantum mechanics is intrinsically random , and one can explore this feature to generate random numbers 1956isida,1990svozil,1994rarity,2000jennewein . however , in real experiments , the intrinsic randomness of quantum mechanics is always mixed - up with an apparent randomness due to noise or imperfect control of the experiment @xcite .
the latter can be exploited by an adversary opponent and leads to security loopholes in various applications of randomness .
recently , a nice idea has been put forward to certify genuine randomness generated by a quantum device through test of violation of the bell - chsh ( clauser - horn- shimony - holt @xcite ) inequality 2007colbeck,2010pironio , and the idea has been demonstrated in a proof - of - principle experiment using remote entangled ions @xcite .
this implementation is not fault - tolerant yet as the remote entanglement is sensitive to noise and the quantum gates have limited precision which can all lead to security loopholes .
we show here that all the operations for generation and certification of genuine randomness can be realized through topological manipulation of majorana fermions .
this implementation is inherently fault - tolerant and automatically closes security loopholes caused by influence of noise .
the implementation of certification of a quantum random number generator with majorana fermions is tricky .
first of all , one can not use the bell - chsh inequality anymore as proposed in ref .
@xcite , since it is impossible to violate this inequality through topological manipulation of majorana fermions alone @xcite .
in fact , to observe violations of the chsh inequality , measurements in the non - clifford bases are required .
however , topological operations on majorana fermions can only give gates in the clifford group , and thus not able to achieve the measurements required for the chsh inequality violation for randomness certification .
consequently , we have to consider certification of randomness based on extension of the bell inequalities in the multi - qubit case . for simplicity , here we use the mabk inequality for three logical qubits @xcite .
we show that first , this inequality can be used to certify randomness , and second , the inequality can be tested with topological manipulation of majorana fermions alone . for the mabk inequality
, we consider three qubits , each with two measurement settings .
we denote the measurement settings for each qubit by the binary variables @xmath3 , @xmath4 , @xmath5 , and the corresponding measurement outcomes by @xmath6 , @xmath7 , @xmath8 , where @xmath9 .
the mabk inequality can be rewritten as @xcite @xmath10\leq 2 , \label{mabk - ineq}\]]where @xmath11 and @xmath12 is a sign function defined by @xmath13 ; @xmath14 ( @xmath14 ) is the probability that @xmath15 is an even ( odd ) number when settings @xmath16 are chosen .
the inequality ( mabk - ineq ) is satisfied by all local hidden variable models .
however , in quantum mechanics certain measurements performed on entangled states can violate this inequality .
experimentally , we can repeat the experiment @xmath17 times in succession to estimate the violation .
for each trial , the measurement choices @xmath16 are generated by an independent identical probability distribution @xmath18 .
denote the input string as @xmath19 and the corresponding output string as @xmath20 .
the estimated violation of the mabk inequality can be obtained from the observed data as @xmath21,\]]where @xmath22 ( @xmath23 ) denotes the number of trials that we get an even ( odd ) outcome @xmath15 after @xmath17 times of measurements with the measurement setting @xmath16 .
we need to show that the output string @xmath24 from the measurement outcomes contains genuine randomness by proving that it has a nonzero entropy .
let @xmath25 be a series of violation thresholds with @xmath26 and @xmath27 , corresponding respectively to the classical and quantum bound . denote by @xmath28 the probability that the observed violation @xmath29 lies in the interval @xmath30 .
we can use the min - entropy to quantify randomness of the output string @xmath24 2010pironio,2012pironio,2009koenig : @xmath31,\]]where @xmath32 represents the knowledge that a possible adversary has on the state of the device and the maximum is taken over all possible values of the output string @xmath24 .
the probability distribution @xmath33 is defined in the supplemental material . based on a similar procedure as in ref .
@xcite , we can prove that if @xmath34 , the min - entropy of the output string conditional on the input string and the adversary s information has a lower bound ( see derivation in the supplement ) , given by @xmath35where the parameter @xmath36 with @xmath37 , the smallest probability of the input pairs , and @xmath38 is a given parameter that characterizes the closeness between the target distribution @xmath39 and the real distribution after @xmath17 successive measurements ( see the supplement for an explicit definition ) .
the function @xmath40 can be obtained through numerical calculation based on semi - definite programming ( sdp ) @xcite and is shown in fig .
randomness - show .
the minimum - entropy bound @xmath41 and the net entropy versus the number of trials @xmath17 are plotted in the insets ( a ) and ( b ) of fig . randomness - show .
any observed quantum violation with @xmath42 leads to a positive lower bound of the min - entropy , and a positive mini - entropy guarantees that genuine random numbers can be extracted from the string @xmath24 of the measurement outcomes through the standard protocol of random number extractors @xcite . as some amount of randomness needs to be consumed to prepare the input string according to the probability distribution @xmath18 , the scheme here
actually realizes a randomness expansion device @xcite .
similar to ref .
@xcite , we can show that under a biased distribution @xmath18 as shown in fig .
[ randomness - show ] we generate a much longer random output string of length @xmath43 from a relatively small amount of random seeds of length @xmath44 when @xmath17 is large . versus violation @xmath29 of the mabk inequality .
the function is calculated through optimization based on the semi - definite programming with the details shown in the supplemental material .
the inset ( a ) shows the lower bound of the min - entropy @xmath45 versus the number of trials @xmath17 .
here we assume an observed mabk violation lies within the interval @xmath46 with probability @xmath47 .
the parameters are chosen as @xmath48 and @xmath49 .
the bound @xmath50 depends on the input probability distribution @xmath18 through the parameter @xmath51 .
the blue - square line represents the bound under a uniform distribution ( @xmath52 for all @xmath53 ) , while the red - dotted line shows the bound under a biased probability distribution with @xmath54 and @xmath55 with @xmath56 .
it consumes less randomness to generate a biased distribution for the input bits , so the net amount of randomness , defined as the number of output random bits minus that of the input , becomes positive when @xmath17 is large ( typically @xmath17 needs to be of the order @xmath57 ) .
the inset ( b ) plots the net amount of randomness generated after @xmath17 trails under a biased distribution of the inputs .
the parameters are the same as those in the inset ( a ) .
, width=332 ] we now show how to generate and certify random numbers using majorana fermions .
the key step is to generate a three - qubit entangled state and find suitable measurements that lead to violation of the mabk inequality .
majorana fermions are non - abelian anyons , and their braiding gives nontrivial quantum operations . however ,
this set of operations are very restricted .
first , all the gates generated by topological manipulation of majorana fermions belong to the clifford group , and it is impossible to use such operations alone to violate the chsh inequality @xcite .
we have to consider instead the multi - qubit mabk inequality .
second , it is not obvious that one can violate the mabk inequality as well using only topological operations .
there are two ways to encode a qubit using majorana fermions , using either two quasiparticles ( majorana fermions ) or four quasiparticles ( see the details in the supplement ) . in the two - quasiparticle encoding scheme , although the braiding gates exhaust the entire two - qubit clifford group , they can not span the whole clifford group for more than two qubits @xcite .
furthermore , braiding majorana fermions within each qubit can not change the topological charge of this qubit which fixes the measurement basis . thus
, no violation of the mabk inequality can be achieved using the topological operations alone in the two - quasiparticle encoding scheme . in the four - qusiparticle encoding scheme , it is not straightforward either as braidings in this scheme only allows certain single - qubit rotations and no entanglement can be obtained due to the no - entanglement rule proved already for this encoding scheme 2006bravyi .
fortunately , we can overcome this difficulty by taking advantage of the non - destructive measurement of the anyon fusion , which can induce qubit entanglement @xcite .
in a real physical device , the anyon fusion can be read out non - destructively through the anyon interferometry 2010hassler . in the four - qusiparticle encoding scheme : each qubit is encoded by four majorana fermions , with the total topological charge @xmath58 .
the qubit basis - states are represented by @xmath59 and @xmath60 . here
, each @xmath61 represents a majorana fermion ; @xmath62 and @xmath63 represent the two possible fusion channels of a pair of majorana fermions , with @xmath62 standing for the vacuum state and @xmath63 denoting a normal fermion . as explained in the supplemental material , a topologically protected two - qubit cnot gate can be implemented using braidings together with non - destructive measurements of the anyon fusion @xcite . to certify randomness through the mabk inequality , we need to prepare a three - qubit entangled state . for this purpose ,
we need in total fourteen majorana fermions , where twelve of them are used to encode three qubits and another ancillary pair is required for implementation of the effective cnot gates through measurement of the anyon fusion .
initially , the logical state is @xmath64 .
we apply first a hadamard gate on the qubit @xmath65 , which can be implemented through a series of anyon braiding as shown in fig .
braidingsb , and then two effective cnot gates on the logical qubits @xmath65 , @xmath66 , and @xmath67 @xmath68 .
the final state is the standard three - qubit maximally entangled state @xmath69 .
after @xmath70 is generated , the three qubits can be separated and we need only local braiding and fusion of anyons within each qubit to perform the measurements in the appropriate bases to generate random numbers and certify them through test of the mabk inequality . to perform the measurements , we read out each qubit according to the input string @xmath71 through nondestructive detection of the anyon fusion . if the input is @xmath58 , we first braid the majorana fermions to implement a hadamard gate @xmath72 on this qubit ( as shown in fig .
[ braidings]b ) , and then measure the fusion of the first two majorana fermions within each qubit .
the measurement outcome is @xmath58 ( @xmath65 ) if the fusion result is @xmath62 ( @xmath63 ) . if the input is @xmath65 , we first braid the majorana fermions to implement a @xmath73 gate ( see fig . [ braidings]a ) on this qubit before the same readout measurement .
for instance , with the the input @xmath74 , we apply a hadamard gate to the first qubit and @xmath73 gates to the second and the third qubits , followed by the nondestructive measurement of fusion of the first two majorana fermions in each qubit . under the state @xmath70 , the conditional probability of the measurement outcomes @xmath75 under the measurement setting @xmath16 for these three qubits is give by @xmath76where @xmath77 and @xmath78 . with this conditional probability , we find the expected value of @xmath29 defined in eqs .
( 1,2 ) is @xmath79 , achieving the maximum quantum violation of the mabkinequality .
all the steps for measurements and state preparation are based on the topologically protected operations such as anyon braiding or nondestructive detection of the anyon fusion , so the scheme here is intrinsically fault - tolerant and we should get the ideal value of @xmath79 if the majorana fermions can be manipulated at will in experiments .
such a large violation perfectly certifies genuine randomness of the measurement outcomes . and
@xmath68 implements a unitary gate @xmath80 on the corresponding qubit .
( b ) implementation of the hadamard gate through composition of anyon braiding . in both ( a ) and ( b ) , time flows from left to right and @xmath81 means equal up to an irrelevant overall phase .
, width=321 ] in summary , we have shown that genuine number numbers can be generated and certified through topologically manipulation of majorana fermions , a kind of anyonic excitations in engineered materials .
such a protocol is intrinsically fault - tolerant . given the rapid experimental progress on realization of majorana fermions in real materials @xcite , this protocol offers a promising prospective for application of these topological particles in an important direction of cryptography with broad implications in science and engineering .
we thank y. h. chan , j. x. gong , and e. lichko for discussions .
this work was supported by the nbrpc ( 973 program ) 2011cba00300 ( 2011cba00302 ) , the iarpa musiqc program , the aro and the afosr muri program . |
taurus - auriga is the star forming region which has been most thoroughly surveyed for pre - main - sequence binary and multiple systems ( see mathieu 1994 for a review ) . for separations from 15 a.u . to 2000 a.u .
, the binary frequency among ttauri stars in taurus is 1.9 times as high as among nearby main - sequence stars ( khler & leinert 1998 ) . extrapolating over the whole range of separations yields a binary frequency of 100% ,
i.e. , each ttauri star in taurus should be member of a binary or multiple systems .
this apparent overabundance of binaries among pre - main - sequence stars is puzzling .
one possible explanation is a decrease in the binary frequency as a function of the age of a stellar population ( patience et al .
however , a tassociation like taurus might not be the typical birthplace for low - mass stars , as up to 80% of all low - mass stars could originate in ob associations ( miller & scalo 1978 ; see also zinnecker et al .
1992 ) .
scorpius - centaurus is the most nearby ob association at a distance of about 145 parsec ( de zeeuw et al .
it consists of three subgroups ( cf .
figure 1 ) with ages ranging from 5 to 13 myr ( de geus et al .
upper centaurus - lupus ( ucl ) is the oldest subgroup of the association .
star formation started here 13 myr ago and subsequently progressed throughout the parental giant molecular cloud ( e.g. blaauw 1991 and references therein ) . based on observations with the einstein x - ray satellite , walter et al .
( 1994 ) identified 28 weak - line ttauri stars ( wtts ) in upper scorpius ( us ) , the youngest subgroup .
10 of these have been surveyed by ghez et al .
( 1993 ) for binary or multiple systems , and three binaries have been detected .
the einstein fields covered only a small fraction of us ( fig . 2 ) and the 28 wtts did not allow for a statistical meaningful study of binary frequencies and separations . a search for visual binary stars among 74 rosat selected wtts and post - ttauri stars in us ( kunkel et al .
, in prep ) was carried by brandner et al .
this survey was sensitive to binary separations down to 08 , and revealed a rather high binary frequency in the region located between us and ucl ( ` us - b ' ) and an apparent absence of wide visual binary stars in the center of us ( ` us - a ' ) . in order to identify closer binary systems and to get a better statistics on possible spatial variations of binary star properties , we have carried out a speckle survey of 114 wtts in us based on the lists by walter et al .
( 1994 ) and kunkel et al .
( in prep ) .
the speckle data were obtained with the sharp - camera at the eso 3.5 m new technology telescope on la silla , chile , in may 1994 and july 1995 .
all observations were performed in the k - band . in order to estimate the local density of background sources
, we obtained additional infrared images with the eso / mpia @xmath0 telescope on la silla in march 1996 using the irac2b camera .
the detailed results , i.e. , separations , position angles , photometry of the individual binary components , and de - biased binary frequencies will be presented in a forthcoming paper ( khler et al .
, in prep ) .
here we concentrate on the distribution of the binary separations . in total
we observed 68 wtts in us - a ( 26 from the list by walter et al .
1994 , and 42 new wtts from kunkel et al .
, in prep ) and 46 wtts in us - b ( kunkel et al . , in prep ) . as us is projected onto the galactic bulge and in the k - band
has a 6@xmath1 higher density of background sources than taurus ( khler 1997 ; khler et al .
, in prep . )
, we used a separation of 3@xmath2 as the upper cut - off , in order to reduce contamination by background giants .
the distribution of brightness ratios as a function of binary separations is rather uniform for separations between 01 and 3@xmath2 .
the limiting magnitude for undetected companions has a 1/d dependence with a typical brightness ratio limit of 0.16 at 05 and 0.1 at 08 . among the binaries with separations between 08 and 3@xmath2 , only two have brightness ratios @xmath3 0.1 . for separations
@xmath4 4@xmath2 we find a larger number of visual binaries with brightness ratios @xmath3 0.1 , most of which are very likely chance projections .
counting binaries with separations between 01 and 3@xmath2 , we detect 21 binaries in us - a and 18 binaries ( if we count one hierarchical triple system as two binaries ) in us - b .
the age of the wtts population in us is under debate .
walter et al . ( 1994 ) derive ages of 1 - 2 myr for the wtts and conclude that the wtts formed 3 - 4 myr _
after _ the early - type stars in us .
kunkel et al .
( in prep ) and martn ( 1998 ) find that the age of wtts in upper scorpius is in agreement with the age of the b stars .
our speckle survey uncovers that almost all of the brighter m - type wtts in us - a and us - b are in fact binaries .
only if they are not recognized as such , they are erroneously placed too high in the h - r diagram ( cf .
simon et al .
1993 , brandner & zinnecker 1997 , ghez et al .
1997 ) and an age of @xmath3 1 myr is derived .
the wtts in us - a and us - b show the same mix in spectral types and ages .
the average lithium abundance is slightly lower in us - b compared to us - a .
if this is not due to age , then it might be related to the difference in binary separations .
wider binaries sustain massive circumstellar disks longer than closer binaries . without a massive circumstellar disk
a ttauri star can not effectively dissipate angular momentum and thus spins up while contracting ( e.g. bodenheimer 1995 ) .
faster rotation should enhance the dynamo effect and the resulting stronger internal magnetic fields could in turn aggravate convection and thus slow down the depletion of lithium on the surface .
based on counts of background sources , we expect two of the 39 binaries in our sample to be pure chance projections .
a 5% contamination by bogus binaries should not strongly affect our analysis .
without correction for selection effects , the ` raw ' binary frequency in us - b ( 39%@xmath59% ) is somewhat higher than in us - a ( 31%@xmath57% ) , but the difference is not statistically significant .
the distribution of binary separations , however , is clearly distinct in the two regions : about 80% of the binary systems in us - a have separations less than 08 , whereas in us - b 70% of the binary systems have separations greater than 08 ( fig . 2 ) .
the average separation of binary systems in us - b is 147 , more than twice as large as the average separation of binary systems in us - a ( 063 ) . in order to transform angular to physical separations ,
we need to know the distance to our stars .
the average distance to scorpius - centaurus , as derived from hipparcos measurements of early type stars , is 145 parsec for us and 140 parsec for ucl ( de zeeuw et al .
1998 ) . for four of the wtts in us - b hipparcos derived meaningful distances ( neuhuser & brandner 1998 ; wichmann et al . 1998 ) .
one of these stars ( hd 140637 ) is clearly a foreground star at a distance of 41@xmath6 parsec .
the three other stars ( all of them are binaries ) have hipparcos distances between 90 parsec and 160 parsec , albeit with rather large uncertainties ( @xmath520 to @xmath580 parsec ) .
we see no evidence that the wtts in us - b are more than a factor of 2 closer than the wtts in us - a as would be necessary in order to explain the difference in the distribution of angular binary separations by the distance . in the following we assume that the wtts in us - a and us - b are at the same distance as the early type stars in scorpius - centaurus .
figure 3 shows a histogram of the distribution of binary separations in us - a and in us - b . while the distribution peaks around 90 a.u . in us -
a , the peak of the distribution is clearly shifted towards larger separations ( @xmath7 215 a.u . ) in us - b .
a kolmogorov - smirnov test indicates that both binary populations are drawn from different distributions at a significance level of 98% .
removing the two widest binaries in us - b from the sample ( assuming that they might be chance projections ) lowers the significance level to % 96 .
in scorpius - centaurus the molecular cloud in which the star formation originated has long been dispersed by ionizing radiation and fast stellar winds from the massive stars and by supernova explosions .
today , the lupus dark clouds to the south and the @xmath8 ophiuchi dark clouds to the east of us are the only sites of ongoing ( low - mass ) star formation in the outskirts of a once giant molecular cloud .
no direct trace of the original physical conditions remains .
we can , however , deduce that the same environment which facilitated the formation of massive stars in us - a also facilitated the formation of closer binaries among the low - mass stars , whereas in us - b the original physical conditions were such that almost no massive stars formed , and preferentially wide binaries among the low - mass stars .
other studies might reveal a similar trend . in their search for visual binaries in nearby dark clouds
reipurth & zinnecker ( 1993 ) found a weak anti - correlation between the binary frequency and the number of young stars in the clouds .
this could be due to a shift in the peak of the distribution of binary separations , with the larger clouds producing preferentially closer binaries than the smaller clouds .
padgett et al .
( 1997 ) and petr et al .
( 1998 ) studied the binary frequency in orion ( distance @xmath7 460 pc ) and found a deficit of `` wide '' ( @xmath401 , i.e. , 50 a.u . ) pre - main - sequence binaries among ttauri stars closely associated to the high - mass stars in the trapezium cluster .
binary stars form due to fragmentation during collapse ( e.g. boss 1997 and references therein ) .
the binary separation depends on the initial angular momentum and on the critical density at which the magnetic field decouples from the matter ( mouschovias 1977 ) .
the critical density @xmath9 relates to the separation @xmath10 of a binary systems as @xmath11 .
part of the initial angular momentum of a protobinary might be transported outwards by means of a circumbinary disk or might be dissipated due to the gravitational interaction of the circumstellar disks of the individual components .
the temperature structure of a molecular cloud could influence the number of binary stars ( durisen & sterzik 1994 ) as well as their typical separation ( durisen et al .
, in prep . ) .
the limited parameter space due to a higher ambient cloud temperature would preferably lead to the formation of closer binaries .
molecular clouds in the process of forming ( binary ) stars have to be studied in order to learn which physical quantities determine the outcome of the binary formation process .
we have shown that the distributions of binary separations among weak - line ttauri stars in two adjacent fields ( ` us - a ' and ` us - b ' ) in the scorpius - centaurus ob association are clearly distinct from each other and considerably more peaked than the ( broad ) distribution of binary separations observed among main - sequence field stars . in us - a , the wtts are closely associated with b type stars , whereas in us - b only a few early type stars are present .
we conclude that the same physical conditions which facilitate the formation of massive stars also facilitate the formation of closer binaries among low - mass stars , whereas physical conditions unfavorable for the formation of massive stars lead to the formation of wider binaries among low - mass stars .
the outcome of the binary formation process might be determined by the critical density at which the magnetic field support breaks down , the internal turbulence and the angular momentum of molecular cloud cores , the initial temperature within a cloud , or - most likely - a combination of all of these .
we further conclude that the distribution of binary separations is not a universal quantity .
instead , both the peak and the width of the distribution might vary from one star forming region to the next . the broad distribution of binary separations observed among main - sequence field stars can be understood as a superposition of binary populations originating in various star forming environments with very distinct peaks in the distribution of binary separations .
the apparent overabundance of binaries among ttauri stars in the taurus - auriga tassociation might be explained by the fact that the distribution of binary separations there is strongly peaked towards @xmath7 30a.u .
extrapolating from this very pronounced peak over the whole range of possible binary separations then leads to an erroneously high estimate of the overall binary frequency .
zinnecker , h. , brandner , w. , reipurth , b. 1992 pre - main - sequence binaries , in `` complementary approaches to double and multiple star research '' , eds .
mcalister & w.i .
hartkopf , asp conf . | we have carried out k - band speckle observations of a sample of 114 x - ray selected weak - line ttauri stars in the nearby scorpius - centaurus ob association .
we find that for binary ttauri stars closely associated to the early type stars in upper scorpius , the youngest subgroup of the ob association , the peak in the distribution of binary separations is at 90 a.u . for binary ttauri stars located in the direction of an older subgroup , but
not closely associated to early type stars , the peak in the distribution is at 215 a.u .
a kolmogorov - smirnov test indicates that the two binary populations do not result from the same distribution at a significance level of 98% .
apparently , the same physical conditions which facilitate the formation of massive stars also facilitate the formation of closer binaries among low - mass stars , whereas physical conditions unfavorable for the formation of massive stars lead to the formation of wider binaries among low - mass stars .
the outcome of the binary formation process might be related to the internal turbulence and the angular momentum of molecular cloud cores , magnetic field , the initial temperature within a cloud , or - most likely - a combination of all of these .
we conclude that the distribution of binary separations is not a universal quantity , and that the broad distribution of binary separations observed among main - sequence stars can be explained by a superposition of more peaked binary distributions resulting from various star forming environments .
the overall binary frequency among pre - main - sequence stars in individual star forming regions is not necessarily higher than among main - sequence stars . |
the study of complex systems , a name that is frequently used for systems having a large number of elements that interact in ( usually ) non trivial ways , has been greatly advanced in recent years by the use of graph theory @xcite .
one can map a complex system to a complex network by representing the interacting elements of the system with nodes and their interactions with links between the nodes .
examples of complex systems that have been recently investigated in this perspective include the internet @xcite , the world wide web @xcite , communication networks @xcite , food webs @xcite , sexual contacts among individuals @xcite , economic networks @xcite , the network of collaborations in eu funded projects @xcite , etc .
complex systems are not always static , meaning that they may evolve dynamically over time .
this evolution can provide a wealth of information about the processes driving the system .
one way to study such systems is to record the time dependence of some specific and well defined property , and thus obtain a set of time series that are able to depict the time evolution of the entire system .
these time series can be transformed into a graph by using a similarity measure based in the cross - correlation among its elements , as very effectively has been done , for example , for the study of equity portfolios @xcite , which has led to an established method for the investigation of financial systems .
therefore , we can use the cross - correlations between variables to create a correlation based network from any complex system that we have data available . on the other hand ,
if the system is static , meaning that its interactions do not change with time , but there are different kinds of interactions among its elements , we can create a static graph for each different interaction , and study this set of graphs . in this work
we describe an approach that can be used for both types of the aforementioned analyses .
it is based on the minimum spanning tree ( mst ) technique @xcite , but it allows the creation of a more representative network of the system ( compared to a simple mst ) that maintains information about its dynamics and its temporal evolution . as application , we use this method to analyze the european scientific collaboration network for the projects carried out with the support of the framework program fp6 , and the network of the 100 highest capitalized stocks traded in the nyse in the period 1995 - 2003 .
any complex system with @xmath0 interacting elements can be mapped to a network with @xmath0 nodes .
the links connecting the nodes of the network represent the interactions among the system elements , and the strength of these interactions is used as weights of the links between the nodes .
the total number of links of a network depends on the information that we have about interactions between its nodes . for the special case that we know the interaction strength between all pairs of nodes , as we do in correlation based graphs , then the network is fully connected and has @xmath1 number of links . for such cases
it is essential to use filtering techniques in order to reduce the number of connections , so that we can study properties of the network that generally are hidden due to its complexity .
the most drastic filtering of a network can be achieved by the extraction of the mst @xcite , a technique that has been used extensively in the literature for the study of financial correlation matrices @xcite , leading to the identification of clusters of stocks that result in a meaningful taxonomy .
the mst was recently used to extract the mode of collaboration in research projects funded by the european commission framework programs @xcite .
the mst is a graph with the same number of nodes @xmath0 as the original network but having a total of only @xmath2 edges , with a total minimum weight .
it is constructed starting with to the disconnected graph that contains all the nodes of the network , and then by adding links in increasing weight order , as long as they do not form loops , until all nodes are connected .
such structure is a much simpler graph than the fully connected network , but it still gives us interesting information about the system .
depending on the structure of the system under investigation , and on the available data , we could create a series of representative networks that contain information about the evolution and the dynamics of the original system .
for the case of networks that are stationary , we can construct different snapshots of them by using as links different interactions among the .
this allows us to calculate msts separately for all different snapshots of the system . on the other hand , for the case of correlation based graphs , where the network is constructed using time series data
, we divide the time series into smaller segments ( time windows ) and construct different network snapshots of the system at different time periods .
let us name @xmath3 the set of all the links present in each calculated mst , and let us assume that we have calculated a total of @xmath4 msts from the @xmath4 different snapshots of the original network that we can obtain .
we can create the graph @xmath5 that has as links the union of all sets @xmath3 for all the @xmath4 trees that we have calculated @xmath6 the number of links of graph @xmath5 lies inside the interval @xmath7 $ ] , where @xmath2 is the number of links of one mst with @xmath0 nodes and @xmath8 is the total number of links of @xmath4 different msts . obviously , the graph @xmath5 will have @xmath2 links if the investigated system is so stable that for all the @xmath4 investigated snapshots we find the same mst , while it will have @xmath8 links if the system is so unstable that for all the @xmath4 investigated snapshots we find completely different msts .
for the case of a highly unstable system , if we choose @xmath9 , we will get a fully connected graph with @xmath1 links . in the msts that we calculate
, each link has a weight , which is representative of the strength of the interaction between nodes @xmath10 and @xmath11 .
we use these weights of the individual msts to calculate the new weights of the combined graph @xmath5 .
we set the weight of a link between nodes @xmath10 and @xmath11 in graph @xmath5 as the mean value of the strength that the same link has for the entire set of the @xmath4 msts . because of this construction method , we name the resulting graph `` overlapping tree network '' ( otn ) .
a representative example of the above procedure is given in figure [ fig:1 ] .
in the following section , we will implement this technique to the different msts that we obtain from the european collaboration network of the fp6 , and to the network of the 100 highest capitalized stocks traded in the nyse from 1995 to 2003 .
joint scientific research in europe has been funded through large programs called framework programs ( fp ) . in
all fps partners come from different countries , and thus international collaborations are strongly encouraged . in the following
we construct the collaboration networks of all countries participating in the fp6 , which is the last concluded fp that ran during the period 2002 - 2006 .
such dataset can be obtained from cordis @xcite .
the collaboration networks are constructed separately for every thematic area , out of a total of 16 , by considering each country as a node , and by representing the collaborations among countries as links between the nodes .
this means that an edge connecting two nodes , @xmath10 and @xmath11 , represents the presence of at least one collaboration between institutions from country @xmath10 with institutions from country @xmath11 . at each edge
we assign a weight @xmath12 , that represents the total number of collaborations between institutions of the two countries , on a specific thematic area .
we transform this weight to a distance measure @xmath13 , in such a way that the smaller the distance , the stronger the collaboration between countries . by default @xmath14
is defined in the interval @xmath15 $ ] , and takes its maximum value when there is only one collaboration between a pair of countries , @xmath16 .
the use of spanning trees as subnetworks that retain only the most meaningful connections of the original network , is an approach that has enhanced our understanding in various complex systems . following this approach in a previous work @xcite we used the mst to measure the role of a country in the collaboration network that resulted due to its participation in fp6 .
we found that each mst of the fp6 collaboration network has star like structure around some specific countries , for different thematic areas .
these countries , that are found to act as hubs ( strongly connected nodes ) , are germany , united kingdom , france , and italy .
more specifically , we find that germany ( de ) is the central hub for 62.5% of the thematic areas , united kingdom ( uk ) 25% , france ( fr ) 6.25% , and italy ( it ) 6.25% . here , for each one of the 16 thematic collaboration networks we calculated again these minimum spanning trees , and we combined them applying the aforementioned technique to obtain the otn of the collaboration network .
examples of mst , where the star like structure becomes apparent and the resulting otn are shown in figure [ fig:2 ] .
the otn is still a well connected network , but its connections represent only the stronger collaboration links , as they are captured by the mts for every thematic area . in order to get a more detailed view
, we zoom more to the otn , by examining only the collaborations among the 25 eu member countries . a strongly interconnected community of the most frequent hubs of the network is identified .
this community is like a nucleus of countries with very strong collaboration links , while the other countries play only a satellite role around it .
this picture is much richer in comparison to the mst because , not only it highlights the importance of the hubs of the network ( this information we can get from the mst as well @xcite ) , but it also shows quantitatively the interconnecivity pattern between them , in such a way that a fully connected community is formed between de , uk , fr , it , and es .
this means that these five countries not only play important role in the network , but they are connected among themselves with very strong connections , as it is shown by the thickness of the links between them .
this valuable information is not possible to be extracted from the mst since , by definition , the mst does not allow loops to be formed .
we now apply the otn technique to the network of the 100 most capitalized stocks traded in the nyse , using daily returns in the period 1995 - 2003 .
the basis of the creation of an equity network , using a portfolio of @xmath0 stocks , is the analysis of the cross - correlation among time series of returns for all pairs of stocks .
the correlation coefficient provides a similarity measure that can be used as weight for a link between each pair of stocks .
thus , a correlation based network is a fully connected weighted graph , and the weights of the graph are obtained from the correlation matrix of the system .
the extraction of the mst from such graphs gives a wealth of useful information @xcite , but because the mst is a very drastic filtering method , it can not capture more structured entities , such as the communities of stocks connected with strong links @xcite .
the need to find filtering techniques that will create richer graphs than the mst was first addressed by tumminello et al . with the creation of the planar
maximally filtered graph ( pmfg ) technique @xcite .
the pmfg is a graph that is created by adding links to a disconnected network in decreasing weight order , with the constrain that the resulting graph will be planar . in what follows
, we will use the otn technique and we will compare our results with the results of the pmfg . in order to apply the otn technique for the case of financial correlation based networks that originate from time series data , as explained in the `` methods '' section , we divide the time series to a set of smaller segments using a sliding time window of length @xmath17 .
the length @xmath17 of the time window is a fraction @xmath18 of the original size of the return time series , for which @xmath19 days .
the time step that we use to move the time window is one day . for every time step
we calculate the correlation matrix , which contains @xmath20 entries , determined from @xmath0 time series of length @xmath17 .
if @xmath17 is not very large compared to @xmath0 , it is shown using arguments from random matrix theory @xcite that the determination of the correlations is noisy . as a results , in order to have a more reliable determination of the correlation matrix , we make sure that @xmath21 .
using this procedure we calculate the otns for different time window lengths . as an example we show in fig .
[ fig:3 ] the otn that was calculated using time window of length @xmath22 days . from this figure
we see that most of the stocks belonging to the same sector of economic activity are , as expected , clustered together . but
from the otn of fig .
[ fig:3 ] we identify how communities of stocks belonging to the same sector of economic activity are connected , and how strong the links between the communities are , just by examining the thickness of these links .
furthermore , it is now possible to identify certain stocks that have large number of connections with stocks from different sectors .
such stocks are the stock of general electric ( ge ) , which is the most connected stock of the network , the stock of american international group ( aig ) that is an insurance and financial conglomerate company , etc .
this shows that these stocks have activities not only in their main field of operations but they also extend to totaly different fields , as for example ge , which is known for electrical machinery ( from airplane engines to light bulbs ) extending to the financial sector with ge capital .
our method is able to extract interdisciplinary activities that other methods can not accomplish effectively . a widely used technique that quantifies the tendency of the nodes of a graph to cluster is the clustering coefficient @xmath23 @xcite , which is defined as follows .
assuming that a vertex @xmath10 has @xmath24 neighbors then at most @xmath25 edges can exist between all the neighbors of vertex @xmath10 .
if we denote with @xmath26 the fraction of such existing edges for node @xmath10 , then @xmath23 is defined as the average of @xmath26 over all connected nodes of the network . in fig .
[ fig:4](a ) we plot the values of the clustering coefficient obtained for otns constructed using various time window lengths . as it is expected the clustering coefficient @xmath27 for the case of the mst , since there are no loops in a tree . in the same plot
the horizontal discontinuous line , for comparison reasons , shows the value of the clustering coefficient obtained using the pmfg technique . as we see from fig .
[ fig:4](a ) the pmfg , constructed using the same time series length that we used to calculate the mst , has always higher clustering coefficient in comparison to the otn .
the above result is expected , since pmfg is constructed in such way that many small cliques are formed .
but all these cliques are not necessary meaningful , in the sense that a cluster of stocks belonging to the same sector is , since the formation of cliques is forced by the construction algorithm . on the other hand ,
clustering information is important , because it plays central role in understanding the hierarchical structure of an equity network @xcite , and could be effectively applied in portfolio selection @xcite . in what follows
we introduce an empirical function that can be used to measure ho meaningful are the clusters that we get by filtering methods applied to correlation based networks , if we assume that the best method will cluster all stocks of the same sector together .
of course , as we discussed above , there are stocks that are strongly connected to stocks of different sectors of economic activity , but such stocks are usually the exception , not the rule . in order to calculate the structure of the network
we introduce a structure function @xmath28 .
this function can take values in the range @xmath29 , and it obtains the maximum value @xmath30 for a complete graph that all its nodes belong to the same partition .
the @xmath28 function is defined as , @xmath31 where @xmath32 is the total number of links of the network , @xmath33 is the total number of nodes of the network , @xmath34 is the number of different partitions ( in this case is the number of different economic sectors ) , @xmath35 is the number of nodes that belong to partition @xmath11 , @xmath36 is the number of different clusters that are created from nodes belonging to partition @xmath11 , @xmath37 is the number of possible links between all the nodes of partition @xmath11 , @xmath38 is the number of nodes that form cluster @xmath39 , and @xmath40 is the number of links that connect the nodes of cluster @xmath39 .
we should make clear that in order to use the structure function @xmath28 , the partitioning of the network must be known _ a priori _ , therefore this function does not give information about the partitioning of a graph , it only compares the output of different partitioning methods .
we applied this measure to extract the structural information obtained using the otn for different time windows and we compared it with the structural information obtained using the pmfg .
the results are shown in fig .
[ fig:4](b ) , where we see that for large time window lengths the otn is lower in comparison to the pmfg , but by decreasing the time window length eventually the otn is able to capture more structural properties than the pmfg .
this happens because the otn is based on the mst , which is a relatively stable structure .
thus , for large time window lengths the otn includes only a small fraction of the links of the original fully connected graph , just a few more than the links included in the mst .
but as we decrease the time window length , we capture more information about the evolution of links that become strong only for some period of time .
such links are included in the otn and make it optimal for studying the network at different time periods . if we compare the set of the common links between the mst that is calculated using the full length time series , and the otns that we have calculated for all the time window lengths that are shown in fig .
[ fig:4 ] , we find that on average @xmath41% of the links included in the mst is also included in the otn .
this means that almost all the connectivity information that we extract from the mst is included in the otn , but the otn allows to examine even more details of the system .
we introduced a general method that extracts information stored in complex networks , by using only a subset of the network strongest links , resulting in the overlapping tree network ( otn ) .
our method is based upon the well established technique of the extraction of the minimum spanning tree , but it allows for the filtered network to have loops , and therefore it retains more of its original complexity that the mst .
the added information in the otn is that besides the clustering together of similar nodes as in the mst , we now have the full picture of the connectivity pattern between the hub nodes , and the strength of these connections ( @xmath12 ) .
such strongly interconnected clusters form easily distinguishable communities that can be used to partition the network .
furthermore , the otn includes links that could be strong for only a certain time period or for some specific set of interactions , and then become weak again .
such dynamical transition is not detected by the use of the mst alone . as a consequence the importance of some central nodes of the network detected by the mst , such as , for example the stock of ge , is highlighted by the use of otn .
this method was applied to two different systems and gave interesting insights for both cases .
the first system was the collaboration network of countries participating to at least one of european sponsored research projects . with this approach
we were able to identify a fully connected community of the most frequent hubs of the network .
the strength of the internal links connecting the nodes of this community are much higher than the average strength of the links of the remaining network . therefore this structure
could be described as a nucleus of countries with very strong collaboration links .
all other participating countries were found to play a satellite role around this nucleus .
such information is not palpable by the use of the mst alone because with the mst approach we lose information about the interconnectivity .
but interconnectivity information is very important both for policy makers and for scientists that create consortia and apply for funding to the european commission .
the second system was the network of the 100 most capitalized stocks traded in the nyse .
these stocks form a correlation based network , which is calculated using time series of the daily returns of these equities .
the otn was extracted using different lengths of the sliding time window .
we verified the expected clustering of stocks according to the sectors of economic activity that they belong , but we were able to identify the stocks of general electric ( ge ) , and the american international group ( aig ) as the ones that have the strongest links with stocks from different sectors in the set of 100 highest capitalized stocks . since clustering is an important information , we introduced an empirical function that is able to quantify the result , in terms of clustering information , of different filtering techniques applied in correlation based networks .
we applied this function to the outcome of the otn and of the pmfg methods .
comparison of these methods showed that the pmfg gives better clustering information if the length of the time window used for the otn method is large . for smaller time
windows the otn is able to capture more structural properties , and it is more suitable for studying the network at different time periods .
the authors would like to thank m. tumminello for his valuable comments during the preparation of this work , and l.k .
gallos for his contribution in the preparation of the eu database .
this work was partially supported by a european research nest / pathfinder project , dysonet 012911 , and by the greek general secretariat for research and technology of the ministry of development , pened project 03ed840 . | we introduce a technique that is capable to filter out information from complex systems , by mapping them to networks , and extracting a subgraph with the strongest links .
this idea is based on the minimum spanning tree , and it can be applied to sets of graphs that have as links different sets of interactions among the system s elements , which are described as network nodes . it can also be applied to correlation - based graphs , where the links are weighted and represent the correlation strength between all pairs of nodes . we applied this method to the european scientific collaboration network , which is composed of all the projects supported by the european framework program fp6 , and also to the correlation - based network of the 100 highest capitalized stocks traded in the nyse . for both cases we identified meaningful structures , such as a strongly interconnected community of countries that play important role in the collaboration network , and clusters of stocks belonging to different sectors of economic activity , which gives significant information about the investigated systems . |
we examine the out - of - equilibrium dynamical evolution of density profiles of ultrasoft particles under time - varying external confining potentials in three spatial dimensions .
the theoretical formalism employed is the dynamical density functional theory ( ddft ) of marini bettolo marconi and tarazona [ j. chem .
phys . * 110 * , 8032 ( 1999 ) ] , supplied by an equilibrium excess free energy functional that is essentially exact .
we complement our theoretical analysis by carrying out extensive brownian dynamics simulations .
we find excellent agreement between theory and simulations for the whole time evolution of density profiles , demonstrating thereby the validity of the ddft when an accurate equilibrium free energy functional is employed .
density functional theory ( dft ) is a very powerful tool for the quantitative description of the equilibrium states of many - body systems under arbitrary external fields .
it rests on the exact statement that the helmholtz free energy of the system , @xmath0 $ ] , is a unique functional of the inhomogeneous one - particle density @xmath1 .
moreover , the equilibrium profile @xmath2 minimises @xmath0 $ ] under the constraint of fixed particle number @xmath3 @xcite .
the task is then to approximate the unknown functional @xmath0 $ ] from which all equilibrium properties of the system follow .
much more challenging is the problem of studying _ out - of equilibrium dynamics _ of many - body systems , for which analogous uniqueness and minimisation principles are lacking . in this paper , we present results based on a recently - proposed dynamical density functional theory ( ddft ) formalism and we demonstrate that the latter is capable of describing out - of - equilibrium diffusive processes in colloidal systems at the brownian time scale .
we are concerned with the dynamics of typical soft - matter systems , such as suspensions of mesoscopic spheres and polymer chains in a microscopic solvent @xcite .
the enormous difference in the masses of the suspended particles and the solvent molecules implies a corresponding separation in the relaxational time scales of the two . at times of the order of the fokker - planck scale , @xmath4 , the solvent coordinates are long relaxed to thermal equilibrium . on the brownian diffusive time scale , @xmath5
, the momentum coordinates of the solute particles relax to equilibrium with the heat bath of the solvent molecules and thus a statistical description involving only the positions of the colloids is feasible @xcite . in this regime ,
the evolution of the coordinates @xmath6 of the @xmath3 colloidal particles is described by the set of stochastic langevin equations : @xmath7 + { \bf w}_i(t ) .
\label{bd : eq}\ ] ] in eq .
( [ bd : eq ] ) above , @xmath8 is the pair ( effective ) interaction potential between the mesoscopic particles @xcite , @xmath9 is the externally acting potential and @xmath10 $ ] is a stochastic term representing the random collisions with the solvent molecules and having the properties : @xmath11 where the averages @xmath12 are over the gaussian noise distribution and @xmath13 , the cartesian components . the constants @xmath14 and @xmath15 stand for the mobility and diffusion coefficients of the particles , respectively , and the einstein relation gives @xmath16 . applying the rules of the it stochastic calculus , marconi and tarazona
@xcite recasted the above equations into the form @xmath17 \\ & + & \nabla_{\bf r}\left[\int{\rm d}^3 r'\ , \langle \hat\rho({\bf r } , t ) \hat\rho({\bf r ' } , t)\rangle \nabla_{\bf r } v({\bf r } - { \bf r'})\right ] .
\label{exact.eq}\end{aligned}\ ] ] here , @xmath18 is the usual one - particle density operator and @xmath19 is the noise - average of this quantity . up to this point , all is exact .
now , the following _ physical assumption _ ( a ) is introduced : as the system follows its relaxative dynamics , the instantaneous two - particle correlations can be approximated with those of a system in thermodynamic equilibrium with the same , _ static _
one - particle density @xmath1 as the noise - averaged dynamical one - particle density @xmath20 .
then , eq . ( [ exact.eq ] ) can be cast into a form involving exclusively the _ equilibrium _ density functional @xmath0 $ ] as @xcite @xmath21}{\delta \rho({\bf r } , t)}\right ] , \label{ddft.eq}\ ] ] with the free energy functional @xmath22 = k_bt\int { \rm d}^3 r\rho({\bf r})\left\ { \ln\left[\rho({\bf r})\lambda^3\right ] - 1 \right\ } + f_{\rm ex}[\rho ] + \int{\rm d}^3 r v_{\rm ext}({\bf r } , t)\,\rho({\bf r } ) .
\label{f : eq}\ ] ] the dynamical equation of motion ( [ ddft.eq ] ) was in fact first derived in a phenomenological way by dieterich , frisch and majhofer @xcite . in carrying out concrete calculations with the theory put forward above and in comparing them with brownian dynamics ( bd ) simulation results based on the microscopic equations of motion , eq .
( [ bd : eq ] ) , two sources of possible discrepancies exist : first , the fundamental assumption ( a ) and second the approximate nature of the equilibrium density functional @xmath23 $ ] of eq .
( [ f : eq ] ) .
in this work we focus our attention to _ ultrasoft particles _ for which a very accurate and simple functional @xmath0 $ ] is known , namely the _ mean - field _ or _ random - phase approximation _ ( rpa ) functional given by eq .
( [ mfa : eq ] ) below .
this guarantees that one can explore the accuracy of the fundamental assumption ( a ) under well - defined external conditions .
consider a one - component system of ultrasoft particles .
it has been demonstrated that for such systems the following rpa - functional is quasi - exact @xcite : @xmath24 = \frac{1}{2}\int\int{\rm d}^3r\,{\rm d}^3r ' v(|{\bf r } - { \bf r'}|)\rho({\bf r})\rho({\bf r ' } ) .
\label{mfa : eq}\ ] ] eq .
( [ ddft.eq ] ) takes now with the help of eqs .
( [ f : eq ] ) and ( [ mfa : eq ] ) the form @xmath25 given an initial density field @xmath26 and a prescribed external potential @xmath27 , eq .
( [ explicit.eq ] ) can be numerically solved to yield @xmath28 . in this work
we apply an ultrasoft gaussian pair potential between the interacting particles that has been shown to describe the effective interaction between the centres of mass of polymer chains in athermal solvents @xcite @xmath29.\end{aligned}\ ] ] we set @xmath30 providing the energy unit for the problem , whereas @xmath31 , which corresponds to the gyration radius of the polymers , will be the unit of length henceforth .
accordingly , the natural time scale of the problem , providing the unit of time in this work , is the brownian time scale @xmath32 .
eq .
( [ explicit.eq ] ) is solved using standard numerical techniques , and for a variety of time - dependent confining external potentials @xmath33 to be specified below .
brownian dynamics simulations of eq .
( [ bd : eq ] ) are also straightforward to carry out .
the langevin equations of motion including the external field are numerically solved using a finite time - step @xmath34 in all simulations , and the technique of ermak @xcite . in order to obtain the time - dependent density @xmath28 we perform a large number @xmath35 of independent runs , typically @xmath36 , and average the density profile over all configurations for a fixed time @xmath37 .
we focus to external fields that correspond to a sudden change , i.e. , @xmath38 .
these force the system to relax from the equilibrium density @xmath39 , compatible to the external potential @xmath40 , to the new equilibrium density @xmath41 , corresponding to the external potential @xmath42 .
important questions related to such processes are what is the typical relaxation time @xmath43 for such a procedure and how does the system cross over from one equilibrium density to the other .
we consider two kinds of confinements : spherical ones , @xmath44 , where @xmath45 , and planar ones between two walls perpendicular to the @xmath46-cartesian direction , @xmath47 . in these cases
we obtain @xmath48 and @xmath49 , correspondingly , and the solution of eq .
( [ explicit.eq ] ) is greatly simplified since the integrals take the form of one - dimensional convolutions that can be evaluated very rapidly by use of fast fourier transform techniques .
three different external confinements have been specifically investigated .
two spherical ones @xmath50 ; \label{v1:eq } \\ v_{\rm ext}^{(2)}(r , t ) = \phi_0\left[(r / r_{1})^{10}\theta(-t ) + ( r / r_{2})^{10}\theta(t)\right ] ; \label{v2:eq } \label{spherical : eq}\end{aligned}\ ] ] and one slab confinement @xmath51 .
\label{v3:eq}\end{aligned}\ ] ] the energy scale @xmath52 sets the strength of the confining potential and is fixed to @xmath53 for all three confinements .
the only difference between the external potential for times @xmath54 and for @xmath55 lies in the different length scales @xmath56 for eqs .
( [ v1:eq ] ) and ( [ v2:eq ] ) , and @xmath57 for eq .
( [ v3:eq ] ) . for each confinement
we consider two cases that give rise to two different dynamical processes : @xmath58 ( @xmath59 ) , enforcing an _ expansion _ of the system and @xmath60 ( @xmath61 ) , bringing about a _
compression _ of the same . for the spherical confinements
an additional parameter is the particle number @xmath62 which is a conserved quantity , as is clear from eq .
( [ ddft.eq ] ) that has the form of a continuity equation .
@xmath3 enters the formalism through the normalisation of the density field at @xmath63 . for both spherical confinements
the particle number is @xmath64 . in the slab confinement ,
eq .
( [ v3:eq ] ) , the conserved quantity is the density per unit area @xmath65 .
we choose @xmath66 . in all cases
examined , the typical relaxation time was found to be of order @xmath67 ; after typically @xmath68 , the system fully relaxes into the new equilibrium profile . in the spherical confining potential @xmath69 with ( a ) @xmath70 and @xmath71 and ( b ) @xmath72 and @xmath73 .
the shown profiles are for the times @xmath74 , @xmath75 and @xmath76 , all in @xmath67-units .
the last time is practically equivalent to @xmath77 , since the system there has fully relaxed into equilibrium . in all figures ,
red curves denote the initial and blue ones the final static profile.,title="fig:",width=264 ] in the spherical confining potential @xmath69 with ( a ) @xmath70 and @xmath71 and ( b ) @xmath72 and @xmath73 .
the shown profiles are for the times @xmath74 , @xmath75 and @xmath76 , all in @xmath67-units .
the last time is practically equivalent to @xmath77 , since the system there has fully relaxed into equilibrium . in all figures ,
red curves denote the initial and blue ones the final static profile.,title="fig:",width=264 ] in fig . [
harmon : fig ] we show the results for the harmonic confining potential of eq .
( [ v1:eq ] ) .
it can be seen that the theory reproduces the time evolution of the density profile , both for the expansion [ fig .
[ harmon : fig](a ) ] and the compression [ fig .
[ harmon : fig](b ) ] processes .
an asymmetry in the two processes can be already discerned : the compression is not the ` time reversed ' of the expansion and this effect will be much stronger in the examples to follow .
though the profiles of the system are considerably different from those of an ideal gas , i.e. , effects of the interparticle interaction are present , the confining potential is smooth enough , so that the profiles are devoid of pronounced correlation peaks . in the spherical confining potential @xmath78 with ( a ) @xmath70 and @xmath79 and ( b ) @xmath80 and @xmath73 .
the shown profiles are for the times @xmath74 , @xmath81 and @xmath82 ( in units of @xmath67).,title="fig:",width=340 ] in the spherical confining potential @xmath78 with ( a ) @xmath70 and @xmath79 and ( b ) @xmath80 and @xmath73 .
the shown profiles are for the times @xmath74 , @xmath81 and @xmath82 ( in units of @xmath67).,title="fig:",width=340 ] the situation is different for the external potential of eq .
( [ v2:eq ] ) . here
, the power - law dependence is much more steep , so that the gaussian fluid develops correlation peaks close to the ` walls ' of the confining field .
the dynamical development of the profiles for the expansion and compression processes are shown in fig .
[ r10:fig ] . here , the asymmetry between the expansion and the compression processes is evident . in the former case , seen in fig .
[ r10:fig](a ) , the expansion of the confining potential leaves behind a density profile that has very strong density gradients close to the boundary of the initial confinement . since the latter ceases to act at @xmath83 , this leaves at @xmath84 instantaneously a region @xmath85 that is devoid of particles but in which the new external potential is essentially zero .
this leads to a collective diffusion of the particles towards the boundaries set by the new potential .
correspondingly , the high density peaks decrease rapidly and leak outward . in the inner region , @xmath86 of the profile ,
the dynamics is much slower and the relaxation to the final plateau there takes place at the end of the process , causing thereby the final development of the new , weaker correlation peaks close to the location of the boundary , @xmath87 . the compression process , depicted in fig .
[ r10:fig](b ) runs very differently .
there , the initial ` closing ' of the potential from @xmath88 to @xmath89 leaves at @xmath84 a region of high density at @xmath85 that now finds itself within a strongly repulsive external field .
there is an extremely rapid shrinking there , accompanied by the development of very high correlation peaks that actually ` overshoot ' in height with respect to the final equilibrium profile .
initially , the region in the centre of the sphere remains unaffected and only as the high peaks start diffusing does material flow toward the centre and at the latest stage of the dynamics the profile at @xmath90 reaches its new equilibrium value . , defined in eq .
( [ moment ] ) plotted against the time @xmath37 for the spherical confinement @xmath78 .
circles correspond to radii @xmath70 and @xmath79 ( expansion ) and squares show the resulting curve for the inverse process , @xmath80 and @xmath73 ( compression ) .
the lines are the analytical fits shown in the text .
solid line : eq .
( [ m2plus : eq ] ) ; long - dashed line : eq .
( [ m2minus : eq ] ) .
the arrows mark the characteristic time scales defined in these two equations.,width=302 ] in order to quantify better this asymmetry and also extract characteristic time scales for the two dynamical processes , we consider the second moment of the density , @xmath91 , defined through @xmath92 the quantity @xmath91 is a quantitative measure of the spread of @xmath93 around the centre of the external field and its time evolution is shown in fig .
[ m2:fig ] .
let the superscript ` @xmath94 ' denote the expansion and the superscript ` @xmath95 ' the compression process .
obviously , it holds @xmath96 .
we notice that for both processes @xmath91 is a monotonic function of @xmath37 but some interesting differences arise when one fits the two curves by analytic functions , shown as lines in fig .
[ m2:fig ] .
the expansion can be very accurately described by a single exponential : @xmath97 [ 1-\exp(-t/\tau^{+ } ) ] , \label{m2plus : eq}\ ] ] with the characteristic time scale @xmath98 . however , a double - exponential fit is necessary to parameterise the compression process , namely @xmath99\exp(-t/\tau_2^{- } ) , \label{m2minus : eq}\ ] ] with the fit parameter @xmath100 and the _ two _ characteristic time scales @xmath101 and @xmath102 . since @xmath103
, it follows that the compression process is at any rate faster than the expansion one .
the occurrence of the two distinct time scales @xmath104 in the compression requires some explanation .
the fast process that takes place at times @xmath105 corresponds to the abrupt shrinking of the profile at the wings of the distribution and is caused exclusively by the ` closing ' of the external field .
this is the same mechanism that brings about the overshooting of the density peaks .
once this is over , diffusion within the now already confined system takes place and the second characteristic time scale , @xmath106 , is solely determined by the interaction potential @xmath107 and the average particle density . in the expansion process , the first mechanism is absent thus a single time scale , @xmath108 , shows up , which is of intrinsic origin exclusively . since stronger density gradients occur during the compression than during the expansion process , even the larger of the two time scales of the compression , @xmath106 , is smaller than @xmath108
. the denser the system is , the faster the collective diffusion towards equilibrium . in a slab
confining potential @xmath109 with ( a ) @xmath110 and @xmath111 and ( b ) @xmath112 and @xmath113 .
the shown profiles are for the times @xmath114 and @xmath115 ( in units of @xmath67).,title="fig:",width=340 ] in a slab confining potential @xmath109 with ( a ) @xmath110 and @xmath111 and ( b ) @xmath112 and @xmath113 .
the shown profiles are for the times @xmath114 and @xmath115 ( in units of @xmath67).,title="fig:",width=340 ] finally , we turn our attention to the slab confinement .
the results from theory and simulation are shown in fig .
[ slab : fig ] .
once more it can be seen that the ddft offers an excellent description of the dynamics of the system .
the same asymmetry between expansion and compression that was seen in the spherical confinement also shows up for the case of the slab , including the overshooting of the peaks during the compression process .
in addition , the density profiles develop during their evolution secondary oscillations that are also very well reproduced by the theory . in summarising
, we have demonstrated that the dynamical density functional theory of marini bettolo marconi and tarazona @xcite , when supplemented by an accurate equilibrium density functional , can provide an excellent description of out - of - equilibrium dynamics of colloidal systems at the brownian time scale .
the accuracy of the ddft formalism has already been successfully tested for the system of one - dimensional hard rods @xcite , for which the exact density functional @xmath0 $ ] is known . to the best of our knowledge ,
this is the first study of the validity of ddft in three dimensions . as the phenomenology in 3d is much richer than in 1d , including the possibility of phase transitions , many intersecting ways for future applications open up .
discussions with andrew archer , bob evans , wolfgang dieterich and pedro tarazona are gratefully acknowledged .
this work has been supported by the deutsche forschungsgemeinschaft through the sfb tr6 . |
|
braneworld models with large extra dimensions @xcite have attracted wide attention owing to their success in solving some open questions in modern physics @xcite . to realize the braneworld idea one should find a mechanism of localization , however , there are some difficulties in this matter within standard brane approaches @xcite .
it is especially economical to consider models with a pure gravitational localization mechanism , since gravity has universal couplings with all matter fields . in this paper
we consider one such scenario within the standing waves braneworld @xcite and explicitly show existence of the universal gravitational trapping of all kinds of matter fields on the brane .
main differences of the model with the most brane scenarios are that our metric _ ansatz _ is time - dependent and contains increasing warp factor
. the standing waves braneworld is formed by the background solution @xcite : @xmath0 of the coupled einstein and massless ghost scalar field equations . in ( [ metric ] ) @xmath1 denotes the extra space - like coordinate , @xmath2 is a curvature scalar ( we choice increasing warp factor ) , the determinant , @xmath3 is @xmath4-independent and the oscillatory function @xmath5 has the form : @xmath6 here @xmath7 denotes the oscillation frequency , @xmath8 is an integration constant and @xmath9 is the second - order bessel function of the second kind .
the solution ( [ metric ] ) describes the brane located at the node of the standing bulk wave ( [ u ] ) . in this paper we consider the case with one node , i.e. the function @xmath9 in ( [ u ] ) has a single zero at the position of the brane , @xmath10 .
this is accomplished by the assumption : @xmath11 in the equations of the matter fields , the oscillatory function ( [ u ] ) enters via exponentials .
we suppose that the frequency @xmath7 of standing waves is much larger than the frequencies associated with the energies of particles on the brane and in these equations we can perform the time averaging of oscillating exponents .
non - vanishing averages are expressed by the formula @xcite : @xmath12 where @xmath13 is the modified bessel function .
the brane at the node of the standing wave ( the point where @xmath14 and @xmath5 vanish ) can be considered as a 4d island , where the matter particles are assumed to be bound .
indeed the system of 5d geodesic equations of motion for a classical particle ( or a photon ) reduces to @xcite : @xmath15 where @xmath16 is the velocity along the brane and the constant @xmath17 corresponds to the energy of the particle per unit mass ( for massive particles @xmath18 and for photons @xmath19 ) . from ( [ dr ] )
it is clear that the motion towards the extra dimension @xmath1 is possible when @xmath20 the term @xmath21 plays the role of the trapping gravitational potential and there exists maximal distance in the bulk @xmath22 . in the standard brane approach with decreasing warp factor ( @xmath23 ) localization
was achieved due to the fact that the extra space actually is finite @xcite . in our case ,
the increasing of the brane warp factor , @xmath24 , creates the potential well that confines particles .
let us now consider the localization problem of quantum fields with various spin using the brane solution ( [ metric ] ) @xcite . to have localized fields on a brane coupling constants appearing after integration of their lagrangians over the extra
coordinate , @xmath1 , must be non - vanishing and finite . using numerical solutions we shall demonstrate existence of normalizable zero modes of various fields on the brane within the model
in this section , using numerical solutions , it is shown localization of massless modes of the spin-0 , spin-1 and spin-2 particles on the brane .
at first let us consider 5d massless scalar field .
we assume that for the case @xmath25 ( @xmath26 is the typical energy of scalar particles on the brane ) variables can be separated @xcite , @xmath27 capital latin letters numerate 5d coordinates , while greek ones are used for 4d indices .
then the action integral for scalar fields can be split in two parts : @xmath28~,\ ] ] where prime denotes the derivative with respect to extra coordinate @xmath1 .
we are looking for zero mode solution on the brane , @xmath29 , and assume that energy - momentum along the brane , @xmath30 , obeys the dispersion relation : @xmath31 than the 5d klein - gordon equation after time averaging reduces to @xcite : @xmath32 where @xmath33 close to the brane the function @xmath34 , which describes oscillatory properties of standing waves , and the extra dimension part of wave function @xmath35 , behave as : @xmath36
so for the equation ( [ xi-2 ] ) we use the boundary conditions : @xmath37 figure 1 shows the numerical solution of the equation ( [ xi-2 ] ) for these boundary conditions for the following values of the parameters : @xmath38 we see that @xmath39 , which governs of probability of scalar particles to reach the bulk , falls off from the brane to zero .
the integrals over @xmath1 in the action ( [ sphi ] ) are convergent if integrand functions decrease stronger than @xmath40 .
figure 2 displays the products of @xmath1 on these integrand functions .
we see that they decrease , i.e. the integrals over @xmath1 in the action ( [ sphi ] ) are convergent , hence 4d scalar fields are localized on the brane .
( [ sphi]).,scaledwidth=70.0% ] recall that the action of scalar fields ( [ sphi ] ) due to the determinant ( [ determinant ] ) contains increasing exponential factor ( @xmath41 ) .
this is the reason why in the original brane models @xcite the scalar field zero modes with the constant extra dimension parts can be localized on the brane only in the case of decreasing warp factor ( i.e. @xmath23 ) . in our model
the extra part of wave function ( [ solution - s ] ) is not constant , moreover , from figure 1 it is clear that it strongly decreases and according to figure 2 the integrals over @xmath1 in the action ( [ sphi ] ) are convergent .
it is known that the transverse traceless graviton modes obey the equations similar to the massless scalar fields in a curved background .
accordingly , the condition of localization of spin-@xmath42 graviton field is equivalent to that of scalar field considered above @xcite . in the case of 5d abelian vector fields ( the generalization to the case of non - abelian gauge fields is straightforward ) we are looking for the solution of field equations in the form @xcite : @xmath43 than the 5d action of vector fields
can be written as : @xmath44 ~.\end{aligned}\ ] ] we require the existence of 4d vector zero mode on the brane , @xmath45 where @xmath30 obey dispersion relation ( [ dispersion ] ) .
then 5d maxwell equations after time averaging yields the single equation for @xmath46 : @xmath47 where @xmath34 is done by ( [ p2 ] ) .
close to the brane , @xmath48 , and we set the boundary conditions : @xmath49 from figure 3 we see that the solution of the equation ( [ v ] ) for these boundary conditions , or the probability of photon to leave the brane , falls off in the bulk down to zero .
as for the case of scalar field the integrals over @xmath1 in the vector functions action ( [ vectoraction ] ) are convergent if integrand functions decrease stronger than @xmath40 . from the figure 4
we see that the products of @xmath1 with integrand functions decrease , i.e. the integral over @xmath1 in ( [ vectoraction ] ) is convergent and 4d photon is localized on the brane . )
.,scaledwidth=70.0% ] in 5d brane models the implementation of pure gravitational trapping mechanism of vector field particles remains the most problematic @xcite .
the reason is that in the vector action ( [ vectoraction ] ) the extra dimension parts of the determinant ( [ determinant ] ) and two metric tensors with upper indices cancel each other .
because of this in the original brane models the vector field zero modes ( with the constant extra dimension part ) can not be localized on the brane for any sign of @xmath50 . in our model
the extra dimension part of vector field @xmath46 is rapidly decreasing function ( fig .
3 ) and the integral over @xmath1 in ( [ vectoraction ] ) is convergent .
now we investigate the localization problem for massless fermions @xcite . for minkowskian @xmath51 gamma matrices we use the weyl basis , @xmath52 and 5d gamma matrices @xmath53 , where @xmath54 .
the inverse _ fnfbein _ for the metric ( [ metric ] ) , we choice as : @xmath55 non - vanishing components of the spin - connection in the background ( [ metric ] ) , which enter covariant derivatives defined as : @xmath56 after time averaging are given by : @xmath57 for the wavefunction of the bulk fermion field we use the chiral decomposition : @xmath58 where @xmath59 and @xmath60 are the extra dimension factors of the left and right fermion wavefunctions respectively .
then the 5d dirac action for massless fermions can be written as : @xmath61 - \overline \psi_l \psi_r \int dr e^{4a|r|}\lambda \left[\rho ' + 2a~ sgn(r ) \rho \right]\right\}~.\nonumber\end{aligned}\ ] ] we assume that 4d left and right dirac spinors ( @xmath62 ) are solutions of the free 4d massless dirac equations : @xmath63 where @xmath30 satisfy the dispersion relation ( [ dispersion ] ) and the constant 2-spinors @xmath64 and @xmath65 obey : @xmath66 derived from ( [ spinoraction ] ) 5d dirac equation , after time averaging , takes the form : @xmath67\psi -\gamma^5 \left[2a ~sgn(r ) + \partial_r \right]\psi = 0~.\ ] ] using the expressions ( [ psi - free ] ) and ( [ relations ] ) it can be rewritten as the system : @xmath68&\sigma^i{\cal p}_i(r)\\ - \sigma^i{\cal p}_i(r)&e^{a|r|}\left[2a ~sgn(r ) + \partial_r \right ] \end{array } \right)\left ( \begin{array}{*{10}{c } } \rho ( r)r\\ \lambda ( r)l\end{array } \right ) = 0~.\end{aligned}\ ] ] here we have introduced the functions @xmath69 : @xmath70 p_x , \nonumber \\ { \cal p}_y ( r ) & = & \left(\left\langle e^{-u/2 } \right\rangle - 1 \right)p_y = \left[i_0\left(z/2\right)-1\right ] p_y , \nonumber \\ { \cal p}_z ( r ) & = & \left(\left\langle e^{u } \right\rangle - 1 \right)p_z = \left[i_0\left(z\right)-1\right ] p_z,\end{aligned}\ ] ] where @xmath71 is defined in ( [ u ] ) . to find boundary conditions for the system ( [ l - r ] ) note that from the last two terms of ( [ spinoraction ] ) we see that the action is even under the @xmath72 orbifold symmetry
if @xmath73 is an even function of @xmath1 and @xmath74 is odd ( or vice versa ) with respect to the transformation @xmath75 .
we require , @xmath76 what can be accomplished if we impose that the bulk fermion wavefunction is even under five - dimensional parity @xcite : @xmath77 then from the symmetry of the problem with respect to the transformation @xmath75 we get the following important boundary conditions : @xmath78 solutions of the dirac equations ( [ l - r ] ) with the boundary conditions ( [ baundary ] ) are displayed on the figure 5 .
we see that in our model the extra dimension part of the left spinor wave function @xmath59 has maximum on the brane and decreases in the bulk .
while , as a consequence of ( [ baundary ] ) , in our setup the right fermionic modes are absent on the brane and @xmath79 has maximum in the bulk outside the brane .
integrals over @xmath1 in the spinor field action ( [ spinoraction ] ) are convergent if integrand functions decrease stronger than @xmath40 .
this feature for ( [ spinoraction ] ) is demonstrated on the figure 6 .
so in our model left massless fermions are localized on the brane and right fermions are localized in the bulk .
( [ spinoraction]).,scaledwidth=70.0% ]
in this section we want to estimate masses of kaluza - klein excitations . for simplicity let us consider the case of 5d scalar particles . for a massive mode the dispersion relation ( [ dispersion ] )
takes the form : @xmath80 where @xmath81 denotes the mass of lightest kk mode on the brane .
then the equation for scalar particles ( [ xi-2 ] ) transforms to : @xmath82\varsigma = 0~.\ ] ] it is more convenient to put ( [ varsigma - m ] ) into the form of an analogue non - relativistic quantum mechanical problem by making the change : @xmath83 for @xmath84 we find : @xmath85 \psi = 0~,\ ] ] where the function , @xmath86 is the analog of non - relativistic potential .
because this potential contains dirac delta function , it becomes infinitely large at the point @xmath87 .
this means that if we interpret @xmath88 as a wave function , it must vanish at the point @xmath87 .
profile of the @xmath89 is shown on the figure 7 . )
.,scaledwidth=70.0% ] to show existence of mass gap between zero and massive modes we approximate @xmath89 by the rectangular potential and study quantum mechanical problem in potential well with the parameters : @xmath90 here @xmath64 is the width of the rectangular effective potential @xmath89 , which can be estimated from ( [ u ] ) . inside of the potential well the equation ( [ psi ] )
obtains the form : @xmath91\psi_1 = 0~.\ ] ] thus in the first region , if @xmath92 the solution can be written in the form : @xmath93 where @xmath94 , @xmath95 and @xmath96 , @xmath97 are integration constants and bessel functions , respectively , and we have introduced the parameter : @xmath98 in the second region , where the potential is infinitely large , we have : @xmath99 the boundary and continuity conditions at the point of junction : @xmath100 as usual , lead to the quantization of the mass : @xmath101 which can be satisfied if @xmath102 so we find that in our model kaluza - klein excitations have discreet spectrum and the mass of the lightest kk massive mode is of order of the curvature scalar , @xmath103 this miens that kk states are very heavy since @xmath50 correspond to the energy scale characterizing the brane as a topological defect in higher - dimensional space - time .
in this letter within the 5d standing wave braneworld model we numerically find profiles of massless matter field wave - functions in the bulk . using numerical solution in action integrals
we have demonstrated the existences of gravitational localization of all kinds of matter fields on the brane . for the case of fermions we found that , while left ones are localized on the brane , the right massless fermions are absent on the brane , they are localized in the bulk .
we have also estimated masses of kk modes and found that , as in the standard single brane models , they are very heavy - of order of the energy scale associated with the brane .
this research was supported by the grant of shota rustaveli national science foundation @xmath104 . | within the 5d standing wave braneworld model numerical solutions of the equations for matter fields with various spins are found .
it is shown that corresponding action integrals are factorizable and convergent over the extra coordinate , i.e. 4d fields are localized on the brane .
we find that only left massless fermions are localized on the brane , while the right fermions are localized in the bulk .
we demonstrate also quantization of kaluza - klein excited modes in our model . 0.3 cm pacs numbers : 04.50.-h , 11.25.-w , 11.27.+d keywords : brane ; standing waves ; zero modes |
in @xcite , a gauge - invariant calculation was presented showing that the propagator for a charged particle acquires an interesting fractal structure from its self - interaction in quantum electrodynamics . in simple terms , the usual dirac propagator @xmath0 is replaced by the nonlocal expression @xmath1^{\big(1+(\alpha /\pi ) \big)}}\right\ } , \label{gai27}\end{aligned}\ ] ] where @xmath2 , @xmath3 is a short distance length scale , the fractional exponent @xmath4 is a function of the coupling strength @xmath5 which we find to be @xmath6 and the usual gamma function is @xmath7 this represents a rather radical departure from the usual textbook discussions of particle propagators ( with the notable exception of @xcite ) and from the usual wigner classification of elementary particles in terms of a sharp mass and spin ( see , for example , @xcite ) .
this is despite the fact that it is well - known that particles coupled to long - range massless fields can not have sharp masses@xcite
. a good review of the basic issues can be found in @xcite . aside from having been neglected in textbooks , the actual exponent in the above expression has been the subject of some controversy
. there are calculations in the literature based on either infinite sums of logarithmic feynman diagrams@xcite or non - perturbative schwinger@xcite computations which argue that such an electron propagator should be of the _ form _ we obtained : @xmath8^{(1+\gamma ) } } \right\ } , \label{appelquist}\ ] ] . in the literature
there has not been full agreement about what constitutes the correct function @xmath9 , nor has there been a consensus as to whether or not it can be set to zero by a suitable choice of gauge .
appelquist and carazzone@xcite argued that @xmath10 to leading order , in conflict with earlier work based on summing logarithms@xcite .
reference @xcite does not actually derive the exponent , rather citing the earlier work , so there is a possibility that a typographical error may be involved .
the fourth volume of the landau and lifschitz course of theoretical physics@xcite starts off in agreement with @xmath11 but ultimately sets @xmath12 by giving a small mass to the photon .
such a photon mass explicitly breaks both gauge invariance and conformal / scale invariance and must be rejected on physical grounds .
intuitively , what goes wrong is that the fractal structure of the propagator for a charged particle must be due to the fact that ever increasing wavelengths one can produce more and more soft photons without limit .
a photon mass implies a maximum photon wavelength and a loss of scale invariance .
there have been other approaches in the literature , often tackling only the case of charged scalar fields , or again arguing that the anomalous dimension can somehow be set to zero .
for example , a path integral@xcite approach using the schwinger@xcite proper time representation of the propagator and some work by bloch and nordsieck@xcite on soft photon emission gives the same sort of result as ours , but with the final answer given only for charged scalar fields .
these results are argued to be gauge invariant in such a way that the singularity structure can be returned to a simple pole by a choice of gauge . as argued above
, this is unphysical as there is a real meaning to a fractional exponent and the failure of charged particles to have sharp masses .
fried also discusses this problem@xcite as do johnson and zumino@xcite , and stefanis and collaborators@xcite .
batalin , fradkin and schvartsman have made a similar gauge dependent calculation for scalar particles@xcite .
in addition , there are possible experimental consequences of the radical change in the nature of the singularity in the propagator .
handel has argued@xcite that the change from a pole into a branch point has measurable physical implications for `` @xmath13 '' noise in the schrdinger ( non - relativistic ) limit .
the result is clearly not analytic in @xmath14 and requires a non - perturbative approach .
such non - analyticity in @xmath14 for an all - orders or non - perturbative calculation could be anticipated on physical grounds from the arguments of dyson against the convergence of perturbative expansions in qed@xcite .
given the apparent confusion in the literature , we present a simple derivation which preserves gauge invariance throughout . with this in hand
, we then discuss the physical interpretation of the result , extend it to include newtonian gravity , discuss the fractal dimension one can associate with paths involved , and finally introduce a new way to think about confinement in these terms .
for completeness , we rederive the expression for the dressed electron propagator presented in @xcite . for an electron in an external electromagnetic field , @xmath15 the dirac propagator @xmath16 may be solved employing the function @xmath17 ; @xmath18 the hamiltonian of the electron can be written @xmath19 and with @xmath20 , one may define the amplitude for the electron to go from _ y _ to _ x _ in a proper time @xmath21 as the matrix element @xmath22 from eqs.([gia3 ] ) , ( [ gia4 ] ) and ( [ gia5 ] ) follows the electron propagator expression @xmath23 one also has a lagrangian @xmath24 as we are interested here in the infrared limit , spin - flips are suppressed and we can neglect the corresponding term in the hamiltonian .
this corresponds to usual bloch - nordsieck replacement of @xmath25 by four velocity @xmath26
one simply thinks of the dirac spinor having only one nonvanishing component representing an electron of given spin state and allows no coupling to the other spin state ( and , of course , no coupling to positrons ) . with this approximation , eq .
( [ gia5 ] ) which gives the propagation amplitude for an electron can be written in the lagrangian _ path integral _ formulation @xmath27/\hbar } \prod_\sigma dx(\sigma ) , \nonumber \\
{ \cal s}[x;a]=\int_0^\tau { \cal l}(\dot{x}(\sigma ) , x(\sigma ) ; a)d\sigma .
\label{gia10}\end{aligned}\ ] ] this expression deserves some comments .
the paths @xmath28 which are integrated over represent virtual histories for the electron .
the proper time @xmath29 for each path in the sum is not given by the classical expression @xmath30 , so in integrating over paths one is integrating here over all proper times .
for each one of these paths , with its own proper time , the interaction between the electron and an elecromagnetic vector potential @xmath31 is described by the action @xmath32 where the integral is along the worldline @xmath33 . for the case of no external field , we want the action due to self - interaction ( interaction with vacuum fluctuations ) , so one wants to apply the rule @xmath34 in the above eq.([gia12 ] ) , the subscript `` @xmath35 '' denotes time ordering , @xmath36 denotes the operator vector potential field and the photon propagator is given by @xmath37 the action form in eq.([gia12 ] ) is of a well known form@xcite . as noted earlier , we have bypassed the usual bloch - nordsieck replacement of @xmath25 by four velocity @xmath26 by simply evaluating a phase in the soft photon infrared limit that we are considering here . the propagator may be written @xmath38 where the parameter @xmath39 fixes a gauge . because the worldline of the electron never begins nor ends the partial derivative terms in eq.([gia14 ] )
do not contribute to the self action in eq.([gia12 ] ) .
this requirement that the worldline have no endpoints is required by charge conservation ( or , equivalently , by gauge invariance ) .
independent of any choice of the gauge parameter @xmath39 then , we have @xmath40 in the absence of any external field ( other than that due to vacuum fluctuations ) we have now derived expressions for the renormalized vacuum electron propagator @xmath41 the functional integral expression for @xmath42 is given by @xmath43/\hbar } \prod_\sigma dx(\sigma ) , \nonumber \\
\tilde{\cal s}[x]=\int_0^\tau { \cal l}_0(\dot{x}(\sigma ) ) d\sigma + s_{self}[x ] , \label{gia17}\end{aligned}\ ] ] wherein the _ free _ electron lagrangian is @xmath44 and the self action is given by eqs.([gia14 ] ) and ( [ gia15 ] ) as @xmath45=\frac{i\hbar \alpha } { 2\pi } \int_0^\tau \int_0^\tau \frac{\dot{x}^\mu ( \sigma_1)\dot{x}_\mu ( \sigma_2)d\sigma_1 d\sigma_2 } { ( x(\sigma_1 ) -x(\sigma_2))^2+i0^+}\ . \label{gia19}\ ] ] the divergent piece of the self action @xmath46=\frac{\delta m}{2}\int_0^\tau ( \dot{x}^\mu ( \sigma ) \dot{x}_\mu ( \sigma ) -c^2)d\sigma , \nonumber \\
\label{gia20}\end{aligned}\ ] ] the formally infinite self - mass can be absorbed into a redefinition of the finite physical mass @xmath47 .
thus , eq.([gia17 ] ) is renormalized to @xmath43/\hbar } \prod_\sigma dx(\sigma ) , \nonumber \\
\tilde{\cal s}[x]=\int_0^\tau { \cal l}_m(\dot{x}(\sigma ) ) d\sigma + iw[x ] \nonumber \\ { \cal l}_m(\dot{x})=\frac{1}{2}m(\dot{x}^\mu \dot{x}_\mu -c^2 ) \nonumber \\
w[x;\tau ] = { \im } ms_{self}[x ] , \nonumber \\ w[x;\tau ] = \frac{\hbar \alpha } { 2\pi } \int_0^\tau \int_0^\tau \frac{\dot{x}^\mu ( \sigma_1)\dot{x}_\mu ( \sigma_2)d\sigma_1 d\sigma_2 } { ( x(\sigma_1 ) -x(\sigma_2))^2}\ . \label{gia21}\end{aligned}\ ] ] for a straight - line path @xmath48 with @xmath49 , one finds @xmath50 this expression as it stands is infinite and requires regularization .
using differential regularization@xcite
we have @xmath51 the solution to the differential equation eq.([gia23 ] ) with a logarithmic cut - off @xmath3 is @xmath52 from eqs.([gia21 ] ) and ( [ gia24 ] ) , one finds @xmath53 wherein @xmath54 is the proper time green s function for a particle of mass @xmath55 with the corresponding free lagrangian @xmath56 .
to exponentially lowest order in @xmath57 , one then has @xmath58 eqs.([gia16 ] ) and ( [ gai26 ] ) then imply @xmath59 and we obtain eq .
( [ gai27 ] ) .
we chose differential regularization as simple and convenient , and preserving gauge invariance , but other regularizations can also be used as long as they are also gauge invariant . we plan to return to this issue in a later publication in more detail . in a sense ,
the form of the answer is intuitively clear since , on dimensional grounds , the expression to be regularized can only be logarithmic .
the sign , as we shall see below , has a physically sensible interpretation , and is in agreement with reference @xcite and with reference @xcite if one assumes that there was a typographical error going from reference @xcite .
it is interesting to consider what the physical interpretation of the non - integer exponent in the radiatively corrected dirac propagator means .
first of all , the fact that the exponent is non - integer means that the renormalized dirac operator is non - local@xcite .
this was , of course , to be expected since the electromagnetic field has infinite range .
non - locality has been previously introduced ad - hoc@xcite as a regularization tool . here
it appears naturally , suggesting that some form of regularization of otherwise formally divergent expressions may be implicit in at least some quantum field theories with massless fields , but appearing only when one goes beyond perturbation theory .
we will also see later that there is evidence that the corrections to the exponent can be changed when other fields are added , and need not always be of the same sign as in pure qed .
the appearance of this sort of non - locality also makes the analytic but rarely used regularization proposed by speer@xcite more physically motivated .
so far we have only been able to treat the long - wavelength approximation since we expect a quenched approximation in which we can ignore electron loops to be a reasonable one .
the physical picture would be one where , as one backs away from the worldine of an electron , one continues to see photons radiated and absorbed , but now of longer and longer wavelength .
this would suggest a fractal@xcite structure , which is made precise by the above derivation .
such notions of scaling and fractality are not new in qed and in quantum field theory in general , but are often considered in the high energy , ultraviolet limit@xcite .
this continues without limit at longer and longer distance scales since there is no minimum energy photon ( photons are massless ) and this self - similarity reflects the scale invariance of maxwell s equations .
if the photon is given a mass , however small , this structure will break down asymptotically , since now there is a minimum energy required to create a virtual photon , and at distances greater than the corresponding compton wavelength one will get the non - interacting dirac propagator .
this was by lifshitz _
et al._@xcite , and this argument makes clear how breaking gauge invariance , _
i.e. _ including a photon mass , removes the anomalous scaling behavior here derived for gauge invariant qed .
the fact that a particle is non - localizable , at least in part due to its electromagnetic field which extends over all space , is interesting .
the feeling of this calculation is such that at least part of what one thinks of as quantum - mechanical about an otherwise point - particle ( its lack of localizability ) may arise from the non - perturbative quantum mechanics of its self - interaction@xcite . at shorter
distance scales one might again expect an anomalous dimension , but now the calculations are more complicated since one must imagine more and more electron ( and other charged particle ) loops contributing , with an ever `` frothier '' structure at smaller and smaller distance scales .
in fact , one would not even expect a constant exponent which is independent of momentum since one expects a running @xmath14 with values which change with momentum scale . in the infrared limit
one simply goes to @xmath60 which is the thompson value and there are no additional complications . in the ultraviolet limit ,
one expects a continuously changing exponent and thus a multifractal as opposed to fractal structure .
in addition , the lack of asymptotic freedom leads one to expect trouble at very short distance scales unless , as noted above , other fields enter significantly .
for quantum gravity one can do the analysis in much the same way as for quantum electrodynamics .
the ultraviolet divergences need not worry us since we are dealing with a strictly infrared problem .
there should be graviton - graviton interactions , but if we neglect these as small compared to graviton - electron interactions we can just repeat what was done for electrostatics but now with the newtonian limit of gravity .
this approximation would not be as reasonable in the qcd case for a quark propagator since gluon - gluon couplings are comparable to gluon - quark couplings , but there is evidence from other calculations of the appearance of anomalous dimensions for infrared propagators .
( for a review see @xcite .
very recent calculations can also be found in references @xcite . )
perhaps the simplest way to think of this is to look at eq .
( [ gia15 ] ) and regard the ( singular ) double integral as the limit of two paths which must be taken as approaching each other arbitrarily closely .
this is an alternative to the differential regularization we used in eq.([gia23 ] ) and physically corresponds to two electron worldlines interacting via the exchange of an arbitary number of photons between them and at all points of their paths . in the limit of the paths coinciding ,
this becomes self - interaction , with the electron continuously exchanging photons with itself along its worldline . for gravity ( at least in the newtonian limit )
we want to replace the repulsive electrostatic self - interaction , say @xmath61 , with the attractive gravitational self - interaction , say @xmath62 , suggesting an asymptotic form of the dirac propagator exponent @xmath63 if @xmath64 , which is the adm@xcite mass of charged shell of charge @xmath65 , regularized by its own gravity , then one recovers an effectively free propagator .
since one generally makes measurements using the electromagnetic interaction , and the suggestion that quantum mechanics might be linked to self - interaction@xcite , it is interesting to consider what this might imply for the role of gravity in the quantum measurement problem@xcite . in particular , since one has a connection between mass and charge which is non - perturbative in newton s @xmath66 , and implies a mass near the planck mass @xmath67 gm , which may be thought to be in the neighborhood of a putative classical - quantum boundary .
intuitively , electromagnetic interactions roughen the path of an electron , making it `` spread out more '' , which is natural since the electromagnetic force between two identical charges ( or an electron and itself ) is repulsive .
gravitational interactions , on the other hand , make it `` spread out less '' , since the gravitational force between two identical objects ( here an electron and itself ) is attractive .
the notion of a path somehow `` spreading out '' can be made precise with the notion of fractal dimensions .
the fractal nature of particle paths has been discussed in the literature ( see , for example @xcite ) .
abbott and wise@xcite , working within nonrelativistic quantum mechanics , found 2 as the dimension of a quantum mechanical path , as opposed to 1 for a classical path .
cannata and ferrari@xcite extended this work for spin-1/2 particles and find different results not only in the classical and quantum mechanical limits , but also in the non - relativistic and relativistic limits .
intuitively one can understand the dimension 2 result for the nonrelativistic quantum mechanical case by thinking of the schrdinger equation as a diffusion equation in imaginary time@xcite . for diffusion one
has the distance @xmath68 a particle covers in time @xmath69 satisfying a relationship of the form @xmath70 wherein @xmath71 is the fractal dimension of the `` path '' .
for example , @xmath72 in the diffusion limit , and @xmath73 in the ballistic ( simple path ) limit@xcite .
here we have a closely analogous situation but with a 4-dimensional hamiltonian @xmath74 and with fractal diffusion in proper time , and as was showed in @xcite , one has @xmath75 all previous discussions of fractal propagators in quantum field theory and quantum mechanics have , to the best of our knowledge , ignored the effects of self - interaction via long - range fields .
the fact that self - interaction can qualitatively change the nature of a propagating particle even when coupled to a weak ( small coupling constant ) abelian gauge field naturally leads one to wonder whether or not coupling to a strong ( large coupling constant ) non - abelian gauge field might have significant physical effects .
we turn now to a discussion of possible implications of this sort of phenomenon for an understanding of confinement in qcd .
in this section we consider how propagators with anomalous dimensions could shed light on the confinement mechanism in qcd .
precise and rigorous calculations are beyond the scope of this paper , but we do indicate approaches to such calculations and what one would expect qualitatively .
quantum chromodynamics is an incredibly difficult theory in which to say anything precise .
the fundamental degrees of freedom carry strong ( colour ) charges , interact via highly nonlinear interactions , and ( by hypothesis ) , are not even observable asymptotically .
perturbative treatments make qcd look very much like qed , and involve propagators which look pretty much like those of qed with the main difference being in the nonabelian nature of the gauge fields and their associated self - couplings .
the question which we now raise is whether or not one can imagine a situation whereby anomalous exponents in the infrared will give rise to such strong qualitative changes in how propagators behave that the originally postulated degrees of freedom will simply not propagate ( at least at low energies ) that is , that some form of confinement appears .
what could confinement mean in terms of an anomalous dimension ?
[ pi1 ] we had a @xmath76 which drove @xmath71 to zero , one would have a zero - dimensional path , which is no path at all !
this would correspond to a confined particle .
let us see how far we can argue that such a phenomenon would occur .
one could start from eq.([gia15 ] ) with an appropriate form of @xmath77 for qcd valid for any separation .
unfortunately , we do not have an exact expression for @xmath77 in the infrared limit and in fact
every expectation is that ( even aside from colour indices ) it is very different from the qed one presumably having a confining term , roughly linear in separation .
of course this would be _ assuming _ confinement .
a simpler approach is to imagine that @xmath78 where @xmath79 and @xmath80 .
if the strong coupling constant @xmath81 is large enough then one could get @xmath82 .
let us follow this line of argument more closely .
the first thing to check is that the energy of self - interaction of a quark with itself is indeed of the same sign as that due to gravity ( and opposite to that due to electromagnetism ) .
while the interaction energy between a quark and an antiquark in the singlet state separated by a distance @xmath68 is given perturbatively by @xmath83 , the corresponding quark - quark expression in the triplet is @xmath84 . in other words , it is of the correct sign to lead to confinement .
now let us revisit eq.([gia15 ] ) and regard the ( singular ) double integral as the limit of two paths which must be taken as approaching each other arbitrarily closely as we did in section [ sec : gravity ] .
we are now looking at two paths of a red , say , quark , exchanging an arbitrary number of gluons .
if we now argue as we did for gravity , we have @xmath85 .
it is well - known@xcite ( and can be shown on very general grounds from dispersion relations ) that @xmath81 must increase without bound as @xmath68 goes to infinity ( or squared momentum transfer @xmath86 goes to zero ) .
happily here one only needs @xmath81 to reach a value making @xmath87 . at the corresponding value of @xmath86 the dimension of a quark path goes to zero and
the quark is confined .
lower values @xmath86 ( and negative dimensions ) make no sense since the integrals in eq .
( [ gia15 ] ) now have no paths any more as soon as @xmath71 hits zero .
note that this approach to thinking about confinement does not require any singular ( i.e. infinite ) values of @xmath86 ( or length ) but should happen at some well - defined value of @xmath86 ( or length ) presumably related to the confinement scale .
a rigorous calculation along these lines would be difficult , but a few points can be made in defense of this overall picture .
first of all , why would one imagine in the regularized view of eq .
( [ gia15 ] ) that one was always looking at a red quark interacting with another red quark ? surely each emitted gluon changes the colour of the particle and complicates matters .
this is indeed true , but if one has defined the quark as being red ( in some gauge ) then _ on average _ , even with various colours of gluons being continuously emitted , one expects that it would remain _ on average _ red .
any tendency to be green or blue would be a violation of @xmath88 symmetry since it would indicate a preferred direction in colour space other than `` red '' ( the effective value of the colour charge would be excpected to be scale - dependent , but this is exactly what a running coupling constant is meant to describe . )
this sort of argument could be the basis for a more rigourous mean field theory approach .
second @xmath81 is usually studied as a function of @xmath86 , but its form is not even known at long distance scales and thus a rather difficult ( ! ) function to work out in coordinate space to use in eq.([gia15 ] ) .
the point we want to make here is that it _ is _ very generally known that @xmath14 increases in the infrared limit , and this alone is enough to argue for confinement . even without an explicitly known confining form for @xmath89 ,
all one needs to know is that @xmath14 continues to rise .
at some point it will rise to a value high enough to drive @xmath71 to zero . beyond that , the quark is non - propagating ( our version of `` confined '' ) and there is no need to evalute paths in eq.([gia15 ] ) since the quark has none !
the expected results from asymptotic freedom are , of course , reproduced , since as @xmath81 goes to zero the propagator goes over to its free form ( aside from whatever corrections remain due to charge and mass as discussed earlier
particles coupled to long - range fields never really get free ! ) . a similar argument
would be expected to hold for any coloured objects , including gluons and hypothetical objects in other representations of @xmath88 colour .
the physical picture is an interesting one and rather different from the assumption that quarks are bound due to strong attractive forces to other quarks which increase linearly with separation in some approximation . here , below some energy scale (
that at which the dimension @xmath71 goes to zero ) , coloured objects interact so much with their own glue that they get `` stuck '' in the sense that their would - be paths are reduced to dimension zero . at higher energy scales they partially escape , with a dimension which rises as the effective coupling strength drops until they become , to a good approximation , free .
it is interesting to note that the view of confinement suggested here applies to any coloured object with no need for it to have neighbours present to `` bind '' to it and ensure that a colour singlet state propagates here interaction of a single coloured object with its own colour field is enough to confine it .
we have reviewed the simple and intuitive path integral description of how the propagator for a charged dirac particle is modified by soft self - energy radiative corrections as shown in reference @xcite .
the result is a self - similar ( fractal ) object with the non - locality one would expect for a particle carrying an infinite range field .
arguments are made for a similar , but qualitatively different , effect due to attractive self - interactions such as gravity and a calculation made for newtonian gravity .
the results are linked to the fractal dimensions of the paths that particles take in quantum field theory , and the effects of repulsive ( qed ) and attractive ( gravity ) self - interactions are discussed .
finally an attempt is made to estimate what effects would be expected in qcd , with a link made to confinement at a finite energy scale in terms of a fractal dimension which goes to zero .
we would like to thank the organizers of the irqcd 2006 conference for a wonderful and stimulating conference where one of us ( j.s . ) presented an earlier version of this work .
we would also like to thank reinhard alkofer for pointing out references @xcite and @xcite to us , and christian fischer for pointing out @xcite .
this work was supported in part by the national science foundation under grant nsf-0457001 .
kibble , _ j. math
* 9 * , 315 ( 1968 ) ; _ phys .
rev . _ * 173 * , 1527 ( 1968 ) ; _ phys .
rev . _ * 174 * , 1882 ( 1968 ) , v. chung , _ phys .
rev . _ * 140 * , 1110 ( 1965 ) , d. zwanziger , _ phys .
rev . _ * d7 * , 1082 ( 1973 ) ; _ phys .
lett . _ * 30 * , 934 ( 1973 ) ; _ phys .
_ * d11 * 3481 and 3504 , l. fadeev and p. kulish , _ teor . mat .
_ * 4 * , 153 ( 1970 ) ; _ theor .
phys . _ * 4 * , 745 ( 1970 ) .
a. a. abrikosov , _ sov .
. jetp _ * 3 * , 71 ( 1956 ) , a. v. suidzuiski _ sov .
jetp _ * 4 * , 179 ( 1957 ) , l. d. soloviev , _ sov .
* 17 * , 209 ( 1963 ) , d. v. shirkov , _
il nuovo cim . _ * 3 * , 845 ( 1956 ) , l. d. soloviev , _ sov .
dokl . _ * 1 * , 536 ( 1957 ) . | we show that qed radiative corrections change the propagator of a charged dirac particle so that it acquires a fractional anomalous exponent connected with the fine structure constant .
the result is a nonlocal object which represents a particle with a roughened trajectory whose fractal dimension can be calculated .
this represents a significant shift from the traditional wigner notions of asymptotic states with sharp well - defined masses .
non - abelian long - range fields are more difficult to handle , but we are able to calculate the effects due to newtonian gravitational corrections .
we suggest a new approach to confinement in qcd based on a particle trajectory acquiring a fractal dimension which goes to zero in the infrared as a consequence of self - interaction , representing a particle which , in the infrared limit , can not propagate .
pacs numbers : 12.20.-m,03.70.+k , 12.38.aw , 12.38.lg,05.45.df keyword : qed , theory of quantized fields , qcd , confinement , nonperturbative methods , fractals |
a geyser can be defined as a periodically eruptive spring resulting from earth s thermal activity .
one of the most famous geyser is the old faithful [ fig .
[ fig : geyser](a ) ] in the yellowstone national park in the united states .
it erupts about every 90 minutes and projects hot water up to 56 m high@xcite .
probably the first scientist to describe the principle of a geyser was robert bunsen in 1845 after the danish government sponsored him to study the eruption of mount hekla in iceland @xcite .
he lowered a thermometer into geyser depths and discovered that they had at their base a reservoir of superheated water .
he understood that when the water in the reservoir boils , the liquid in the vent is ejected upwards and produces an eruption of hot water and steam .
bunsen and his student tyndall built a toy geyser based on the same principle as illustrated in fig .
[ fig : geyser](b ) @xcite .
such a demonstrative experiment showed eruptions regularly spaced in time that confirmed their intuitions on the fundamental principles underlying natural geysers .
the realization of a geyser at reduced scale contributed to promote the bunsen s theory for geyser among the geophysical community @xcite .
nowadays , similar set - ups are present in various science museum like the exploratorium in san francisco ( usa ) and source - o - rama in chaudfontaine ( be ) and are also currently used for research purpose in order to understand mechanisms underlying a geyser eruption @xcite . in nature
, the dynamics of a geyser can be much more complex time series that the one of a toy geyser as revealed by _ in situ _ data collection in yellowstone usa @xcite or at el tatio in chile @xcite .
such a complexity can be attributed to a high sensitivity of geysers dynamics towards water and steam inflow @xcite , barometric pressure , tidal forces , tectonic stresses @xcite or reservoirs interconnection @xcite . over the past decades ,
crucial steps in the theoretical modelling of geyser eruptions have been achieved @xcite but an extensive description of the problem is still to uncover . throughout this paper
we aim at studying and predicting the dynamics of a toy geyser and understanding its similarities with natural geysers .
first , we reproduced a model geyser similar to the one of tyndall .
the erupting dynamics of such a geyser has been inspected as a function of the injected heat power and its geometrical characteristics .
the different stages of a geyser life and the evolution of the reservoir temperature over time are described in section [ sec : experiments ] .
afterwards , we develop in section [ sec : model ] a theoretical model based on thermodynamic notions and fluid dynamics in order to rationalize our observations . besides , we observed that the geyser effect can vanish because of intense vaporization or thermal convection in the vent .
the required conditions to produce a geyser effect are inspected in section [ sec : discussion ] and compared with conditions occurring in nature .
finally , we showed that the complex dynamics of natural geysers can be approached by interconnecting two reservoirs to a single vent as detailed in section [ sec : two_reservoirs ] .
\(a ) + of water at a temperature @xmath0 was connected to a pool by a long vent of radius @xmath1 .
the water in the pool was maintained at a constant temperature @xmath2 .
the system was completely filled with a column of water of height @xmath3 chosen between 1.6 and 11.5 m and the pool has a depth @xmath4 .
a heating plate provided a specific heat power @xmath5 to the water of the reservoir.,title="fig:",height=188 ] \(b ) + of water at a temperature @xmath0 was connected to a pool by a long vent of radius @xmath1 .
the water in the pool was maintained at a constant temperature @xmath2 .
the system was completely filled with a column of water of height @xmath3 chosen between 1.6 and 11.5 m and the pool has a depth @xmath4 .
a heating plate provided a specific heat power @xmath5 to the water of the reservoir.,title="fig:",height=188 ] \(c ) + of water at a temperature @xmath0 was connected to a pool by a long vent of radius @xmath1 .
the water in the pool was maintained at a constant temperature @xmath2 .
the system was completely filled with a column of water of height @xmath3 chosen between 1.6 and 11.5 m and the pool has a depth @xmath4 .
a heating plate provided a specific heat power @xmath5 to the water of the reservoir.,title="fig:",height=188 ]
a toy geyser was realized with a glass reservoir connected to a pool through a long and narrow vertical vent as sketched in fig .
[ fig : geyser](c ) .
the system was completely filled with water .
the total height of the liquid column from the bottom of the reservoir to the liquid surface is denoted @xmath3 .
the depth of the pool was @xmath4 and the vent radius @xmath1 .
the reservoir , which contained a mass of water @xmath6 , was heated from below by a hot plate maintained at a constant temperature .
thus , the water of the reservoir received a specific heat power @xmath5 .
when heated continually , the toy geyser undergoes periodic erupting events spaced between quiet phases .
this periodic behavior is illustrated by the evolution of the temperature @xmath0 measured in the middle of the reservoir as shown in fig .
[ fig : t_t_r](a ) for a geyser height @xmath7 m and a specific heating power @xmath8 w / kg . after a first heating phase
, the temperature in the reservoir follows a saw - tooth evolution between about 93@xmath9c and 102@xmath9c . a single period of the temperature of the reservoir
is reported typically in fig .
[ fig : steps ] and consists of several steps .
first , the water in the reservoir is heated up to the boiling point @xmath10 at a pressure corresponding to the water column of height @xmath3 [ stage ( a ) in fig .
[ fig : steps ] ] .
the boiling temperature of water as a function of the height of the overlying column is recalled in fig [ fig : t_t_r](b ) and is equal to 102@xmath9c for @xmath7 m. when the water in the reservoir reaches the corresponding boiling point , an intense production of vapor bubbles occurs .
buoyancy induces the upward motion of vapor bubbles in the vent which entrains the water above and produces an eruption of hot water and steam at the top of the toy geyser [ stage ( b ) in fig .
[ fig : steps ] ] .
after the ejection of the liquid in the vent , the water in the reservoir keeps on boiling and the vapor continues to escape throughout the vent toward the pool [ stage ( c ) in fig .
[ fig : steps ] ] .
the total amount of water transferred from the reservoir to the pool at this stage is denoted @xmath11 .
as the boiling goes on , the vapor production cools down the water remaining in the reservoir . when this latter reaches the boiling temperature @xmath12 corresponding to the pressure of a water column of height @xmath4 , the vapor emission decreases till the vapor flow can not sustain the water in the pool which finally falls back into the reservoir [ stage ( d ) in fig .
[ fig : steps ] ] . at this moment
, the eruption phase is finished .
the eruption has last a time @xmath13 .
the temperature in the reservoir reaches @xmath14 , which is the equilibrium temperature resulting from the mixture of a mass of water @xmath15 at the temperature @xmath16 and a mass of water @xmath11 coming from the pool maintained at a constant temperature @xmath2 [ stage ( e ) in fig .
[ fig : steps ] ] .
after that , the temperature in the reservoir increases from @xmath14 to @xmath10 in a time @xmath17 before an other eruption starts .
meanwhile , the temperature of the pool is thermalized to @xmath18c throughout the duration of the experiment .
\(a ) + in the reservoir over time for a toy geyser of height @xmath7 m , specific heating power @xmath8 w / kg , pool height @xmath19 cm and vent radius @xmath20 mm .
dots indicate the time of eruptions .
( b ) boiling temperature @xmath21 of water as a function of the height @xmath3 of the overlying column extracted from @xcite .
symbols show the parameters of the experiments presented in this study.,title="fig:",height=188 ] \(b ) + in the reservoir over time for a toy geyser of height @xmath7 m , specific heating power @xmath8 w / kg , pool height @xmath19 cm and vent radius @xmath20 mm .
dots indicate the time of eruptions .
( b ) boiling temperature @xmath21 of water as a function of the height @xmath3 of the overlying column extracted from @xcite .
symbols show the parameters of the experiments presented in this study.,title="fig:",height=188 ] ; ( b ) water boiling in the reservoir provokes an eruption ; ( c ) vapor emission throughout the vent ; ( d ) end of the eruption , the water from the pool refills the reservoir ; ( e ) the system returns to its initial state .
the typical evolution of the reservoir temperature @xmath0 over time is shown by the way of the red solid line in the graphic at the top .
the pool temperature is maintained at @xmath18c .
bottom and top pictures correspond to the reservoir and the pool at different times of the geyser cycle.,height=529 ] the two characteristics time @xmath17 , @xmath13 and the exchanged mass @xmath11 were measured as a function of the parameters of the experiments : the specific heating power @xmath5 and the geyser height @xmath3 .
the specific heating power received by the water in the reservoir is estimated during the first heating step throughout the relation @xmath22 where @xmath23 is the specific thermal capacity of water ( @xmath24 ) .
one remarks that @xmath5 takes into account the heat transfer lost by the reservoir toward its environment . in our experiments
the specific heating power could be varied between 30 and 220 w / kg . also , we inspected two different geyser heights in this study , @xmath7 m and @xmath25 m , for which the boiling temperatures of water are respectively @xmath26 and @xmath27 for a standard outside atmospheric pressure .
the mean values of @xmath17 , @xmath13 and @xmath11 were determined over more than 10 events and the experimental results are shown in fig .
[ fig : experimental_results ] and [ fig : experimental_results_dmm ] .
\(a ) + as a function of the specific heat power @xmath5 for different heights of the water column @xmath3 .
the blue dots and purple squares correspond to @xmath28 m and @xmath29 m. the inset presents the data with logarithmic scales .
solid lines show the predictions of eq .
( [ eq : geyser_period ] ) .
( b ) mean erupting time @xmath13 as a function of the specific heat power @xmath5 for different heights of the water column @xmath3 .
the blue dots and purple squares correspond to @xmath28 m and @xmath29 m. solid lines show the predictions of eq .
( [ eq : geyser_tau_e]).,title="fig:",width=264 ] \(b ) + as a function of the specific heat power @xmath5 for different heights of the water column @xmath3 .
the blue dots and purple squares correspond to @xmath28 m and @xmath29 m. the inset presents the data with logarithmic scales .
solid lines show the predictions of eq .
( [ eq : geyser_period ] ) .
( b ) mean erupting time @xmath13 as a function of the specific heat power @xmath5 for different heights of the water column @xmath3 .
the blue dots and purple squares correspond to @xmath28 m and @xmath29 m. solid lines show the predictions of eq .
( [ eq : geyser_tau_e]).,title="fig:",width=264 ] of the reservoir exchanged with the pool during an eruption as a function of the specific heating power @xmath5 for different water column heights @xmath3 .
the blue dots and purple squares correspond respectively to @xmath7 m and @xmath25 m. solid lines show the predictions of eq .
( [ eq : mass_exchanged]).,width=264 ] one observes in fig .
[ fig : experimental_results](a ) that the elapsed time @xmath17 between two eruptions lasts from 2 to 30 min in our experiments .
also , this characteristic time increases with the geyser height @xmath3 and decreases with the specific heating power @xmath5 . the logarithmic scale diagram in the inset of fig .
[ fig : experimental_results](a ) reveals that the time @xmath17 between eruptions is inversely proportional to @xmath5 .
besides , the erupting time @xmath13 is between one and two minutes in the present set of experiments reported in fig [ fig : experimental_results](b ) .
this characteristic time increases with the geyser height @xmath3 and shows a small dependency with the specific heating power @xmath5 .
one notices in fig .
[ fig : experimental_results_dmm ] that the exchanged mass fraction @xmath30 increases with the specific heating power and the water column height .
the duration between two successive geyser eruptions is approached by estimating the time needed to heat the water in the reservoir from the initial temperature @xmath14 to the boiling temperature @xmath31 corresponding to a water column of height @xmath3 [ fig .
[ fig : t_t_r](b ) ] .
if a specific heat power @xmath5 is transferred to the reservoir , its temperature increases from @xmath14 to @xmath31 in a time @xmath32 \label{eq : initial_temperature}\ ] ] during an eruption , a mass of water @xmath11 moves from the reservoir to the pool .
the eruption stops when the temperature of the water remaining in the reservoir is close to the boiling temperature @xmath16 corresponding to the pressure imposed by a water column of height @xmath3 and vaporization weakens . at this moment , the exchanged mass @xmath11 moves from the pool to the reservoir and fills it completely .
thus , the temperature @xmath14 of the reservoir at this particular time is @xmath33 combining eqs .
( [ eq : initial_temperature ] ) and ( [ eq : mix_temperature ] ) yields @xmath34 \label{eq : geyser_period}\ ] ] one observes that eq .
( [ eq : geyser_period ] ) gives two contributions for the characteristic time between geyser eruptions .
the first contribution corresponds to the cool down of the reservoir by the entry of a mass @xmath11 of water coming from the pool maintained at a temperature @xmath2 .
the second contribution is the time required to reach the boiling temperature @xmath31 below a water column of height @xmath3 .
these contributions imply that the geyser height @xmath3 , the pool temperature @xmath2 and the geometry of the plumbing system impact the time between eruptions . in our experiments ,
all quantities involved in eq .
( [ eq : geyser_period ] ) are known except @xmath30 which has been measured , thus an estimation of the time between eruptions @xmath17 can be deduced .
such predictions are plotted by the way of solid lines in fig .
[ fig : experimental_results](a ) for the two geyser heights inspected experimentally ( @xmath35 and @xmath36 ) .
the good agreement between experiments and predictions validate the approach considered above . after the beginning of an eruption ,
the temperature of the reservoir is @xmath31 whereas the overlying pressure is close to the atmospheric one and imposes an equilibrium temperature @xmath16 .
the erupting time is associated with the time needed to cool down the reservoir from @xmath31 to @xmath16 by vaporizing water [ stage ( c ) in fig .
[ fig : steps ] ] .
in our situation , we assume that this process is limited by the thermal transfer occurring in the reservoir with bubbles formation on nucleation sites , their growth and their ascension .
such a thermal transfer depends on the difference between the temperature of the water in the reservoir and the boiling temperature corresponding to the imposed pressure . to our knowledge , such a phenomenon has never been described quantitatively .
thus , we assume empirically a power law dependency for the power rate lost by the superheated water , _
i.e. _ @xmath37 . in first approximation ,
the power lost by the vaporization is balanced by the temperature variation of the reservoir @xmath38 .
solving the previous equation from a reservoir temperature @xmath31 to a threshold temperature @xmath39 , for which an eruption stops , provides the following prediction for the time of eruptions @xmath40 \label{eq : geyser_tau_e}\ ] ] one notices that if @xmath41 , the erupting time predicted by eq .
( [ eq : geyser_tau_e ] ) increases with the geyser height as observed experimentally in fig .
[ fig : experimental_results](b ) .
more quantitatively , the previous model can be adjusted on experimental data .
solid lines in fig .
[ fig : experimental_results](b ) correspond to the prediction of eq .
( [ eq : geyser_tau_e ] ) for @xmath42 , @xmath43 si and @xmath44 .
the deviation of experimental data from the theoretical predictions at large heating power ( @xmath45 w / kg ) can be attributed to the omission of the energy injected by the heating plate in the energy balance of the model .
indeed , the specific heating power @xmath5 delays the cooling of the water in the reservoir by vaporization .
+ when an eruption has occurred and water has left the vent , the vaporization in the reservoir results from both the injected heat power @xmath46 and the heat flux @xmath47 coming from the water superheat .
thus , the power balance in such a situation provides : @xmath48 where @xmath49 is the mass flow rate of vapor produced in the reservoir and @xmath50 the latent heat of water .
the vaporization generates bubbles which grow and rise in the reservoir . due to the conical shape of the reservoir , the bubble volume fraction @xmath51 increases along the vertical axis @xmath52 as sketched in fig .
[ fig : sketch_bubbles ] .
when the volume fraction reaches a critical value @xmath53 , the bubbles coalesce and finally form the vapor phase . considering that the critical bubble volume fraction corresponds to the fraction for which a single bubble can not be added to a cubic close - packed arrangement of monodispered bubbles
, we estimate that @xmath54 .
thereby , the equilibrium position @xmath55 of the liquid / vapor interface during an eruption can be deduced from the critical bubble volume fraction in the liquid . introducing the bubbles upwards velocity @xmath56 in the reservoir ,
the vapor density @xmath57 and @xmath58 is the radius of the reservoir at the altitude @xmath52 relatively to the bottom of the reservoir , we get : @xmath59 .
haberman and morton showed that the terminal velocity of vapor bubbles in water is between 20 cm / s and 30 cm / s for bubble of radius comprised between 0.5 mm and 15 mm @xcite .
for the sake of simplicity , we consider that all the bubbles in the vent rise at the same velocity @xmath60 cm / s , a fact which is consistent with our observations .
the critical bubble volume fraction @xmath53 is reached for a reservoir radius @xmath61 we deduce from eq .
( [ eq : mass_exchanged ] ) and geometrical considerations , the mass of water @xmath11 ejected out of the reservoir during an eruption for the parameters corresponding to the experiments of section [ sec : experiments ] ( @xmath62 kg , @xmath63 , @xmath64
j / kg , @xmath65 and @xmath66 si ) and a critical bubble volume fraction @xmath54 .
the prediction of the theory is show in fig .
[ fig : experimental_results_dmm ] by the way of solid lines .
one observes the relative agreement between the theory and experimental data . .
all the vapor bubbles have the same upwards velocity @xmath67 . due to the reduction of the reservoir radius @xmath58 with increasing height @xmath52 , the bubble volume fraction @xmath51 increases along this direction up to a critical volume fraction @xmath53 which defines the equilibrium position @xmath55 of the liquid / gas interface.,width=132 ]
the aim of this section is to inspect under which conditions a toy geyser can lead to periodic eruptions .
first , we inspect the physical origin of toy geyser eruptions , and second we consider the domain of existence of three non - periodic regimes : the fumarole regime , the hot springs regime and the boiling pool regime .
first , we investigate the effect of the vent radius on the toy geyser .
experimentally , we observed that if the vent radius is too large ( @xmath68 mm ) , the vapor bubbles produced by the boiling in the reservoir are able to rise without entraining the surrounding water and no eruptions occur .
thus , the birth of a geyser eruption is caused by the transition from a dispersed flow regime to an annular flow regime in the vent .
the regimes of a two - phase flow in a vertical pipe have been studied extensively by mishima _
the transition between a dispersed and an annular flow in the vent can be approached with a critical air volumic fraction above which bubbles clog and entrain the surrounding water . in the present situation ,
the vapor volumic fraction in the vent @xmath69 can be estimated by assuming that when the reservoir reaches its boiling temperature @xmath31 , all the specific heating power @xmath5 is used to vaporize water .
the balance between the vapor departure through the vent and the vapor production in the reservoir yields @xmath70 using the critical vapor volume fraction @xmath71 introduced in section [ sec : exch_mass ] , we predict the critical vent radius @xmath72 below which eruptions occur @xmath73 for @xmath74 , @xmath75 , @xmath76 , @xmath63 , @xmath77 and @xmath78 , we get @xmath79 .
the vent radius used in the experiments presented in section [ sec : experiments ] ( @xmath80 ) is smaller than @xmath72 which explains why periodic eruptions were observed . in order to verify the previous criteria
, we built a model geyser with the same reservoir and the same height but a different vent radius @xmath81 and no eruptions have been observed , consistently with our approach .
such a regime is called the hot spring regime because only tiny and hot vapor bubbles reach the pool . beyond the case of a toy geyser , bubble clogging in a vertical constriction
is a phenomenon observed in a large variety of situations such as volcanic eruptions @xcite , cold co@xmath82 geysers @xcite , mentos in diet coke @xcite and beer taping @xcite . in the particular case of volcanic eruptions , the elevation of the internal pressure in the volcano and thus the explosiveness of the eruption depend on possibility for gas to escape .
depending on the volatile content of the magma , its viscosity and the configuration of the magmatic chamber , different types of eruption occur from hawaiian to plinian eruptions @xcite . for large specific heating power @xmath5 , we observed a first eruption followed by a continuous vaporization of the water in the reservoir .
in such a case , the liquid in the pool is unable to go down into the reservoir . because of permanent emission of vapor through the vent , this situation
is called the fumaroleregime .
besides , we have noticed that a toy geyser working in a fumarole regime can be brought back to a periodic regime by increasing the height @xmath4 of the water in the pool . the minimal pool height which leads to a periodic regime
can be discussed quantitatively . during an eruption , when the water in the reservoir has reached its equilibrium temperature @xmath16 , the specific heating power @xmath5 only induces water vaporization .
thus , the mass of water vapor produced per unit time is @xmath83 and imposes a mean gas velocity in the vent @xmath84 . if the dynamic pressure @xmath85 associated to the vapor flow at the exit of the vent is larger than the hydrostatic pressure @xmath86 ( where @xmath87 is the water density ) at the bottom of the pool , the water will not re - entry in the vent and the system will stay in a fumarole regime .
the critical specific heating power @xmath88 above which the system does not follows periodic eruptions is @xmath89 for @xmath90 , @xmath91 , @xmath76 , @xmath63 , @xmath92 and @xmath93 , this relation gives a critical value of @xmath94 . in the experiments presented in section [ sec : experiments ] ,
the maximal specific heat power is @xmath74 which explains why a geyser regime was always observed . in order to test the validity of eq .
( [ eq : power_threshold_fumarole ] ) , we decrease the height @xmath4 of the water in the pool till the transition towards a fumarole regime occur . for @xmath74
, we found a critical height of water @xmath95 mm which is consistent with the previous law .
in the description of the toy geyser done previously , we have considered that the heat power injected in the system is entirely transferred to the reservoir .
actually , thermal convection can transfer a part of the injected heat power from the reservoir to the vent and to the pool .
the thermal energy per unit time conveyed in the vent can be expressed as @xmath96 with @xmath97 the heat transfert coefficient in the case of vertical natural convection in a vertical pipe heated from below .
this coefficient depends on the rayleigh number and the grashof number but stays between 100 and 1000 @xmath98 in the range of parameter encountered in our experiments @xcite .
the limit case where the convective flux is equal to the injected heat power @xmath99 occurs for a critical vent radius @xmath100 for @xmath74 , @xmath75 , @xmath101 and @xmath102 we get @xmath103 .
this approach explains why thermal convection in the vent has been neglected in the experiments presented in section [ sec : experiments ] for which @xmath90 .
indeed , in such a situation the convective flux is about 200 times smaller than the injected heat power .
finally , one expects that for a vent radius larger than @xmath104 , the injected heat power will be transferred to the entire volume of water and boiling will occur everywhere .
this regime of the toy geyser is called the boiling springregime in the following of this paper .
the three transitions discussed previously between the different regimes of a toy geyser can be gathered on a single diagram which is presented in fig . [
fig : phase_diagramme ] .
such a diagram shows the behavior of a toy geyser depending on the injected heat power @xmath99 and the vent radius @xmath1 . if the specific heating power is larger than the critical one @xmath88 corresponding to eq .
( [ eq : power_threshold_fumarole ] ) , the set - up exhibits a fumarole regime .
differently , if the vent radius is between the two critical radii @xmath72 and @xmath104 predicted by eq .
( [ eq : radius_clogging ] ) and ( [ eq : radius_convection ] ) , the system behaves as a hot spring . finally , if @xmath105 than the water in the reservoir convects and the toy geyser adopts a boiling spring regime .
in between these regimes , lies the regime corresponding to geyser periodic eruptions which required @xmath106 and @xmath107 . also , we report in fig . [
fig : phase_diagramme ] the range of experimental parameters used for the experiments detailed in section [ sec : experiments ] and section [ sec : fumarole ] where a geyser regime and a hot spring regime have been reported .
( c ) as a function of the vent radius @xmath1 and the injected heat power @xmath99 .
the white solid line represents transition between a fumarole and a geyser regime as predicted by eq .
( [ eq : power_threshold_fumarole ] ) .
the dashed line indicates the limit between the geyser and the hot spring regimes according eq .
( [ eq : radius_clogging ] ) .
the dotted line corresponds to the transition between a hot spring and a boiling spring regime as predicted by eq .
( [ eq : radius_convection ] ) .
the white vertical stripe indicates the range of parameter inspected experimentally in section [ sec : experiments ] where a geyser regime has been reported . on the contrary ,
the dark stripe shows the range of parameter inspected experimentally in section [ sec : fumarole ] and for which a hot spring regime has been observed . also , the estimations of the mean heating power and minimal radius of the constriction for the old faithful are reported in the diagram by the way of a diamond.,width=453 ] as underlined in the introduction , most of natural geysers adopt a complex dynamics rather than a precise frequency of eruption .
such behavior may be attributed to a change in external parameters such as the water inflow , the atmospheric temperature and pressure , tidal forces and tectonic stress @xcite or be the result of the internal mechanism of geysers .
indeed , the interconnection between the plumbing systems of different geysers is known to induce a chaotic and complex dynamics of eruptions @xcite . despite the numerous differences that distinguish toy geysers from natural ones , it is interesting to observe where natural geysers are located in the previous phase diagram .
for this purpose , we collected in the literature estimations of the minimal radius of constriction and the mean injected heat power of the old faithful in yellowstone @xcite and report them in fig .
[ fig : phase_diagramme ] .
one notices that the parameters corresponding to this natural geyser is in the domain where a toy geyser adopts a periodic regime .
we hope that future characterizations of geothermal systems will allow to compare more rigorously this theoretical phase diagram with experimental facts .
such a procedure can be a way to determine if natural geysers work on the same principle than toy geysers .
a way to approach the complexity of natural geysers is to build a set - up with two interconnected reservoirs .
both reservoirs received a different specific heating power and they are connected to a single pool as sketched in fig .
[ fig : setup_2 ] .
the connection between the vents of the two reservoirs is made at a fraction @xmath108 of the total height @xmath3 of the water column . and
the two vents are connected at a height @xmath110 proportional to the total height @xmath3 of the system.,width=226 ] figure [ fig : double_exp](a ) shows the evolution over time of the temperature in each reservoir of an interconnected toy geyser .
one notices that the saw - tooth evolution of the temperature in a reservoir is not periodic anymore differently from the case of a geyser with a single reservoir inspected in section [ sec : experiments ] .
the lost of regularity is underlined in fig .
[ fig : double_exp](b ) where the times between eruptions for each reservoir and for the total system are presented as a function of time .
the aim of this section is to describe and rationalize the dynamics of toy geyser with two coupled reservoirs .
the total height of the geyser is @xmath7 m and @xmath111 .
the specific heat power injected in the left and right reservoirs are respectively @xmath112 w / kg and @xmath113 w / kg .
( a ) measured temperatures as a function of time in each reservoir .
symbols indicate the time of eruptions .
( b ) times @xmath17 between eruptions over time for each reservoir ( red dots and blue square ) and for the global system ( gray diamonds).,title="fig:",width=264 ] \(b ) + .
the total height of the geyser is @xmath7 m and @xmath111 .
the specific heat power injected in the left and right reservoirs are respectively @xmath112 w / kg and @xmath113 w / kg .
( a ) measured temperatures as a function of time in each reservoir .
symbols indicate the time of eruptions .
( b ) times @xmath17 between eruptions over time for each reservoir ( red dots and blue square ) and for the global system ( gray diamonds).,title="fig:",width=264 ] a different way to present the data of fig .
[ fig : double_exp](b ) is to plot the time between eruption of an event numbered @xmath114 as a function of the previous time between eruption ( numbered @xmath115 ) .
such a procedure allows to visualize the correlation between successive eruptions and is done in fig .
[ fig : double_exp2 ] for three different coupling parameters @xmath108 .
the figure shows the correlation for each reservoir ( red dots and blue squares for left and right reservoirs respectively ) and for the total system ( gray diamonds ) .
one observes in fig .
[ fig : double_exp2](a ) that even for small coupling parameter @xmath111 , the time between eruptions in each reservoir undergoes a large dispersion ( about two minutes of dispersion while the mean period is about six minutes ) .
besides , the data for the whole system ( gray diamonds ) occupy a trapezoidal shaped domain . when the coupling parameter increases to @xmath116 [ fig . [
fig : double_exp2](b ) ] , the dispersion for each reservoir increases and the domain of the global eruption adopt a banana shape . finally , for large coupling parameter @xmath117 , fig .
[ fig : double_exp](c ) indicates a reduction of the dispersion of the time between eruptions in each reservoir and for the whole system .
\(a ) + and the previous event numbered @xmath115 .
red dots and blue squares indicate the time between eruption in each reservoir whereas gray dots represent the time between eruptions for the total system .
( a ) @xmath111 .
( b ) @xmath116 .
( b ) @xmath117 .
the dashed rectangles show the predictions for the dispersion for the time between eruptions based on the model introduced in section [ sec : model_double].,title="fig:",width=188 ] \(b ) + and the previous event numbered @xmath115 .
red dots and blue squares indicate the time between eruption in each reservoir whereas gray dots represent the time between eruptions for the total system .
( a ) @xmath111 .
( b ) @xmath116 .
( b ) @xmath117 .
the dashed rectangles show the predictions for the dispersion for the time between eruptions based on the model introduced in section [ sec : model_double].,title="fig:",width=188 ] \(c ) + and the previous event numbered @xmath115 .
red dots and blue squares indicate the time between eruption in each reservoir whereas gray dots represent the time between eruptions for the total system .
( a ) @xmath111 .
( b ) @xmath116 .
( b ) @xmath117 .
the dashed rectangles show the predictions for the dispersion for the time between eruptions based on the model introduced in section [ sec : model_double].,title="fig:",width=188 ] in order to rationalize the previous observations , we develop a theoretical modeling of an interconnected geyser .
the model is based on the idea that when a reservoir reaches its boiling temperature and erupts , it empties partially the vent and reduces the height of the water column for the other reservoir .
thus , depending on the temperature of the second reservoir , the eruption of the first reservoir may entrain the second one . in the model , we assume that the temperature of each reservoir follows a perfect saw - tooth evolution which is governed successively by the equations : @xmath118 and @xmath119 in order to stay in the range of temperature @xmath120 .
we consider that the evolution of the temperature in both reservoirs are independent unless the temperature of one reservoir is larger than a temperature threshold @xmath121 when the other erupts .
in such case , the eruption of a reservoir provokes the eruption of the other one .
this approach provides a prediction of times between eruptions for each reservoir .
figure [ fig : double_model ] shows the correlation plot of the times between eruptions for the three values of the coupling parameters inspected experimentally in paragraph [ sec : double_exp ] .
\(a ) + and the previous event denoted @xmath115 .
( a ) @xmath122 .
( b ) @xmath123 .
( b ) @xmath124.,title="fig:",width=188 ] \(b ) + and the previous event denoted @xmath115 .
( a ) @xmath122 .
( b ) @xmath123 .
( b ) @xmath124.,title="fig:",width=188 ] \(c ) + and the previous event denoted @xmath115 .
( a ) @xmath122 .
( b ) @xmath123 .
( b ) @xmath124.,title="fig:",width=188 ] one notices in fig .
[ fig : double_model ] that despite the large spreading of the experimental data is not reproduced by the model , the general trends are similar to the ones observed in section [ sec : double_exp ] .
indeed , the times between eruptions of the total system occupy a trapezoidal domain which vanishes at large coupling parameters @xmath108
. moreover , the spreading of the times between eruptions for each reservoir increases with @xmath108 up to a critical value where the two reservoirs act as a single one . besides
, the dispersion observed in fig .
[ fig : double_exp2 ] for @xmath17 can be approached by the theory developed in section [ sec : model ] for a toy geyser with a single reservoir . indeed , eq . ( [ eq : geyser_period ] ) and ( [ eq : geyser_tau_e ] ) yield the time between two eruptions of a reservoir if the other one does not experience any eruption . the substitution in eq .
( [ eq : geyser_period ] ) and ( [ eq : geyser_tau_e ] ) of the boiling temperature @xmath10 by its value @xmath125 when the other reservoir has erupted , provides the minimal time between eruptions that a reservoir can follow .
the difference between these two extreme cases provides an estimation of the dispersion observed for @xmath17 .
we delimited in fig .
[ fig : double_exp2 ] the domain in which the time between eruptions is supposed to be confined according this approach .
the discrepancy observed between the experimental dispersions and the theoretical ones may come from the variability of boiling processes .
the study of a toy geyser reveals the successive operations which lead to periodic eruptions of steam and hot water . in a geyser ,
a geometrical constriction prevents thermal convection of water from a heated reservoir toward a temperated pool .
thus , the water in the reservoir slowly accumulates heat energy by being superheated relatively to the pool level .
such a situation holds till the water in the reservoir boils and the vaporization ejects the water in the vent , a phenomena which generates a large release of energy in a short period of time .
the slow accumulation of energy and its fast release is characteristic of catastrophic natural events such as earthquakes , volcanic eruptions or thunderbolts .
we showed with the example of a toy geyser that the conditions to get such an accumulation of thermal energy and a periodic release are strict .
this may explain why geysers are rare objects among other geothermal sources such as fumaroles , hot springs and boiling springs .
moreover , toy models of geyser allow us to understand their sensitivity towards external parameters such as water inflow , barometric pressure or pool temperature .
however , a major difference exists between toy geysers and natural ones : the latter generally adopt a complex dynamics which strongly differs from the clock - like precision of toy geysers .
the complexity of real geysers has been approached with a toy geyser having two interconnected reservoirs .
such a set - up showed a complex time behavior which has been studied statistically and rationalized theoretically .
the rareness of faithful geysers in nature reveals that very few of them have a single and independent reservoir .
interconnected and complex plumbing systems seem to prevail in nature leaving room for the complexity rather than simplicity .
besides , the present study lets several open questions such as the deep understanding of the phenomenon of bubble clogging , the geyser operation for large vent radius or the exact description of heat transfer during a boiling process in a superheated liquid .
also , this work opens new perspectives of research such as geysers made with a liquid differing from water or with a liquid ables to let solid deposition in order to study the stability of the plumbing systems over time .
this research has been funded by the inter - university attraction poles programme ( iap 7/38 micromast ) initiated by the belgian science policy office .
sd thanks the fnrs for financial supports .
mb acknowledges support as a fnrs - fria fellow .
we are grateful to alice dubus who realised the glass reservoirs we dreamed about .
tadd truscott and zhao pan are thanked for their fascinations concerning this subject and the excellent idea of interconnecting two reservoirs .
the authors acknowledge herv caps , christophe clanet , laurent maquet , julien schokmel , alexis duchesne for eruption of ideas and precious comments concerning this work . c. munoz - saez , m. manga , s. hurwitz , m. l. rudolph , a. namiki , and c .- y
dynamics within geyser conduits , and sensitivity to environmental perturbations : insights from a periodic geyser in the el tatio geyser field , atacama desert , chile . , 292:4155 , 2015 .
han , m. lu , b.j .
mcpherson , e.h .
keating , j. moore , e. park , z.t .
watson , and n .- h . jung .
characteristics of co2-driven cold - water geyser , crystal geyser in utah : experimental observation and mechanism analyses . , 13(3):283297 , 2013 . | a geyser can be reproduced by a toy experiment composed of a water pool located above a water reservoir , the two being connected by a long and narrow tube . when the bottom reservoir is heated , the system experiences periodic eruptions of hot water and steam at the top similarly to the geyser effect occurring in nature .
the eruption frequency of a toy geyser is inspected experimentally as a function of the heating power and the height of the set - up .
we propose a thermodynamic model of this system which predicts the time between eruptions .
besides , the geometric and thermodynamic conditions required to observe a geyser effect are discussed . the study of the toy geyser
is then extended to the case of two reservoirs connected to the same tube .
in such a configuration , an eruption in a reservoir may entrain the eruption of the other .
such a coupled system adopts a complex time - evolution which reflects the dynamics of natural geysers .
we analyze the behavior of a toy geyser with two reservoirs by the way of statistical tools and develop a theoretical model in order to rationalize our observations . |
the tw hya association , a loose grouping of very young stars that does not appear to be related to any nearby molecular cloud material , has been the subject of considerable attention in recent years .
the prototype of the association , tw hya itself , was the first recognized example of an isolated t tauri star .
the first few members of the association were discovered more or less by accident , and were only later recognized to be related . aside from these , the systematic search for additional stars belonging to the group has relied on their x - ray properties ( such as their detection as a rosat source ) , or their similar kinematics ( proper motions ) .
follow - up studies have been made in many cases to establish the youth of the candidate members , for example by the presence of strong lithium @xmath06708 absorption , h@xmath1 emission , infrared excess , etc . to date
there are 19 objects recognized as true members in the published literature ( see , e.g. , kastner et al .
1997 ; webb et al.1999 ; sterzik et al .
1999 ; zuckerman et al . 2001 ) , along with many other potential members that need to be studied further .
one of the goals of the present work is to investigate the kinematics of the candidate members , but in the direction to previous studies that have focussed only on proper motions .
we present measurements along the line of sight ( radial velocities ) , which are complementary to the proper motions and help define the space motion of these objects .
another goal of this project is to detect any spectroscopic binaries that might be present through repeated measurements of the radial velocities of our targets , and to determine their orbits .
our observations were obtained using nearly identical echelle spectrographs on the mmt prior to conversion ( mt . hopkins , az ) , the 1.5 m tillinghast reflector at the f. l. whipple observatory ( also on mt .
hopkins , az ) , and on the 1.5 m wyeth reflector at the oak ridge observatory ( harvard , ma ) .
a single echelle order was recorded with a photon - counting reticon diode array , at a resolving power of @xmath2 and a central wavelength of 5187 .
the spectral coverage is 45 .
we report here on the results based on some 300 spectra , and an additional 150 archival spectra from the cfa echelle database .
table 1 presents the sample of stars studied in this program , including for completeness hd 98800 ( twa-4 ) , which was observed several years ago with the same instrumental setup and was discovered to be a quadruple system ( torres et al .
the second column in the table gives the official " designation within the association , and the last column collects the available measurements of the li @xmath06708 equivalent width for these stars . -5pt
rllcccc@ c & & & r.a . & dec .
& @xmath3 & & li @xmath06708 eq . + & twa # & other name & ( j2000 ) & ( j2000 ) & ( mag ) & spt & width ( ) + 1 & & hip 48273 & 09:50:30.1 & @xmath404:20:37 & 6.24 & f6 & ... + 2 & & tyc 6604 - 0118 - 1 & 09:59:08.4 & @xmath522:39:35 & 10.09 & k2 & 0.143 + 3 & & hip 50796 & 10:22:18.0 & @xmath510:32:15 & 10.80 & ... & ... + 4 & & hip 53486 & 10:56:31.0 & @xmath407:23:19 & 7.37 & k0 & ... + 5 & & rxj1100.0@xmath53813 & 11:00:02.4 & @xmath538:13:20 & 12.29 & ... & ... + 6 & & hd 95490 & 11:00:51.2 & @xmath535:33:38 & 8.79 & f7 & 0.15 + 7 & twa-1 & tw hya & 11:01:52.0 & @xmath534:42:16 & 10.92 & k8e & 0.43 + 8 & twa-2a & cd@xmath68887a & 11:09:13.9 & @xmath530:01:39 & 11.07 & m2e & 0.52 + 9 & & rxj1109.7@xmath53907 & 11:09:40.1 & @xmath539:06:48 & 10.58 & g3 & 0.190 + 10 & twa-3a & hen 3 - 600a & 11:10:28.0 & @xmath537:31:53 & 12.04 & m4e & 0.57 + 11 & & hd 97131 & 11:10:34.2 & @xmath530:27:19 & 9.01 & f2 & 0.03 + 12 & & cd@xmath77097 & 11:12:42.7 & @xmath538:31:04 & 10.24 & f5 & ... + 13 & & rxj1115.1@xmath53233 & 11:15:06.9 & @xmath532:32:46 & 12.36 & ... & ... + 14 & twa-12 & rxj1121.1@xmath53845 & 11:21:05.5 & @xmath538:45:17 & 12.85 & m2 & 0.530 + 15 & twa-13a & rxj1121.3@xmath53447n & 11:21:17.3 & @xmath534:46:47 & 11.46 & m2e & 0.650 + 16 & twa-13b & rxj1121.3@xmath53447s & 11:21:17.5 & @xmath534:46:51 & 12.00 & m1e & 0.51 + 17 & twa-4a & hd 98800a & 11:22:05.3 & @xmath524:46:40 & 9.41 & k5 & 0.425 + 18 & twa-4b & hd 98800b & 11:22:05.3 & @xmath524:46:39 & 9.94 & ... & [email protected] + 19 & twa-5a & cd@xmath87795a & 11:31:55.4 & @xmath534:36:27 & 11.37 & m3 & 0.55 + 20 & twa-9a & cd@xmath97429a & 11:48:24.3 & @xmath537:28:49 & 11.26 & k5 & 0.46 + all recognized members are seen to have strong li in absorption , while some of the other objects ( proposed by a number of authors as candidate members ) have much weaker li .
not all of the potential members have estimates of this crucial youth indicator , so their status is unclear .
radial velocities were obtained using standard cross - correlation techniques ( the xcsao task running under iraf ) , as described , e.g. , by latham ( 1992 ) . the templates were taken from an extensive library of synthetic spectra based on the latest model atmospheres by kurucz ( morse & kurucz , in preparation ) . in addition , we derived effective temperatures ( @xmath10 ) and projected rotational velocities ( @xmath11 ) for all our stars by comparison with the synthetic spectra .
a number of objects have turned out to be double - lined . for these we used the two - dimensional cross - correlation algorithm known as todcor ( zucker & mazeh 1994 ) , and in most cases we were able to derive the @xmath10 and @xmath12 for both components .
table 2 lists our results for the temperatures ( k ) and rotational velocities ( km s@xmath13 ) . a comparison with @xmath10 values derived from the spectral types ( adopting the calibration by de jager & nieuwenhuijzen 1987 ) and other sources shows generally good agreement .
the agreement for the cooler stars is poorer due to limitations in the model atmospheres used to compute the synthetic spectra , which do not include several key molecular opacity sources .
our measures of @xmath12 are also fairly consistent with other determinations .
rll@ c@ c@ c@c@c@c & & & & t@xmath14 & t@xmath14 & t@xmath14 & @xmath12 & @xmath12 + & twa # & other names & spt & ( cfa ) & ( spt ) & ( other ) & ( cfa ) & ( other ) + 1 & & hip 48273 & f6 & 6300/6150 & 6530 & & 22/22 & + 2 & & tyc 6604 - 0118 - 1 & k2 & 5050/4850 & 4840 & & 19/15 & 19 + 3 & & hip 50796 & ... & 4750 & ... & & 8 & + 4 & & hip 53486 & k0 & 5050 & 5150 & & 1 : & + 5 & & rxj1100.0@xmath53813 & ... & 5750 & ... & & 35 & + 6 & & hd 95490 & f7 & 6500 & 6430 & & 13 & + 7 & twa-1 & tw hya & k8e & 4150 & 4150 & 4150 & 4 & 10 , 15 , 14 + 8 & twa-2a & cd@xmath68887a & m2e & 4050 : & 3690 & & 13 & 15 + 9 & & rxj1109.7@xmath53907 & g3 & 5900 & 5710 & 5800 & 27 & 23 + 10 & twa-3a & hen 3 - 600a & m4e & 4750 : & 3350 & & 20 : & 15 + 11 & & hd 97131 & f2 & 6750/6750 & 7180 & & 12/16 & + 12 & & cd@xmath77097 & f5 & 6250 & 6650 & & 14 & + 13 & & rxj1115.1@xmath53233 & ... & 5750/5250 : & ... & & 25/0 : & + 14 & twa-12 & rxj1121.1@xmath53845 & m2 & 4000 : & 3520 & 3600 & 15 & 21 + 15 & twa-13a & rxj1121.3@xmath53447n & m2e & 4150 : & 3520 & 3600 & 12 & 16 , 10 + 16 & twa-13b & rxj1121.3@xmath53447s & m1e & 4100 : & 3660 & 3800 & 12 & 16 , 12 + 17 & twa-4a & hd 98800a & k5 & & 4410 & 4350 & & 5 + 18 & twa-4b & hd 98800b & ... & & ... & 4250/3700 & & 3/2 + 19 & twa-5a & cd@xmath87795a & m3 & 4050 : & 3490 & & 36 : & 58 + 20 & twa-9a & cd@xmath97429a & k5 & 4350 & 4410 & & 11 & + in addition to the well known quadruple system hd 98800 , multiple measurements of the radial velocity of our other targets have revealed several binaries ( some double - lined ) with short orbital periods ( rxj1115.1@xmath53233 , tyc 6604 - 0118 - 1 , hip 48273 , rxj1100.0@xmath53813 ) and one double - lined triple system ( hd 97131 ) .
none of these appear to be true members of the tw hya association , based on their li abundance or kinematics ( radial velocity of the center of mass ) .
details on these orbital solutions will be reported in a forthcoming paper . in table 3
we list the mean heliocentric radial velocities of all our targets . for the multiple systems we give the center - of - mass velocity .
our measurements are consistent with other determinations from the literature , but much more precise . except for the case of hen 3 - 600a ( twa-3a ) and cd@xmath5337795a ( twa-5a ) , which are double - lined or possibly triple - lined but have not yet had their orbits determined ( see also webb et al .
1999 ) , the other recognized members of the association have similar radial velocities . in the case of hd 98800
we have assumed that the average velocity of components a and b corresponds to the center of mass of the quadruple system .
similarly for rxj1121.3@xmath53447 ( twa-13 ) , which is a visual binary .
the mean velocity of twa-1 , twa-2 , twa-4 , twa-9 , twa-12 , and twa-13 is @xmath15 km s@xmath13 .
rllcc@c@l & & & & mean @xmath16 & mean @xmath16 & + & twa # & other names & n@xmath17 & ( cfa ) & ( other ) & remarks + 1 & & hip 48273 & 112 & @xmath416.249 @xmath18 0.071 & & binary ( orbit ) + 2 & & tyc 6604 - 0118 - 1 & 34 & @xmath426.96 @xmath18 0.24 & & binary ( orbit ) + 3 & & hip 50796 & 2 & @xmath413.1 @xmath18 1.0 & & + 4 & & hip 53486 & 6 & @xmath45.54 @xmath18 0.29 & & + 5 & & rxj1100.0@xmath53813 & 18 & @xmath42.03 @xmath18 0.92 & & binary ( orbit ) + 6 & & hd 95490 & 4 & @xmath57.51 @xmath18 0.17 & & + 7 & twa-1 & tw hya & 5 & @xmath412.92 @xmath18 0.23 & @xmath413.5 @xmath18 1.5 & + 8 & twa-2a & cd@xmath529@xmath198887a & 6 & @xmath411.20 @xmath18 0.32 & @xmath20 & + 9 & & rxj1109.7@xmath53907 & 7 & @xmath51.2 @xmath18 1.1 : & @xmath52.0 @xmath18 1.1 & var ?
+ 10 & twa-3a & hen 3 - 600a & 14 & @xmath46.9 @xmath18 1.7 : & @xmath21 & double - lined ?
+ 11 & & hd 97131 & 37 & @xmath527.10 @xmath18 0.26 & & triple ( orbit ) + 12 & & cd@xmath77097 & 3 & @xmath58.19 @xmath18 0.26 & & + 13 & & rxj1115.1@xmath53233 & 11 & @xmath510.0 @xmath18 1.1 & & binary ( orbit ) + 14 & twa-12 & rxj1121.1@xmath53845 & 2 & @xmath412.23 @xmath18 0.60 & @xmath410.9 @xmath18 1.1 & + 15 & twa-13a & rxj1121.3@xmath53447n & 4 & @xmath411.72 @xmath18 0.61 & @xmath410.5 @xmath18 1.2 & + 16 & twa-13b & rxj1121.3@xmath53447s & 4 & @xmath412.41 @xmath18 0.48 & @xmath412.0 @xmath18 1.2 & + 17 & twa-4a & hd 98800a & 152 & @xmath412.75 @xmath18 0.10 & & binary ( orbit ) + 18 & twa-4b & hd 98800b & 152 & @xmath45.73 @xmath18 0.14 & & binary ( orbit ) + 19 & twa-5a & cd@xmath533@xmath197795a & 26 & @xmath46.9 @xmath18 2.0 : & @xmath21 & double - lined ? + 20 & twa-9a & cd@xmath536@xmath197429a & 10 & @xmath410.17 @xmath18 0.36 & & + though it is tempting to assign this mean radial velocity to the group as a whole , and to then use it as a criterion to accept or reject other candidate members , reality is more complex and various kinematical studies have shown that there is probably a gradient in the radial velocity across the large sky area covered by this association ( several tens of degrees ) .
one of such studies , by makarov & fabricius ( 2001 ) , has modeled the kinematics of the association as a moving group using a variant of the convergent - point method .
the authors used proper motions from the tycho-2 catalogue along with hipparcos parallaxes for the few members which have them , and searched an area of more than 3000 deg@xmath22 for additional members with the same motion that are also x - ray sources from the rosat bright source catalog . for each star they computed a
kinematical " distance as well as the predicted radial velocity . comparing the latter with the few measurements available to the authors from the literature for the classical members , they found it necessary to include an expansion term in their model , which not only is not unexpected for this group , but also gives a dynamical age of 8.3 myr , in good agreement with other estimates from pre - main sequence evolutionary tracks ( @xmath23 myr ) .
the internal velocity dispersion they derived is 0.8 km s@xmath13 , and the depth of the group is significant ( tens of parsecs ) .
makarov & fabricius ( 2001 ) produced a list of 23 additional candidates , of which we have observed several ( see table 4 ) .
our radial velocity measurements allow us to test these objects for membership .
the known members of the association ( designation in column 2 ) have radial velocities ( @xmath24 ) very close to the predicted values ( @xmath25 ) in most cases , but two of the other stars ( hip 48273 and tyc 6604 - 0118 - 1 ) , which happen to be double - lined binaries , do not . on the other hand
, the velocity for hip 50796 agrees perfectly with the prediction .
hip 53486 is an especially interesting case because it the nearest candidate member ( 17 pc ) , and its predicted distance ( @xmath26 ) agrees with the value measured by hipparcos ( @xmath27 ) .
it has the lowest expected radial velocity ( @xmath28 km s@xmath13 ) , and we do indeed measure a low value ( @xmath29 km s@xmath13 ) .
further evidence that some of these new candidates may be true members is given by the fact that , unbeknownst to makarov & fabricius , one of their stars ( hip 57524 , too faint for us and not shown in table 4 ) is actually twa-19 , which is li - rich .
follow - up observations of all these candidate are necessary to establish their youth , and our colleagues at this workshop are encouraged to do so .
l@ c@ c@ c@ c@ c@ cc@ c & & r.a . & dec .
& @xmath30 & @xmath27 & @xmath26 & @xmath25 & @xmath24 + name & twa # & ( 2000 ) & ( 2000 ) & ( mas / yr ) & ( pc ) & ( pc ) & ( km s@xmath13 ) & ( km s@xmath13 ) + hip 53911 & 1 & 11:01:51.9 & @xmath534:42:17 & 75.4 & 56.4 & 57.1 & @xmath412.7 & @xmath412.9 + tyc 7201 - 0027 - 1 & 2 & 11:09:13.8 & @xmath530:01:40 & 92.6 & & 47.1 & @xmath410.6 & @xmath411.2 + hip 55505 & 4 & 11:22:05.3 &
@xmath524:46:40 & 96.8 & 46.7 & 45.7 & @xmath49.1 & @xmath49.2 + tyc 7223 - 0275 - 1 & 5 & 11:31:55.3 & @xmath534:36:28 & 86.7 & & 50.9 & @xmath410.0 & @xmath46.9 : + hip 57589 & 9 & 11:48:24.2 & @xmath537:28:49 & 158.1 & 50.3 & 76.3 & @xmath412.6 & @xmath410.2 + hip 48273 & & 09:50:30.1 & @xmath404:20:37 & 157.9 & 45.9 & 26.6 & @xmath410.7 & @xmath416.2 + tyc 6604 - 0118 - 1 & & 09:59:08.4 & @xmath522:39:35 & 163.7 & & 62.8 & @xmath416.5 & @xmath427.0 + hip 50796 & & 10:22:18.0 & @xmath510:32:16 & 179.1 & 34.0 & 53.8 & @xmath413.1 & @xmath413.1 + hip 53486 & & 10:56:30.8 & @xmath407:23:18 & 268.1 & 17.6 & 16.7 & @xmath43.7 & @xmath45.5 + | we present our spectroscopic measurements of the radial velocity , effective temperature , and projected rotational velocity of several of the known members of the tw hya association , as well as measurements for candidate members selected on the basis of their x - ray or kinematic properties .
a number of our targets turn out to be binaries , but most are non - members . the radial velocities for some of the other candidates support the conclusion that they are kinematically associated with the group , although further observations are required to show that they are indeed pre - main sequence objects .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
quantum physics allows us a perfect randomness , so most of all quantum information - theoretic primitives try to offer an unconditional security under the randomness . for examples , quantum key distribution protocols such as bb84 @xcite and b92
@xcite highly depend on a random measurements for given classified non - orthogonal quantum states . instead of the random measurement on non - orthogonal states
, we can consider a direct randomization of quantum states through a quantum channel .
this randomizing procedures are efficiently accomplished via the private quantum channels ( pqc ) _ or _ quantum one - time pads @xcite . in the paper we are interest to some schemes for _ approximate _ encryptions ( no perfect ) and we make an attempt to reducing some classical communication resources .
we would like to call the randomizing procedures or maps as random unitary channels ( ruc ) in terms of quantum channels .
there are several methods for the approximate randomizing quantum states , for examples , @xcite : we here adapt the procedure of hayden et al .
@xcite .
many applications of ruc in quantum protocols ( see e.g. , @xcite . ) are started from the approximate version of pqc . here
we will propose new approximate quantum _ state _ sharing ( aqss ) scheme , which uses two approximate pqcs ( apqc ) and reduces the classical pre - shared secrets about one - half as compared with a perfect protocol .
actually our protocol could be including the ( well - known ) quantum _ secret _ sharing protocols @xcite , because a quantum state itself is able to operate special quantum tasks , though those are impossible in the classical power .
imagine that if there is a quantum computer only activated under a bipartite quantum state ( or _ quantum key _ ) , then our aqss protocol will give a efficient and secure solution for the quantum key .
these approximate quantum state sharing protocols may offer us more opportunities as compared with the quantum secret sharing .
let s take account of the pre - shared secrets for the approximate quantum state sharing protocols under ruc - based pqc roughly .
assume that a sender charlie prepares a quantum state @xmath1 ( two - qudit ) and transmits the state through two independent rucs , then two distant agents alice and bob will receive some output state of including high entropy .
for the state @xmath1 the perfect randomization protocol will require exactly the amount of @xmath2-unitary matrices ( pauli matrices ) . on the other hand , the construction of hayden et al .
@xcite for our aqss scheme implies that only @xmath3-unitaries sufficient . in other words
, the perfect quantum state sharing protocol needs to @xmath4 bits of pre - shared secret information , while the aqss protocol demands about @xmath5 bits of information .
note that the works in @xcite will give a similar result for @xmath5 bits bound .
we will prove the information - theoretic security of the aqss scheme in two kinds of eavesdropping : an interior and exterior attackers .
the proof of having higher entropy condition for the exterior attacks is not easy fact , so we split the input state @xmath1 to separable and entangled cases . as a result
, the von neumann entropy in both cases can be chosen sufficiently larger , and a leakage information will be arbitrarily small . finally the authors show that our bipartite aqss scheme naturally can be generalized to an one - sender and multiparty - receivers schemes . in section [ sec : ruc ] we introduce the definition of random unitary channels , and briefly mention about special property known as the destruction of quantum states on a product random unitary channel .
we present our aqss protocol based on two approximate pqcs in section [ sec : aqss ] , and investigate the security of aqss of considering two attacks : an exterior and interior strategies .
we finally conclude our results in section [ sec : concl ] .
now let us define the random unitary channel , and then construct an approximate private quantum channels .
for all density matrices @xmath6 , a completely positive trace - preserving map @xmath7 is the so - called _ @xmath0-randomizing _ , if @xmath8 where the trace norm is defined by @xmath9 .
this definition directly induces the notion of random unitary channels .
that is , for every @xmath10 , a quantum channel @xmath7 is called the
_ random unitary channel _ , if @xmath11 is @xmath0-randomizing , where the unitary operators @xmath12 , and the probability @xmath13 s are all positives with @xmath14 . ( the notation @xmath15 denotes the set of bounded linear operators from @xmath16 to itself and @xmath17 the unitary group on @xmath16 . ) note that the parameter @xmath18 is the number of kraus operation elements for ruc , so it corresponds to the dimension of arbitrary environment . for the approximate constructions of ruc
, it was known that for all @xmath19 there exist random unitary channels in sufficiently larger dimension @xmath20 , such that @xmath18 can be taken to be @xmath21 in @xcite and @xmath22 in @xcite where @xmath23 s are chosen randomly according to the haar measure .
we here fix the number @xmath18 of having exactly @xmath24 , the theorem 1 in @xcite .
as mentioned in the introduction , most intuitive application of the random unitary channel is the _ approximate _ private quantum channel @xcite , which is a modification of the perfect private quantum channel @xcite via ruc .
the ruc - based apqc is the main tool of constructing the proposed aqss protocol .
approximate private quantum channel : alice applies some @xmath23 s and bob decodes @xmath25 with @xmath26 of having pre - shared @xmath27 bits classical information . ]
the security of pqc is preserved by the argument of the accessible information in which the leakage information is less than @xmath0 .
although small information is leaked to exterior attackers , bob s decoding state is almost equal to alice s original state @xmath10 . the fig .
[ fig : pqc ] describes the total procedure of apqc . in the next section we use two one - way independent pqcs between a sender charlie and a receiver alice , and the sender charlie and another receiver bob .
let s define two rucs , from the definition of ( eq . ( [ eq : ruc ] ) ) , such that @xmath28 where we fix the probability as an equally weighted probabilities @xmath29 and @xmath30 for all @xmath31 , and assume that the number of @xmath32 is equal to @xmath33 , i.e. , @xmath34 .
for an approximate state sharing of any bipartite quantum state , above two channels play an important role in the approximate quantum state sharing scheme . for given two rucs @xmath35 and @xmath36 , and for all input @xmath1 , we must bound the trace norm for the difference between an output state of the product channel @xmath37 and maximally mixed @xmath38 , such that @xmath39 where a security parameter @xmath0 be a positive less than 1 .
the relation above asserts that all encoding states are information - theoretically secure .
unfortunately , for any entangled states proving the bound is not a simple task .
note that the argument for the ( efficient ) randomization is related to a destruction of correlations in quantum states @xcite .
they pointed out that the unitary operations of the amount corresponding to the quantum mutual information @xmath40=s[\varphi_a]+s[\varphi_b]-s[\varphi_{ab}]$ ] efficiently destroy the total correlation of any quantum states @xcite , where @xmath41=-{\mbox{$\mathrm{tr}$}}\varrho\log\varrho$ ] the von neumann entropy . for the maximally
entangled state @xmath42 , @xmath40=2\log d$ ] , which might be related to the eq .
( [ eq:2logd ] ) .
the following section gives the aqss protocol and the security of the protocol .
the last of the section , we briefly describe a multiparty aqss scheme .
let us assume that charlie - alice and charlie - bob have independent two apqcs , and charlie wants to sharing a bipartite quantum state @xmath1 _ securely _ between alice and bob .
approximate qss : if charlie - alice and charlie - bob have shared two independent pqcs each other , then the product channel @xmath37 preserves the security with high probability for any attacks .
the arrow denotes that alice must go to bob s location to obtain the state . ]
the protocol for a bipartite quantum state sharing is simple ( see fig . [
fig : qss3 ] ) : * the sender charlie selects a quantum state @xmath1 and transmits the state through the channel @xmath37 to the receivers alice and bob . * distant two parties alice and bob
just hold the state @xmath43 they received .
* when alice and bob want to reveal the original state @xmath1 , they must cooperate in a single location .
they perform the inverse unitary operations under the locally shared keys .
the security of the aqss protocol is divided two cases of an exterior and interior attacks .
actually the security is based on information - theoretic assumption , which means that the intercepted states must have the higher von neumann entropy .
thus any attackers can not obtain sufficient information for the original states .
first , let us consider an attack accomplished by an exterior eve .
assume that eve intercepts the state @xmath43 .
we here claim that @xmath44\sim2\log d\to\infty\ ] ] as @xmath20 goes to infinity .
we do nt know the accurate description for the state @xmath43 for all inputs , so we will divide the state @xmath1 into the separable and entangled one and investigate the behavior each other . if product state is given , it is possible to infer the inequality eq .
( [ eq : main ] ) easily . by using the triangle inequality with respect to the trace norm for the two rucs , if @xmath45 and @xmath46 , then @xmath47 for @xmath48 .
more formally assume that @xmath49 such that @xmath14 , i.e. , a separable state is given , then @xmath50 \label{eq : triangle}\\ & \le&2{\varepsilon } , \nonumber\end{aligned}\ ] ] where the inequalities eq .
( [ eq : normconv ] ) and eq .
( [ eq : triangle ] ) come from the norm convexity and the triangle inequality , respectively @xcite .
thus any separable inputs for the product channel are very close to the maximally mixed state @xmath51 .
this implies that @xmath52 $ ] is close to @xmath53 . for the separable input cases
, there is another bound that depends on the dimension parameter @xmath20 and @xmath18 : we can prove that the expectation value for the difference between the channel output and the maximally mixed state ( with respect to the trace norm ) is very close , that is , @xmath54 where @xmath55 denotes the total expectation value of @xmath56 and @xmath57 for the independent rucs @xmath35 and @xmath36 , respectively .
the appendix in this paper states that the inequality eq .
( [ eq : exp2ruc ] ) is non - trivial and obtained precisely by exploiting the relation between the trace norm and the hilbert - schmidt norm . as mentioned above ,
let s take @xmath58 and @xmath59 , then @xmath60 this implies that eve s attack is impossible in principle .
what can we do for an entangled input state ?
though a direct proof could be impossible , there is an evidence for the statement , the eq .
( [ eq:2logd ] ) . the theorem iii.3 in @xcite states that , for a positive operator - valued measure ( povm ) @xmath61 which is implemented using local operation and classical communication ( locc ) , @xmath62 , where @xmath63 and @xmath64 with a maximally entangled state s.t . @xmath65 and @xmath66 .
natural extension is possible as adding the channel @xmath36 : define @xmath67 and @xmath68 , then also @xmath62 .
therefore , we can conclude the state @xmath43 is close to @xmath51 under the locc - implemented povm . in this reason
any input state @xmath1 through the product channel @xmath37 have high entropy for @xmath69 .
second , we must consider a situation when alice _ or _ bob is malicious .
assume that bob intercepts the alice s state @xmath70 , bob s decoded state looks like @xmath71 where @xmath72 denotes the inverse operation for bob s ruc @xmath36 , but @xmath73 $ ] has still high entropy values .
the intercepted state @xmath74 is still almost maximally mixed state by the definition of the ruc @xmath70 . as a result
, bob can not obtain any information for @xmath75 without charlie - alice s key information .
symmetrically alice s attack is useless .
in other words , the charlie s aim of sharing a quantum state @xmath1 between alice and bob will be securely accomplished .
at least above - mentioned two attacks ( exterior and interior eavesdropping ) can not break the security of the proposed aqss protocol .
so the cooperation between alice and bob always restores the original state approximately . in the proposed scenarios ,
the perfect protocol for quantum state sharing requires exactly @xmath76 unitary operators , while our protocol only needs to total @xmath77 unitaries for sufficiently larger @xmath20 .
this fact directly means that some pre - shared key bits are reduced by factor 2 , since the aqss is needed @xmath78 secret bits , but the perfect qss is required @xmath2 bits . for any state @xmath79 , and for any channel @xmath80 ( for an @xmath19 is arbitrary ) , let s consider a relation like that @xmath81 then , it is sufficient to construct the perfect qss ( @xmath82 ) with @xmath76 pauli operators for the channel @xmath80 in the sense of pqc @xcite . in the case of our approximate qss , the product channel of two rucs ( @xmath83 ) just consume of half secret bits , so we say that it is _ efficient _ in weak sense ( though small information is always leaking ) . without loss of generality , a direct extension of the bipartite quantum state sharing protocol ( eq . ( [ eq : exp2ruc ] ) )
gives the security of a multiparty approximate quantum state sharing ( maqss ) .
assume that a sender charlie ( @xmath84 ) prepares an @xmath85-qudit @xmath86 .
if they initially have shared pqcs between @xmath84-@xmath87 , @xmath84-@xmath88 and so on , then , for any @xmath19 , @xmath89 the above eq .
( [ eq : maqss ] ) implies that any exterior attacks will be failed .
furthermore all interior attacks ( including group conspiracy ) will be frustrated to obtain the whole state without others secrets , it has similar reason to the two receivers protocol . let s look at the cost of secret bits for the maqss scheme
. roughly speaking , the perfect scheme requires @xmath90 secret bits , but maqss only @xmath91-bits sufficient .
we studied that the approximate quantum state sharing schemes are efficient from the classical information cost of view and those are robust to the two kinds of attacks .
the proposed aqss protocol basically depends on an approximate private quantum channel , which is constructed via two independent random unitary channels .
although the protocol leaks small information corresponding to the security parameter @xmath0 , the scheme preserves its information - theoretic security , and so the aqss and maqss schemes can be interpreted as some high - efficiency state sharing protocols for any bipartite and multipartite quantum states .
for given two random unitary channels @xmath92 and @xmath93 in eq .
( [ eq:2ruc ] ) , and for all pure - separable states @xmath79 , @xmath94 where @xmath95 for any pure states @xmath1 .
( note that this method is just an extension of the statement , the chapter 3 in @xcite . )
recall that the unitary operators are chosen randomly according to the haar measure , and take expectation over the random selection of unitaries : @xmath96 & = & \frac{1}{n_an_b } \nonumber\\ & & + \frac{1}{n_a^2n_b^2}\sum_{i\neq k}\sum_{j\neq l } { { \mathbb{e}}}_{\{u_{i , j}\}}{\mbox{$\mathrm{tr}$}}\left(u_i\otimes u_j\varphi_{ab}u_i^\dagger\otimes u_j^\dagger\right ) \left(u_k\otimes u_l\varphi_{ab}u_k^\dagger\otimes u_l^\dagger\right ) \nonumber\\ & = & \frac{1}{n_an_b}+{\mbox{$\mathrm{tr}$}}\left[{{\mathbb{e}}}_{\{u_{i , j}\ } } \left(u_i\otimes u_j\varphi_{ab}u_i^\dagger\otimes u_j^\dagger\right ) { { \mathbb{e}}}_{\{u_{k , l}\ } } \left(u_k\otimes u_l\varphi_{ab}u_k^\dagger\otimes u_l^\dagger\right)\right ] \label{eq : exp1}\\ &
= & \frac{1}{n_an_b}+{\mbox{$\mathrm{tr}$}}\frac{{{\mathbbm{1}}}}{d^4 } \label{eq : exp2}\\ & = & \frac{1}{n_an_b}+\frac{1}{d^2}. \label{eq : exp3}\end{aligned}\ ] ] in eq .
( [ eq : exp1 ] ) , we have used that @xmath97 and @xmath98 are chosen independently , and eq .
( [ eq : exp2 ] ) inherited from the definition of the haar measure . ( for any @xmath6 , a haar - distributed set @xmath99 satisfies that @xmath100 . ) the eq .
( [ eq : exp2 ] ) exploits the separable condition for @xmath1 .
note that for any rank @xmath20 matrix @xmath101 @xmath102 . for any rank @xmath103 matrix @xmath101 , a generalization of the corollary a.2 in @xcite
directly show that @xmath104 1 c. h. bennett and g. brassard , _ quantum cryptography : public - key distribution and coin tossing _ , in _ proceedings of ieee international conference on computers , systems and signal processing _ , p. 175 , ( 1984 ) .
c. h. bennett , _ quantum cryptography using any two nonorthogonal states _ , phys .
lett . * 68 * , 3121 ( 1992 ) .
a. ambainis , m. mosca , a. tapp , and r. de wolf , _ private quantum channels _ , in _ ieee symposium on foundations of computer sciences ( focs ) _ , p. 547 , ( 2000 ) .
p. hayden , d. leung , p. w. shor , and a. winter , _ randomizing quantum states : constructions and applications _ ,
phys . * 250 * , 371 ( 2004 ) .
a. ambainis and a. smith , _ small pseudo - random families of matrices : derandomizing approximate quantum encryption _ , in _ proceedings of random_. p. 249
a. harrow , p. hayden , and d. leung , _ superdense coding of quantum states _ ,
lett . * 92 * , 187901 ( 2004 ) . c. h. bennett , p. hayden , d. leung , p. w. shor , and a. winter , _ remote preparation of quantum states _ ,
ieee trans .
theory * 51 * , 56 ( 2005 ) .
p. a. dickinson and a. nayak , _ approximate randomization of quantum states with fewer bits of key _ , aip conference proceedings * 864
* , 18 ( 2006 ) .
m. hillery , v. buek , and a. berthiaume , _ quantum secret sharing _ ,
a. * 59 * , 1829 ( 1999 ) .
a. karlsson , m. koashi , and n. imoto , _ quantum entanglement for secret sharing and secret splitting _ ,
a. * 59 * , 162 ( 1999 ) .
b. groisman , s. popescu , and a. winter , _ quantum , clasical , and total amount of correlations in a quantum state _ ,
a. * 72 * , 032317 ( 2005 ) .
g. aubrun , _ on almost randomizing channels with a short kraus decomposition _
* 288 * , 1103 ( 2009 ) . | we investigate the approximate quantum state sharing protocol based on random unitary channels , which is secure against any exterior or interior attackers in principle . although the protocol leaks small information for a security parameter @xmath0 , the scheme still preserves its information - theoretic secrecy , and reduces some pre - shared classical secret keys for a private quantum channel between a sender and two receivers . the approximate private quantum channels constructed via random unitary channels play a crucial role in the proposed quantum state sharing protocol |
small trapped fermi gases with contact or short - range interactions have attracted a great deal of attention recently @xcite . using lithium or potassium , for example
, equal - mass two - component systems can be realized experimentally by occupying two different hyperfine states . for typical experimental conditions , @xmath11-wave or higher partial wave interactions between two like atoms ( say , two spin - up atoms ) and between two unlike atoms ( a spin - up and a spin - down atom )
are negligibly small .
furthermore , by tuning an external magnetic field in the vicinity of a fano - feshbach resonance , the @xmath0-wave scattering length @xmath12 can be adjusted to essentially any value @xcite . in this paper
, we consider the regime where the @xmath0-wave scattering length is much larger than the range @xmath3 of the underlying two - body model potential . in the limit that @xmath3 goes to zero and
@xmath12 goes to infinity , the unitary regime is realized . in this regime , the only meaningful length scale of the system is given by the oscillator length @xmath13 that characterizes the external confining potential @xcite . throughout
, we assume a spherically symmetric harmonic potential with angular trapping frequency @xmath8 ( i.e. , @xmath14 with @xmath15 denoting the atom mass ) . from a theoretical point of view ,
small harmonically trapped fermi gases with central short - range interactions are particularly appealing since they can be treated with comparatively high accuracy by a variety of methods , including techniques that have been developed in the context of atomic physics , nuclear physics and quantum chemistry problems @xcite . for the harmonically trapped equal - mass system ,
the center of mass degrees of freedom separate .
furthermore , the relative orbital angular momentum quantum number @xmath5 , the projection quantum number @xmath16 and the parity @xmath6 are good quantum numbers .
this implies that the hilbert space can be divided into subspaces , which significantly reduces the complexity of the calculations compared to , for example , systems confined to move within a box with periodic boundary conditions @xcite .
the harmonically trapped fermi gas consisting of two spin - up and two spin - down atoms with vanishing angular momentum has been treated by a variety of techniques in the literature ( see refs .
@xcite for reviews ) .
the ground state energy and ground state properties of the @xmath1 system in the zero - range limit , for example , are by now well characterized @xcite .
much less , however , is known about the excitation spectrum @xcite , which contains both natural and unnatural parity states , i.e. , states with parity @xmath17 and @xmath18 , respectively .
while a good portion of the excitation spectrum of the @xmath1 system with natural parity has been determined throughout the crossover and at unitarity @xcite , little is known about the properties of states with unnatural parity . moreover ,
the energy spectra of the @xmath2 , @xmath9 and @xmath10 systems have not yet been characterized in detail .
this paper presents extensive benchmark results for the @xmath1 and @xmath2 energies of natural and unnatural parity states at unitary .
in addition , we present portions of the energy spectra of the @xmath9 and @xmath10 systems .
we then use the energy spectra of the @xmath1 and @xmath2 systems at unitarity to determine the fourth - order virial coefficient @xmath19 of the trapped system in the low - temperature regime .
the fourth - order virial coefficient enters into the virial equation of state , which allows for the determination of the universal thermodynamics of two - component fermi gases in the temperature regime down to about half the fermi temperature @xmath20 @xcite .
for the temperature regime in which we have convergence , i.e. , for @xmath21 , where @xmath22 denotes the temperature and @xmath23 boltzmann s constant , we find that the fourth - order virial coefficient @xmath19 of the trapped system is negative and decreases monotonically with increasing temperature . if we assume that @xmath19 continues to change monotonically with increasing temperature in the medium- and high - temperature regime , our results predict that the fourth - order virial coefficient of the trapped system and through application of the local density approximation ( lda)that of the homogeneous system approach a negative value in the high - temperature limit .
this is in contrast to recent results @xcite based on the equation of state , determined both experimentally and calculated via a diagrammatic monte carlo technique .
these studies predict that the fourth - order virial coefficient of the homogeneous system is positive .
the discrepancy would be resolved if the fourth - order virial coefficient of the trapped system was changing non - monotonically with temperature , allowing for a sign change of @xmath19 in the medium- or high - temperature regime .
analogous non - monotonic behavior was found for one of the third - order virial coefficients of the trapped unequal - mass two - component fermi gas at unitarity @xcite . while we do not have access to sufficiently large portions of the energy spectra of the @xmath1 and @xmath2 systems to determine the fourth - order virial coefficient of the trapped system in the medium- and high - temperature regimes ( thereby preventing us from drawing definite conclusions ) , our results illuminate a number of aspects related to the determination of the virial coefficients from few - body energy spectra .
section [ sec_system ] introduces the system under study , reviews the stochastic variational approach , and presents details regarding our implementation .
compact expressions for the relevant matrix elements for natural and unnatural parity states are presented in appendix [ appendix ] .
section [ sec_energies ] ( see also supplementary material @xcite ) summarizes our extrapolated zero - range energies for the @xmath1 , @xmath2 , @xmath9 and @xmath10 systems in tabular form and discusses their characteristics .
section [ sec_virial ] uses the @xmath1 and @xmath2 energies to determine the fourth - order virial coefficient of the trapped system at unitarity .
lastly , sec .
[ sec_summary ] concludes .
we consider a two - component fermi gas with @xmath24 spin - up and @xmath25 spin - down atoms of mass @xmath15 with @xmath26 .
we assume that the atoms are confined by a spherically symmetric trapping potential with angular frequency @xmath8 .
furthermore , we assume that the spin - up and spin - down atoms interact through a short - range interaction potential @xmath27 , where @xmath28 ( @xmath29 ) denotes the position vector of the @xmath30 atom measured relative to the center of the trap and @xmath31 , and that atoms with like spins do not interact .
the model hamiltonian @xmath32 then reads @xmath33 where @xmath34 throughout , we are interested in the regime where the @xmath0-wave scattering length @xmath12 of the interspecies interaction potential @xmath35 becomes infinitely large . for the @xmath36 and @xmath37 systems , semi
- analytical solutions are known if @xmath35 coincides with the zero - range @xmath38-function potential @xcite . for @xmath39 systems with @xmath40 , however , no such semi - analytical solutions are known . to determine the eigenenergies of @xmath39 systems with @xmath41 and @xmath42 , we separate off the center of mass motion and resort to a numerical technique , the stochastic variational approach @xcite . in this approach , it is convenient to model the interactions between the unlike atoms through a gaussian potential @xmath43 with depth @xmath44 ( @xmath45 ) and range @xmath3 @xcite , @xmath46.\end{aligned}\ ] ] to treat the unitary system , we adjust the depth @xmath47 of @xmath48 for a given @xmath3 such that the two - body system in free space supports one zero - energy @xmath0-wave bound state but no deep - lying bound states . to determine the zero - range energies , we consider a number of @xmath3 , @xmath49 , and extrapolate the finite - range energies to the @xmath4 limit ( see sec .
[ sec_energies ] for examples ) .
we take advantage of the fact that the hamiltonian @xmath32 separates into the center of mass hamiltonian @xmath50 and the relative hamiltonian @xmath51 , @xmath52 . in the following
, we consider the relative hamiltonian @xmath51 and use the stochastic variational approach to determine the eigenenergies and eigenstates of the schrdinger equation @xmath53 . here
, we explicitly indicate the dependence of the eigenenergies on @xmath24 and @xmath25 but , for notational simplicity , not that of the hamiltonian and the wave function . to compact the notation , we write @xmath51 as @xmath54 , where @xmath55 denotes the kinetic energy operator associated with the relative motion , and @xmath56 the contribution of the confining potential associated with the relative degrees of freedom .
the stochastic variational approach is a basis set expansion approach that writes the relative wave function @xmath57 of a given state in terms of a set of basis functions @xmath58
@xcite , @xmath59 here , the @xmath60 denote expansion coefficients and @xmath61 an anti - symmetrization operator that ensures that the wave function is anti - symmetric under the exchange of any pair of like fermions . in eq .
( [ eq_expand ] ) , @xmath62 denotes the number of basis functions . as with other basis set expansion approaches
, the ritz variational principle ensures that the lowest energy as well as the higher - lying energies obtained by the stochastic variational approach are rigorous upper bounds to the exact eigenenergies of the system @xcite . in the following ,
we introduce the basis functions used in this work , which have good orbital angular momentum @xmath5 , projection quantum number @xmath16 and parity @xmath6 ; here , @xmath5 , @xmath16 and @xmath6 are associated with the relative motion . following refs .
@xcite , we write the basis functions @xmath58 as a product of a correlated gaussian [ second line of eq .
( [ eq_basis1 ] ) ] and a `` prefactor '' [ first line of eq .
( [ eq_basis1 ] ) ] that carries the angular momentum @xmath5 of the system , @xmath63_{lm } \nonumber \\
\times \exp \left ( - \frac { \vec{x}^t \underline{a}_k \vec{x}}{2 } \right).\end{aligned}\ ] ] here , @xmath64 collectively denotes the @xmath65 jacobi vectors @xmath66 , where @xmath67 .
the notation @xmath68_{lm}$ ] indicates that the spherical harmonics @xmath69 and @xmath70 are coupled to form a function with angular momentum @xmath5 and projection quantum number @xmath16 . for states with natural parity ,
i.e. , for states whose parity is given by @xmath71 , we choose @xmath72 and @xmath73 @xcite . for states with unnatural parity ( @xmath74 ) , i.e. , for states whose parity is given by @xmath18 , we choose @xmath72 and @xmath75 @xcite .
the basis functions that describe unnatural parity states with @xmath76 have a slightly different form since the construction of states with @xmath76 and @xmath77 requires the coupling of three spherical harmonics with @xmath78 , @xmath79 and @xmath80 @xcite .
the matrix @xmath81 is symmetric and positive - definite , and has dimensions @xmath82 .
the @xmath83 independent elements of @xmath81 are treated as variational parameters and optimized semi - stochastically .
the three - dimensional vectors @xmath84 and @xmath85 , referred to as global vectors since they depend on all @xmath65 jacobi vectors , are defined through @xmath86 and similarly for @xmath85 .
the vectors @xmath87 and @xmath88 are optimized semi - stochastically , where @xmath89 and similarly for @xmath88 . a key benefit of the basis functions given in eq .
( [ eq_basis1 ] ) is that the overlap matrix element @xmath90 , the matrix element for the kinetic energy operator @xmath91 , and the matrix element for the confining potential @xmath92 reduce to compact expressions @xcite . here
, it is understood that the integration is performed over all @xmath93 jacobi coordinates and that @xmath58 is characterized by @xmath81 , @xmath87 and @xmath88 while @xmath94 is characterized by @xmath95 , @xmath96 and @xmath97 .
moreover , a compact expression can also be found for the matrix elements @xmath98 associated with the atom - atom interaction if @xmath35 is modeled by the gaussian potential @xmath48 .
appendix [ appendix ] summarizes explicit expressions of the matrix elements with natural parity ( any @xmath5 ) and unnatural parity ( @xmath74 ) .
the matrix elements for states with @xmath99 symmetry can be found in refs .
@xcite .
we note that the overlap matrix element @xmath100 between two different basis functions does not vanish , i.e. , the basis set employed is not orthogonal .
this implies that the determination of the eigenenergies amounts to the diagonalization of a generalized eigenvalue problem defined by the hamiltonian and overlap matrices @xcite .
while one might think , at first sight , that the non - orthogonality of the basis functions could introduce numerical instabilities , it has been shown in previous work @xcite that numerical instabilities due to linear dependence issues can be avoided completely for the systems of interest in this work if the basis sets are chosen carefully . our strategy to optimize the large number of non - linear variational parameters is quite simple @xcite .
we start with a reference basis set , which could consist of just one basis function or as many as several 100 or 1000 basis functions .
we then enlarge this reference basis set by one basis function , which is chosen from a large number of trial basis functions , typically between several 100 and several 1000 .
each trial function is characterized by a different set of variational parameters . to decide which trial basis function to keep ,
we calculate the energy for each of the enlarged trial basis sets , which consist of the reference basis set plus one of the trial basis functions , and choose the one that results in the largest reduction of the energy of the state of interest
. the state of interest could be the ground state or an excited state .
the procedure is repeated till the basis set is sufficiently complete to describe the state of interest with the desired accuracy . when optimizing a state whose energy is nearly degenerate with that of another state or when optimizing highly excited states , some care needs to be exercised . in the former case
, we find it advantageous to optimize two or more states simultaneously . in the latter case
, we find it beneficial to start with a basis set that provides a reasonably accurate description of the lower lying part of the energy spectrum .
the advantage of our optimization procedure is that the basis set is optimized for a particular state or a particular subset of states .
correspondingly , we work with comparatively small basis sets .
the energies of the @xmath1 and @xmath2 systems at unitarity ( see table [ tab1 ] and supplemental material ) are obtained using basis sets that consist of 700 - 3400 basis functions , while the energies of the @xmath9 and @xmath10 systems ( see supplemental material ) are obtained using basis sets that consist of 1500 - 3800 basis functions .
one key purpose of this paper is to elucidate how we determine a large portion of the energy spectrum of trapped two - component fermi systems with @xmath101 and 5 , and to tabulate the extrapolated zero - range energies .
we believe that the tabulation of the energies is useful as these energies provide much needed highly accurate benchmark results that can be used to assess the accuracy and validity regime of alternative approaches .
we anticipate that the tabulated energies will also prove useful in other applications .
figure [ fig1 ] shows an example of our basis set optimization system at unitarity with @xmath102 symmetry and @xmath103 .
solid and dashed lines show the quantity @xmath104 , where @xmath105/e^{\rm{rel}}_{3,1}(n_b \rightarrow \infty)$ ] , for states 1 and 12 as a function of @xmath106 .
dotted lines show the extrapolation to the @xmath107 limit .
the inset shows a blow - up of the small @xmath106 region .
, width=264 ] for the @xmath2 system with @xmath102 symmetry and @xmath108 at unitarity .
solid and dashed lines show the fractional difference @xmath104 for the ground state ( state 1 ) and state 12 @xcite , respectively , between the relative energy @xmath109 for a basis set of size @xmath62 and the energy for an infinite basis set .
the dotted lines in fig .
[ fig1 ] show the extrapolation to the @xmath107 limit .
it can be seen that the ground state energy converges notably faster than the excited state energy .
the energies for @xmath110 and @xmath111 are @xmath112 for state 1 and @xmath113 for state 12 , respectively .
the basis set errors for these basis sizes are 0.0002% and 0.003% , respectively , i.e. , the energies of states 1 and 12 lie respectively @xmath114 and @xmath115 above the extrapolated energies for the infinite basis set .
the low - lying states of the @xmath2 system with @xmath102 symmetry at unitarity converge relatively quickly with increasing @xmath62 .
the convergence is slower for most other states and , in general , we choose the size of our basis sets for the @xmath1 and @xmath2 systems such that the basis set extrapolation error is smaller than @xmath116 % .
figure [ fig2 ] exemplarily illustrates the range dependence for the relative energy of the @xmath2 system with @xmath102 symmetry at unitarity .
squares show the relative eigenenergies @xmath117 for various ranges @xmath3 of the underlying two - body interaction potential for ( a ) the ground state ( state 1 ) , ( b ) state 12 , and ( c ) state 5 ; @xmath62 is the largest basis set considered .
the energies provide variational upper bounds and the estimated basis set extrapolation error is indicated by errorbars ; in panels ( a ) and ( b ) , the basis set extrapolation error is smaller than the symbol size and thus not visible . in panels ( a ) and ( b ) ,
solid lines show linear fits to the energies @xmath118 [ the fit shown in panel ( a ) includes the energies for the five smallest @xmath3 values ] .
, width=264 ] system with @xmath102 symmetry .
figure [ fig2](a ) shows the range dependence of the ground state energy , fig .
[ fig2](b ) shows the range dependence of the energy associated with state 12 , and fig .
[ fig2](c ) shows the range dependence of the energy for a state that depends comparatively weakly on @xmath3 ( state 5 ) . in figs .
[ fig2](a ) and [ fig2](b ) , the energies vary to a very good approximation linearly with @xmath3 for sufficiently small @xmath119 .
this finding is in agreement with earlier work @xcite .
for the ground state [ see fig .
[ fig2](a ) ] , the range dependence is quite weak and linear behavior is only observed for @xmath120 . in fig .
[ fig2](c ) , the zero - range energy agrees to within @xmath121 with the energy of the non - interacting system .
this , combined with the very weak dependence of the energy on @xmath3 and the fact that the energy approaches the zero - range limit from below , suggests that this state is not affected by @xmath0-wave scattering but only by higher - partial wave scattering .
in the zero - range limit , energy shifts associated with higher - partial wave scattering processes vanish .
our interpretation is corrobated by a perturbative calculation along the lines of that performed in refs .
@xcite , which utilizes zero - range contact interactions . for the @xmath2 system with @xmath122 symmetry , we find , in agreement with our results based on the stochastic variational approach , that there exists one state with relative energy @xmath123 and six states with relative energy @xmath124 that are independent of @xmath12 .
we refer to states that are unaffected by @xmath0-wave interactions as unshifted states .
we find that a relatively large number of states fall into this category .
their existence and likelihood of occurance has already been discussed for the @xmath37 and @xmath1 systems in the literature @xcite . for the @xmath37 system ,
e.g. , all unnatural parity states are unaffected by @xmath0-wave interactions in the zero - range limit . for the @xmath1 and @xmath2 systems ,
unnatural parity states can be affected by @xmath0-wave interactions in the zero - range limit .
the only exception are states with @xmath99 symmetry , which are unshifted .
this behavior can be intuitively understood within a picture that utilizes angular momentum coupling . to construct a state with @xmath99 symmetry ,
the coupling of three finite angular momenta is needed .
these angular momenta can be envisioned as being each associated with one of the three jacobi vectors that characterize the @xmath125 system . as a consequence ,
the @xmath0-wave interactions are effectively turned off by the nodal structure of the wave function . for @xmath126 ,
this argument predicts that states with @xmath99 symmetry can be affected by @xmath0-wave interactions since the system is characterized by one more jacobi vector than angular momenta needed to ensure the @xmath99 symmetry .
indeed , this prediction is in agreement with our results from the perturbative and stochastic variational calculations .
table [ tab1 ] summarizes our extrapolated zero - range energies @xmath127 , @xmath128 , for states with @xmath102 symmetry at unitarity that are affected by @xmath0-wave interactions .
the zero - range energies are obtained by calculating the energies of a given state for several ranges @xmath3 between @xmath129 and by then fitting these energies for the largest basis set considered by a linear function .
.relative energies @xmath130 for the @xmath2 system with
@xmath122 symmetry [ only states that are affected by @xmath0-wave interactions are included ; each energy is @xmath131-fold degenerate ] .
the first column indicates the state number ( st .
the second column shows the extrapolated zero - range energy @xmath127 at unitarity ; the uncertainty is estimated to be 0.1 % or smaller .
the third column indicates the dependence of the energy at unitarity on the range @xmath3 of the gaussian potential @xmath48 .
we assume a linear dependence and write @xmath132 .
the fourth column shows the @xmath133 value determined from the energy ; the value of @xmath133 is only shown for the lowest rung of a ladder , i.e. , for states with @xmath134 .
the last column shows @xmath135 of the non - interacting state that is `` paired '' with the interacting state when determining @xmath136 ( see sec .
[ sec_virial ] ) .
there exist 1 and 6 unshifted states with energy @xmath137 and @xmath138 , respectively . [ cols="^,^,^,^,^",options="header " , ] [ tab_virial ] where the @xmath139 are defined in terms of the canonical partition functions @xmath140 and @xmath141 of the interacting @xmath39 system and the single - component system with @xmath24 atoms , respectively , @xmath142 the temperature - dependent canonical partition functions @xmath140 and @xmath141 , @xmath143\end{aligned}\ ] ] and @xmath144,\end{aligned}\ ] ] are determined by the total energies @xmath145 and @xmath146 of the interacting two - component and non - interacting single - component systems , respectively .
it is important to note that the energies @xmath145 and @xmath146 contain the center of mass energy .
the summation over @xmath147 in eqs .
( [ eq_partint ] ) and ( [ eq_partni ] ) extends over all states allowed by symmetry . for @xmath148 , the sum in eq .
( [ eq_partni ] ) can be performed analytically , yielding eq .
( [ eq_q1 ] ) . in the high - temperature limit ,
one finds for systems with zero - range interactions at unitarity that @xcite @xmath149 the second - order virial coefficient of the trapped system at unitarity takes the simple form @xcite @xmath150\end{aligned}\ ] ] or , performing the infinite sum , @xmath151 the solid line in fig .
[ fig_b2 ] shows the second - order virial coefficient @xmath152 , eq .
( [ eq_b2compact ] ) , as a function of @xmath153 . of the trapped two - component fermi gas at unitarity as a function of the inverse temperature @xmath153 .
the solid line shows @xmath152 , eq .
( [ eq_b2compact ] ) , while the dash - dot - dotted , dash - dotted and dashed lines show @xmath152 obtained by setting @xmath154 in eq .
( [ eq_b2sum ] ) to 0 , 1 and 10 , respectively .
the solid horizontal line shows the high - temperature limit @xmath155 .
, width=245 ] in the high - temperature ( small @xmath153 ) limit , @xmath152 approaches the constant @xmath155 , @xmath156 ( solid horizontal line in fig .
[ fig_b2 ] ) , which can be obtained by taylor - expanding eq .
( [ eq_b2compact ] ) . to illustrate the convergence of @xmath152 with increasing energy cutoff , dash - dot - dotted , dash - dotted and dashed lines show @xmath152 obtained by setting @xmath154 in eq .
( [ eq_b2sum ] ) to 0 , 1 and 10 , respectively . for a finite energy cutoff
, it can be seen that @xmath152 goes to @xmath157 in the small @xmath153 region as opposed to @xmath156 .
as expected , a larger energy cutoff provides an accurate description of @xmath152 over a larger temperature range , i.e. , down to a smaller inverse temperature @xmath153 .
the relative three - body energies at unitarity and for vanishing @xmath0-wave scattering length @xmath12 can be written as @xmath158 ( see sec . [ sec_energies ] ) and @xmath159 , respectively . performing the sum over @xmath160 analytically
, the interaction piece @xmath161 of the trapped three - body system at unitarity takes the form @xcite @xmath162.\end{aligned}\ ] ] using large @xmath163 and @xmath164 , a fully converged pointwise representation of @xmath165 is obtained @xcite ( see solid line in fig .
[ fig_b3 ] ) . of the trapped two - component fermi gas at unitarity as a function of the inverse temperature @xmath153 .
the solid line shows @xmath166 with @xmath163 and @xmath164 set to very large values ( see ref .
@xcite for details ) while the dash - dot - dotted , dash - dotted and dashed lines show @xmath166 obtained by limiting @xmath163 and @xmath164 in eq .
( [ eq_b3sum ] ) such that @xmath167 , @xmath168 and @xmath169 , respectively .
the solid horizontal line shows the high - temperature limit @xmath170 , eq .
( [ eq_b3hight ] ) .
, width=245 ] using the analytical forms for @xmath171 and @xmath152 , eqs .
( [ eq_q1 ] ) and ( [ eq_b2compact ] ) , we find that @xmath172 diverges as @xmath173 in the high - temperature limit .
this divergence is cancelled by a divergence of @xmath174 of opposite sign . as a result
, @xmath165 is well behaved in the small @xmath153 ( high @xmath22 ) limit .
a careful analysis of the high - temperature behavior gives @xcite @xmath175 ( see horizontal solid line in fig . [ fig_b3 ] ) . to illustrate the convergence of @xmath166 with increasing @xmath163 and @xmath164
[ see eq .
( [ eq_b3sum ] ) ] , dash - dot - dotted , dash - dotted and dashed lines in fig . [ fig_b3 ] show @xmath166 calculated using @xmath161 from eq .
( [ eq_b3sum ] ) with @xmath163 and @xmath164 chosen such that @xmath176 , @xmath177 and @xmath169 , respectively .
no cutoff is imposed in calculating @xmath178 . in these calculations ,
we include the same number of @xmath133 and @xmath135 in evaluating @xmath161 , i.e. , each interacting @xmath133 value is paired with the corresponding non - interacting @xmath135 value .
figure [ fig_b3 ] shows that the cutoff introduces a divergence in @xmath165 .
this divergence arises because the cutoff alters the high - temperature behavior of @xmath174 , which implies that the divergencies of @xmath172 and @xmath174 no longer cancel .
importantly , @xmath165 is converged in the low - temperature ( large @xmath153 ) regime even for a relatively small cutoff .
this allows us to use the converged low - temperature tail to constrain @xmath165 in the high - temperature regime . extrapolating @xmath165 ( calculated using a cutoff of @xmath179 ) to the high - temperature limit , we find @xmath180 , which deviates by less than 2% from the exact value .
the validity of the employed extrapolation scheme crucially hinges on the fact that the functional form of @xmath166 changes `` predictably '' as @xmath153 changes from the low- to the medium- to the high - temperature regime .
for example , if @xmath166 changed sign in the medium- or high - temperature regime , as is the case for the coefficient @xmath181 that characterizes the behavior of two identical fermions and one lighter fermion ( with a mass ratio from @xmath182 to @xmath183 ) @xcite , the extrapolation employed above would predict the incorrect high - temperature limit of @xmath181 .
the interaction piece @xmath184 of the fourth - order virial coefficient can be expressed analogously to @xmath161 . in particular , we write the energies at unitarity in terms of the @xmath133 ( see sec . [ sec_energies ] and the supplemental material for a listing of the @xmath133 values ) and
perform , as in the three - body case above , the sum over the hyperradial quantum number @xmath160 analytically . since both natural and unnatural parity states of the four - body systems are affected by the @xmath0-wave interactions , the @xmath133 values corresponding to both natural and unnatural parity states need to be included when evaluating @xmath184 .
the reference piece @xmath185 diverges as @xmath186 in the high - temperature limit .
this divergence must be cancelled by a divergence of @xmath184 of opposite sign .
dash - dot - dotted , dash - dotted and dashed lines in fig . [ fig_b4 ] show @xmath187 at unitarity obtained by using the full expression for @xmath185 and by limiting the sums over @xmath188 and @xmath5 in @xmath184 such that @xmath167 , @xmath189 and @xmath168 , respectively . of the trapped two - component fermi gas at unitarity as a function of the inverse temperature @xmath153 .
the dash - dot - dotted , dash - dotted and dashed lines show @xmath19 obtained by limiting @xmath133 to be smaller than @xmath190 , @xmath191 and @xmath192 , respectively .
the dotted line shows our attempt to extrapolate to the high - temperature limit ; this extrapolation assumes that @xmath19 changes `` predictably '' from the low- to the medium- to the high - temperature regime .
the inset shows the same data as the main figure .
in addition , the solid horizontal line shows the high - temperature limit @xmath193 determined by applying the lda to the fourth - order virial coefficient predicted for the homogeneous system @xcite .
, width=245 ] for the largest cutoff , our calculation includes 169 and 89 @xmath133 values associated with shifted states [ not counting the @xmath131-multiplicity ] of the harmonically trapped @xmath1 and @xmath2 systems with zero - range interactions , respectively .
figure [ fig_b4 ] shows that @xmath19 is negative in the low - temperature ( large @xmath153 ) regime and that neither @xmath19 nor its first or second derivatives with respect to @xmath153 change sign in the regime where @xmath19 is converged .
this motivates us to extrapolate the converged part of @xmath19 to the medium- and high - temperature regime ( see dotted line in fig .
[ fig_b4 ] ) , yielding @xmath194 .
the lda predicts that the virial coefficient @xmath195 of the homogeneous system is related to the high - temperature limit of the @xmath196th order virial coefficient of the trapped system via @xcite @xmath197 application to our extrapolated @xmath193 yields @xmath198 .
this value for the homogeneous system differs in both sign and magnitude from the values @xmath199
@xcite and @xmath200 @xcite determined from experimental data .
these experimental values have been found to be consistent with the equation of state determined by a diagrammatic path integral monte carlo approach @xcite . given the disagreement between our value and that reported in the literature , we speculate that the fourth - order virial coefficient of the trapped system changes sign in the medium- or high - temperature limit , implying that the applied extrapolation scheme does not predict the correct medium- and/or high - temperature behavior of @xmath19 . if this conclusion is correct , it would follow that the determination of the medium- and high - temperature behavior of the fourth - order virial coefficient of the trapped systems requires , if determined via the microscopic energy spectra , knowledge of large portions of the energy spectra of the @xmath1 and @xmath2 systems .
this suggests that other approaches , based on feynman diagrams or based on simulating the finite temperature behavior directly numerically , may be more suitable than the approach pursued here for determining the temperature - dependence of @xmath19
. we also analyzed the low - temperature tail of @xmath201 . using the @xmath9 and @xmath10 energies from the supplemental material , we find that the fifth - order virial coefficient of the trapped fermi gas at unitarity is positive in the low - temperature limit .
high precision measurements of the equation of state in the high - temperature regime might reveal if @xmath201 changes sign as a function of temperature .
more generally , we find that the low - temperature limit of @xmath202 at unitarity is fully determined by the low - temperature behavior of @xmath203 and @xmath171 . to arrive at this result , we derive explicit expressions for @xmath204 and @xmath205 for @xmath206 , and determine the low - temperature behavior of all terms that enter into @xmath204 and @xmath205 . using the ground state energies at unitarity for trapped two - component fermi gases with up to @xmath207 ( with @xmath208 or 1 ) @xcite , we find that @xmath204 falls off faster than @xmath205 with decreasing @xmath22 , thus allowing us to obtain analytic expressions for the leading order low - temperature behavior of @xmath202 : @xmath209\exp[-(2n-3/2 ) \tilde{\omega}]/(2n)$ ] and @xmath210 for @xmath211 .
thus , the sign of @xmath212 in the low - temperature regime is @xmath213 for @xmath214 . for @xmath215
, we have checked that these analytical predictions agree with our numerically determined virial coefficients in the low - temperature regime . while the sign of @xmath212 in the low - temperature regime may not allow one to draw conclusions about @xmath216
, it is interesting , at least from a theoretical point of view , that the sign and functional form of @xmath212 in the low - temperaure regime are fully determined by @xmath171 and @xmath152 .
this paper considered the energy spectra of small trapped two - component fermi gases with vanishing and finite angular momentum as well as natural and unnatural parity .
large portions of the energy spectra of the @xmath1 and @xmath2 systems at unitarity were determined as a function of the range of the underlying two - body model potential and extrapolated to the zero - range limit .
the extrapolated zero - range energies are expected to be universal , i.e. , independent of the underlying gaussian model potential .
portions of the energy spectra of the @xmath9 and @xmath10 systems at unitarity were also determined .
the energies were obtained by solving the relative schrdinger equation using the stochastic variational approach .
compact expressions for the relevant matrix elements were presented in the appendix .
the @xmath1 and @xmath2 energies at unitarity were then used to determine the low - temperature behavior of the fourth - order virial coefficient @xmath19 of the trapped fermi gas .
the high - temperature limit of the fourth - order virial coefficient enters into the universal virial equation of state .
the present study suggests that much larger portions of the microscopic energy spectra are needed to predict the high - temperature limit of @xmath19 . in our view
this is unfortunate . despite this , we believe that the analysis presented illuminates important characteristics relevant to the determination of the virial coefficients . we gratefully acknowledge support by the aro .
kmd and db acknowledge hospitality of the int where part of this work was conducted .
we also gratefully acknowledge communication by the mit - amherst collaboration prior to publication of refs .
this appendix summarizes the expressions for the overlap , kinetic energy , trap potential , and interaction potential matrix elements for states with natural parity ( any @xmath5 ) and unnatural parity ( @xmath74 ) . for notational simplicity , we omit the subscripts of the matrix @xmath81 and the vectors @xmath87 and @xmath88 , and consider the matrix elements between the unsymmetrized basis functions @xmath217 and @xmath218 characterized by @xmath219 and @xmath220 , respectively [ see eq .
( [ eq_basis1 ] ) of sec .
[ sec_system ] ] .
the matrix elements have been derived in the literature @xcite and are summarized here for completeness . before providing explicit expressions for the matrix elements ,
we introduce a number of auxiliary quantities that are utilized in subsecs .
[ appendixa ] and [ appendixb ] .
the product of @xmath218 and @xmath217 can be conveniently written in terms of the matrix @xmath221 , @xmath222 we further define the scalars @xmath223 and @xmath224 ( @xmath225 or @xmath226 ) , @xmath227 and @xmath228 note that the order of the primed and unprimed vectors @xmath229 and @xmath230 matters .
we further define the scalars @xmath231 and @xmath232 ( @xmath225 or @xmath226 ) , @xmath233 and @xmath234 where the diagonal elements of the matrix @xmath235 are given by the inverse of the masses associated with the jacobi vectors and the off - diagonal elements of @xmath235 are zero . in eq .
( [ eq_r ] ) , @xmath236 denotes the trace operator .
the scalars @xmath237 and @xmath238 ( @xmath239 and @xmath240 ) have a similar structure to @xmath231 and @xmath232 , @xmath241 and @xmath242 the matrix @xmath243 is defined as @xmath244 where @xmath245 is the @xmath246-dimensional vector that relates the distance vectors @xmath247 to the jacobi vectors @xmath248 , @xmath249 lastly , we define the total mass @xmath250 , @xmath251 for natural parity states , we use @xmath72 and @xmath73 in eq .
( [ eq_basis1 ] ) , which implies that @xmath218 and @xmath217 are independent of @xmath252 and @xmath253 , respectively . in the following ,
we assume that @xmath218 and @xmath217 are characterizd by the same @xmath5 and @xmath6 values . under these assumptions
the overlap matrix element is given by @xmath254 where @xmath255 is a @xmath5-dependent constant that enters into all matrix elements and thus cancels when calculating expectation values .
the kinetic energy matrix element reads @xmath256 the matrix element for the trapping potential reads @xmath257
lastly , the interaction matrix element for the gaussian potential can be written as @xmath258 \rangle.\end{aligned}\ ] ] the expression for the matrix element @xmath259 | \psi \rangle$ ] reduces to that for the overlap matrix element if the matrices @xmath260 and @xmath261 are replaced by @xmath262 and @xmath263 , respectively . for unnatural parity states with @xmath74
, we use @xmath72 and @xmath75 in eq .
( [ eq_basis1 ] ) . in the following , we assume that @xmath218 and @xmath217 are characterizd by the same @xmath5 and @xmath6 values . under these assumptions
the overlap matrix element is given by @xmath264 where @xmath265 is a @xmath5-dependent constant that enters into all matrix elements and thus cancels when calculating expectation values .
the kinetic energy matrix element reads @xmath266 \left ( \rho_{11}\rho_{22}-\rho_{12 } \rho_{21 } \right ) + \nonumber \\ 2\rho_{11 } \left(\rho_{11 } s_{22}+\rho_{22}s_{11}-\rho_{12}s_{21}-\rho_{21}s_{12 } \right ) \}.\end{aligned}\ ] ] the matrix element for the trapping potential reads @xmath267 \left ( \rho_{11}\rho_{22}-\rho_{12 } \rho_{21 } \right ) + \nonumber \\ 2\rho_{11 } \left(\rho_{11 } \tilde{s}_{22}^{(pq)}+\rho_{22}\tilde{s}_{11}^{(pq)}- \rho_{12}\tilde{s}_{21}^{(pq)}-\rho_{21}\tilde{s}_{12}^{(pq ) } \right ) \}.\end{aligned}\ ] ] as in the natural parity case , the expression for the interaction matrix element for the gaussian potential can be related to that of the overlap matrix element by making the appropriate substitutions .
the states are counted separately for each @xmath268 symmetry . in counting the states
, we do include the unshifted states but we do not account for the @xmath269 degeneracy of the energies ( see column 1 of table [ fig1 ] for the state number ) . | equal - mass two - component fermi gases under spherically symmetric external harmonic confinement with large @xmath0-wave scattering length are considered . using the stochastic variational approach , we determine the lowest 286 and 164 relative eigenenergies of the @xmath1 and @xmath2 systems at unitarity as a function of the range @xmath3 of the underlying two - body potential and extrapolate to the @xmath4 limit .
our calculations include all states with vanishing and finite angular momentum @xmath5 ( and natural and unnatural parity @xmath6 ) with relative energy up to @xmath7 , where @xmath8 denotes the angular trapping frequency of the external confinement .
our extrapolated zero - range energies are estimated to have uncertainties of 0.1% or smaller .
the @xmath1 and @xmath2 energies are used to determine the fourth - order virial coefficient of the trapped unitary two - component fermi gas in the low - temperature regime .
our results are compared with recent predictions for the fourth - order virial coefficient of the homogeneous system .
we also calculate small portions of the energy spectra of the @xmath9 and @xmath10 systems at unitarity . |
among the structures shown by interacting galaxies ( barnes & hernquist 1992 ) , the so - called collisional ring galaxies ( theys & spiegel 1976 , 1977 ) are a rare and most distinctive sub - class ( appelton & struck - marcell 1996 ) .
collisional ring galaxies occur from an almost head - on collision between a smaller galaxy travelling along the minor axis of a larger galaxy passing through , or close to , its centre ( appleton & struck - marcell 1996 ) .
the shock wave sweeps up and expels gas from the system , usually leaving a gas - poor galaxy behind ( freeman & de vaucoleurs 1974 ) .
the pre - eminent example is the cartwheel galaxy ( eso350 - 40 ; am0035 - 335 ) at a distance of 124mpc . the cartwheel was discovered by zwicky ( 1941 ) , and studied in - depth by fosbury & hawarden ( 1977 ) and amram et al .
the cartwheel shows a spectacular ring of star formation , with interior spokes surrounding the host galaxy .
other well - known collisional systems include arp148 ( mayall s object ) at 148mpc ( smith 1941 ; burbidge , burbidge & prendrgast 1964 ; arp 1966 ) , ic 298 ( arp147 ; gerber , lamb & balsara 1992 ) at 135mpc , and am0644 - 741 ( eso34 - 11 or the `` lindsay - shapley ring '' ) at 87mpc ( lindsay & shapley 1960 ; graham 1974 ; arp & madore 1987 ) .
this latter object is often considered to be a typical exemplar of evolved ringed galaxies ( e.g. higdon , higdon & rand 2011 ) . while star - forming events resulting from galactic collisions may be more common than usually thought ( e.g. block et al .
2006 ) , spectacular cartwheel - type galaxies , showing complete rings ( but not necessarily with evident spokes ) , appear to be very rare in volume - limited surveys .
the previous nearest example was the vela ring galaxy ( am1006 - 380 = eso316 - 43 ) , at 72mpc ( dennefeld , lausten & materne 1979 ) .
partial ring galaxies are more numerous .
probably the nearest in the madore , nelson & petrillo ( 2009 ) catalogue is am0322 - 374 at 20mpc .
ugc9893 , at 11mpc ( kennicutt et al .
2008 ) , is also listed as a collisional system , but it lacks a clear ring structure ( gil de paz & madore 2003 ) , appearing indistinct in available imagery .
collisional ring galaxy systems can be identified in high- resolution , deep h@xmath1 imagery where the hii regions of the ring are readily visible .
however , h@xmath1 galaxy surveys such as gil de paz & madore ( 2003 ) , james et al .
( 2004 ) , and kennicutt et al .
( 2008 ) have not added to the modest number of cartwheel - type galaxies currently known and there are @xmath820 convincing systems , e.g. marston & appleton 1995 .
bosch et al . ( 2015 ) present the most recent detailed work on such systems for an apparently empty ring galaxy system eso474g040 at @xmath9mpc , thought to have arisen from a recent merger of two disk galaxies .
collisional rings systems offer an excellent environment to study the evolution of density - wave induced starbursts ( marston & appleton 1995 ; romano , mayya & vorobyov 2008 ) , as well as the dynamic interactions between the individual galaxies ( e.g. smith et al .
a closer example of a collisional ring system would allow its star formation , stellar distribution and evolution to be studied in detail , spatially , kinematically and chromatically . here
we present the discovery of a new collisional ring galaxy at a distance of only 10mpc .
our paper is organised as follows . in section 2
we present some background and describe the discovery of the ring around eso179 - 13 . in section 3
we present our follow - up high - resolution imaging and other archival multi- wavelength images .
optical spectroscopy is presented in section 4 .
parameters for the system , such as masses and star formation rates , are provided in section 5 .
section 6 presents our discussion and give our summary in section 7 .
eso179 - 13 is noted as an interacting double galaxy system in simbad , appearing to consist of a near edge - on late - type spiral with another more chaotic system @xmath380arcsec to the north - east .
it was first identified by sersic ( 1974 ) and independently re - discovered by lauberts ( 1982 ) , longmore et al .
( 1982 ) and corwin , de vaucouleurs & de vaucouleurs ( 1985 ) .
a detailed @xmath10-band image was published by laustsen , madsen & west ( 1987 ) .
woudt & kraan - korteweg ( 2001 ) identify the most prominent galaxy as wkk7460 ( componenta hereafter ) , and wkk7463 for the secondary ( component b ) . a small , third galaxy is also identified as wkk7457 to the west of componenta ( componentc ) .
it agrees with our own positional determination to 1arcsec .
the system , centred on componenta , is entry 391 in the h@xmath1 catalogue of kennicutt et al .
( 2008 ) , but they do not list integrated h@xmath1 ( + [ nii ] ) flux or luminosity estimates .
lee et al .
( 2011 ) , in a related paper , included this system in their galaxy evolution explorer ( galex ) ultraviolet imaging survey of local volume galaxies with a quoted @xmath11 = 15 , but no near - uv or far - uv fluxes are quoted .
owing to its proximity to the galactic plane , the galaxy was not observed by galex prior to mission completion ( bianchi 2014 ) .
componenta has been classified as an sb(s)m spiral in the rc3 catalogue ( de vaucouleurs et al .
1991 ) and similarly classified as sb(s)dm system ( t = 7.5 ) by buta ( 1995 ) . these independent classifications treat the system as a distinct , late - type spiral galaxy , albeit in an interaction .
it can be best described as a magellanic dwarf or a magellanic spiral , similar to the lmc .
a range of heliocentric radial velocities for componenta are available in the literature from both optical spectroscopic and hi measurements .
the most recent optical results are strauss et al .
( 1992 ) who quote 775@xmath1236 kms@xmath0 , di nella et al .
( 1997 ) , 750@xmath1270 kms@xmath0 , and woudt et al .
( 1999 ) , 757@xmath1250 km s@xmath0 .
these are in good agreement so we take an unweighted average of 761kms@xmath0 ( @xmath13=11kms@xmath0 ) for componenta . the earliest hi heliocentric velocities towards the system are 836@xmath125kms@xmath0 by longmore et al .
( 1982 ) , 840@xmath1220kms@xmath0 by tully ( 1988 ) and 843@xmath129kms@xmath0 by de vaucouleurs et al .
the parkes hi ` hipass ' multibeam receiver programme gives 843@xmath125kms @xmath0 ( hipass catalogue ; meyer et al .
2004 ) where it is noted as the 60@xmath14 brightest in terms of integrated flux density ( koribalski et al .
a deeper , pointed parkes observation was obtained by schrder et al .
they give parameters for componenta , but note that a contribution from componentb is likely .
their heliocentric hi radial velocity is 842kms@xmath0 from the hi profile mid - point , with a velocity width of 222kms@xmath0 at 20% of peak .
the integrated hi flux density is [email protected]@xmath0 , in good agreement with the hipass value of 100 .
all the hi velocities are consistent within small errors .
these values include the combined contribution from all the gas in the full system
. we adopt 842@xmath125kms@xmath0 for the combined hi system velocity .
kennicutt et al .
( 2008 ) give a distance of 9mpc based on a virgocentric flow model distance .
we adopt a slightly further distance of 10.0@xmath15 mpc , based on kennicutt et al .
( 2008 ) , but scaling from a hubble constant h@xmath16 of 75 to 68 kms@xmath0mpc@xmath0 due to the latest wmap ( hinshaw et al . 2013 ) and planck results ( ade et al .
2014 ) . at this distance
, 1arcsec equates to 48.5pc .
wouldt & kraan - korteweg ( 2001 ) give a modest colour excess of @xmath17(b - v)@xmath18 mag in the surrounding field , giving @xmath19 mag while schlafly & finkbeiner ( 2011 ) give @xmath20 mag for galactic extinction in this general direction , which we adopt hereafter .
we will also use @xmath21 mag based on the reddening law from howarth ( 1983 ) .
apart from the basic information above , the system has been little studied because of its location in a crowded , low - latitude area near the galactic plane ( @xmath22 - 8 degrees ) and proximity to the bright , 7.7 mag a0iv star , hd150915 , located 72arcsec due south .
there are actually two bright stars here , but only hd150915 is listed in simbad .
the second star is @xmath310arcsec north of hd150915 and @xmath32magnitudes fainter .
it was undoubtedly blended with hd 150915 in the original photographic imagery .
the low heliocentric velocity meant the system was discovered as a strong h@xmath1 emitter in the anglo - australian observatory ( aao ) / u.k .
schmidt telescope ( ukst ) supercosmos h@xmath1 survey of the southern galactic plane ( shs ; parker et al .
2005 ) during searches for galactic planetary nebulae ( parker et al . 2006 ) .
the bandpass of the survey filter ( parker & bland - hawthorn 1998 ) includes all velocities of galactic gas and emission from extragalactic objects out to the virgo and fornax galaxy clusters at @xmath31200kms@xmath0 .
[ shs ] shows the 4 @xmath23 4arcmin discovery images from the shs survey data centred on the system . from left to right there is the h@xmath1 image , the matching broadband short - red ` sr ' image and the image obtained by taking the sr image from the h@xmath1 image ( effectively a continuum subtracted image ) .
componenta is at the centre .
componentb is labelled to the north - east while the smaller compact componentc is indicated with an arrow .
the shs continuum subtracted image reveals a spectacular ring of emission knots not obvious , without context , even in the h@xmath1 image while the main componenta galaxy has almost disappeared .
the ring is approximately ( but not perfectly ) centred on componenta .
surprisingly , given its previous spiral morphological classification , componenta does not show up well in the shs continuum subtracted image and appears to show little star formation .
this indicates that along with its apparently highly elongated nature ( so not an elliptical ) it is has been largely stripped of gas .
the smaller , more irregular galaxy ( componentb ) shows extensive , clumpy h@xmath1 emission . a third , faint compact system ( componentc )
is seen 38arcsec to the west of componenta , identified as wkk7457 .
this object is clear in the broad - band @xmath10 , @xmath24 and @xmath25-band supercosmos images and is also undergoing star - formation .
the strong hi detection noted earlier indicates that a significant , neutral hi component still remains within the overall system environment .
the lack of much star formation in componenta suggests that most of this hi may be located outside of its inner disk .
discovery of this collisional ring system led us first to investigate and compile existing multi - wavelength data from the archives .
we then obtained deeper , higher resolution imaging and spectroscopy of the major emission components ( see below ) , to provide estimates of some key physical parameters for this system .
high resolution imagery at the blanco 4-m telescope at ctio was obtained in june 2008 using the wide field mosaic camera .
we used five filters : @xmath26-band , h@xmath27 , h@xmath1 off - band ( 80 redward of h@xmath1 ) , , and a ( wider ) off - band filter centred at 5300 .
the field of view is 30arcmin on a side and the plate scale 0.26arcsecpx@xmath0 .
the seeing measured from the data frames was typically 1.2arcsec .
exposures were 5 minutes for the emission - line and off - band filters , and 60s for the @xmath26-band filter .
the airmass during the exposures averaged 1.3 . to subtract the continuum ,
the h@xmath1 off - band image was scaled down by a factor of 0.88 , and the off - band by a factor of 0.14 , to compensate for the filter widths .
these factors were established by minimizing residuals for a set of selected field stars .
for that works very well and results are in good agreement with the filter curve .
for h@xmath1 this can be an issue if the chosen stars have h@xmath1 in absorption .
this does not appear to be the case for the field stars chosen which provide a consistent scale - factor to @xmath28 .
it is best to minimize stellar residuals rather than filter curves as stars are a more important contribution to the continuum than bound - free emission from hi regions . a montage of these ctio images is shown in fig .
they reveal the distribution and variety of the emission structures as well as data for quantitative flux estimates . from these deeper
, higher resolution ctio h@xmath1 and continuum subtracted images the distribution of ionised gas is seen to be far more complex than is evident from the shs discovery data .
the leftmost panel is a @xmath29arcminute ctio h@xmath1 image with the off - band frame subtracted .
we avoid the bright star to the south where ccd blooming is serious .
emission associated with componentc and at the southern tip of componenta s disk previously gave the impression of a circular ring in the shs as the bright star obscured emission further south .
the higher resolution ctio data reveals an elliptical emission ring with a major - axis diameter of 127arcsec ( @xmath36.2kpc ) .
the inscribed oval was positioned to fit the prominent emission now seen to extend further south .
the ellipse is not centred on componenta , but at 16h47m19.5s , @xmath3057@xmath3126 44 ( j2000 ) , @xmath313arcsec ( or @xmath30.6kpc ) south - south - east .
some low - level h@xmath1 emission is now seen across componenta , surrounding componentb and with faint , localised emission features and blobs around the entire system .
an interesting feature is the elongated nature of some of these emission knots in a north - east direction from a to b. the main emission features follow the equivalent h@xmath1 structures though at typically a third to a half of their native integrated pixel intensities .
the mid panel is the matching ctio continuum subtracted image ( narrow on and off - band images used ) .
both these images are presented at 90% linear scaling to reveal the full extent of the emission ( this sets the upper and lower limits based on the 90% pixel intensity level where a histogram of the data is created and the limits are set to display the percentage about the mean value ) . with full pixel range
, saturated pixels from the 7.7th magnitude star hd150915 would prevent detail being seen .
low level diffuse flux is seen for componenta but it is not conspicuous .
this shows it has almost been completely cancelled out indicating componenta is mostly composed of normal starlight .
the right hand panel is the off - band filter image ( 5300 ) that effectively represents the @xmath26-band starlight without the + h@xmath32 emission line contribution .
this shows dust lanes along and at the south - west extremity of the disk , concentrated starlight from componentsa , b and c and an envelope of diffuse starlight around the entire system .
this common envelope " was already noted by laustsen et al .
( 1987 ) from old @xmath10-band photographic data taken with the eso 3.6-m telescope in the 1980s .
this raises an interesting question and also a possible explanation for why the apparent axial ratios of the highly elongated componenta differs so much from the surrounding oval ring .
currently known collisional ring systems with low impact parameter have axial ratios that are broadly similar to that of the target galaxy .
such arrangements are a natural consequence of the copious gas and ism being swept up and compressed by the resultant density wave having a similar radial ( and perhaps also vertical ) distribution as the target galaxy stellar content .
this does not appear to be the case here unless what we are actually seeing is the residual , less - inclined bar of a more face - on disk system that extends out to the entire ring and whose presence can still be seen in the low - level diffuse starlight refereed to earlier .
if componenta is not a residual bar but a distinct late - type disk galaxy ( as commonly assumed ) stripped of gas , then the gas distribution could have become less flattened somehow .
alternatively the original target galaxy could have been stretched out along the direction of the impact towards componentb before , during or even after the bulls - eye collision occurred .
though componentb is only 13@xmath33 of the mass ( in stars ) of componenta these effects could have been strengthened due to the low ` magellanic'-type masses of the system as a whole .
finally , fig .
[ ctio - rgb ] is a composite red , green blue ( rgb ) ctio image displaying the system s spectacular nature .
the red channel is h@xmath1 , the green and the blue broadband v. componenta is mostly star - light with dust lanes evident in this higher resolution data .
the few small hii knots appear red , likely due to dust extinction within the disk .
based on the shs data , literature data and our new ctio images , table [ fund_data ] presents some key parameters for galaxy components a , b and c , including the central position of the best fit ellipse to the main emission ring . the position for componenta
is tied to the 2mass ( skrutskie et al .
2006 ) near - ir imagery and astrometry .
the positions refer to the centre of each assumed galaxy .
the centre of componenta is derived from the dominant starlight in the @xmath24-band .
the separation between the centres of componentsa and b is @xmath380arcsec .
using the distance of 10mpc gives a projected separation between them of 3.9kpc .
componenta has a major - axis of @xmath386arcsec ( 4.2kpc ) measured to the outer edge in the stellar continuum from off - band and @xmath26-band ctio images where a decline in the background is seen before the star - forming ring is encountered . when checked against the broad band @xmath24 , @xmath25 and 2mass images when autocut at the 90% pixel intensity level the disk appears to extend only to 70 - 80arcsec . at low isophotal thresholds starlight
extends out to the star - forming ring both along the major and minor axes .
componentb has a major axis of @xmath333arcsec ( 1.6kpc ) determined in the same way .
the strongest h@xmath1 emission in b is also in an oval structure of nucleated knots @xmath332arcsec across .
this is in excellent agreement with the value estimated from the concentrated starlight .
starlight internal to this emission oval is clear in fig .
[ ctio - rgb ] and b may extend to 70arcsec ( @xmath33.4kpc ) in low level , isolated clumps of star formation and diffuse starlight .
the overall system envelope , including all assumed associated broad and narrow band emission and diffuse starlight as seen across the region in the right panel of fig .
[ ctio ] , has a major diameter of 318arcsec and a minor axis of 198arcsec , giving a total system size of @[email protected] . [
cols="<,^,^,^,^,^",options="header " , ] [ summary_data ] eso197 - 13 is an important system that can be used to study collisional rings as it is much closer than any equivalent , opening new possibilities for analysis .
system dynamics can be studied at unprecedented spatial resolution as 0.1arcsec ( e.g. hst and/or ao resolution ) is only 25pc .
this is sufficient to explore details of the system - wide star formation process and to allow detection of individual stars .
it would be a prime candidate to measure the star formation history in the regions recently traversed by the ring .
during visual scans of the shs ( parker et al . 2005 ) , a spectacular new collisional ring galaxy ( or cartwheel analogue ) has been found , eso179 - 13 , which we have named kathryn s wheel . with an estimated distance of only 10.0mpc ,
it is in the local supercluster but has been previously missed due its location near the galactic plane and its projected proximity to a bright 7.7-magnitude star which interferes with observation . the velocity range of all optical emission components from the main ring , componentb and the published velocity for componenta is about 200kms@xmath0
this is completely encapsulated within the fwhm of 222kms@xmath0 reported for the hi profile at the 20% level centred at a heliocentric velocity of 842kms@xmath0 which we consider representative of the overall interacting system . at a distance of 10mpc ,
the system has an overall physical size of @xmath315kpc for all major components and the diffuse surrounding envelope , a total mass of @xmath4m@xmath5 ( stars + hi ) and a metallicity of [ o / h]@xmath3 - 0.4 , that classifies it as a magellanic - type system .
it has a large reservoir of neutral gas and is the 60@xmath14 brightest galaxy in the hipass hi survey . by some margin
it is also the nearest ring system known .
this system offers an unprecedented opportunity to undertake detailed studies of this rare phenomenon , where two reasonably well matched galaxies have a close encounter centred close to the potential well of the primary .
this paper used the simbad and vizier services operated at the cds , strasbourg , the nasa / ipac extragalactic data base ( ned ) and data products of the aao / ukst h@xmath1 survey , produced with the support of the anglo - australian telescope board and uk particle physics and astronomy research council ( now the stfc ) .
this research used data from the wide - field infrared survey explorer , a joint project of the university of california , los angeles , and jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration , and data from the 2mass , a joint project of the university of massachusetts and the infrared processing and analysis center california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
the authors acknowledge help of dr anna kovacevic at the ctio telescope .
we are grateful for stimulating discussion with kate zijlstra .
we thank an anonymous referee who provided excellent feedback to help improve the quality of this paper .
amram p. , mended de oliveira c. , boulesteix j. , balkowski c. , 1998 , a&a , 330 , 881 appleton p.n .
, struck - marcell c. , 1996 , fund . cosmic phys
, 16 , 111 arp h.c . , 1966 ,
apj , 140 , 1617 arp h.c . , madore b. , 1987 , a catalogue of southern peculiar galaxies and associations .
cambridge , new york : cambridge university press baldwin j.a . ,
phillips m.m . , terlevich r. , 1981 , pasp , 93 , 5 barnes j.e . , hernquist l. , 1992 , ara&a , 30 , 705 beichman c.a . , neugebauer g. , habing h.j . , clegg p.e .
, chester t.j . ,
1988 , the infrared astronomical satellite ( iras ) catalogs and atlases .
volume 1 : explanatory supplement bell e.f .
, mcintosh d.h . , katz n. , weinberg m.d . , 2003 , apjs , 149 , 289 bendo g.j .
et al . , 2006 ,
apj , 652 , 283 bendo g.j .
et al . , 2012 ,
mnras , 419 , 1833 bendo g.j .
et al . , 2015 ,
mnras , 448 , 135 bianchi l. , 2014 , ap&ss , 354 , 103b block d.l .
et al . , 2006 ,
nature , 443 , 832 boselli a. et al .
, 2012 , a&a , 540 , 54 boselli a. , fossati m. , gavazzi g. , ciesla l. , buat v. , boissier s. , hughes t. , 2015 , a&a , 579 , 102 brosch n. , vaisanen p. , kniazev a.y .
, moisav a. , 2015 , mnras , in press , arxiv:1506.00409 burbidge e.m . ,
burbidge g.r . ,
prendergast k.h . , 1964 ,
apj , 140 , 1617 buta r. , 1995 , apjs , 96 , 39 childress m. j. , vogt f. p. a. , nielsen j. , sharp r. g. , 2014 , ap&ss , 349 , 617 cluver m. et al . , 2014 , apj , 782 , 90 condon j.j . , 1992 , ara&a , 30 , 575 conn b.c . , pasquali a. , pompei e. , lane r.r . ,
chene a - n .
, smith r. , lewis g.f .
, 2011 , apj , 741 , 80 corwin h.g . , de vaucouleurs a. , de vaucouleurs g. , 1985 , univ .
texas monogr .
astron . , 4 , 1 davies j.i .
et al . , 2012 ,
mnras , 419 3505 dennefeld m. , lausten s. , materne j. , 1979 , a&a , 74 , 123 de vaucouleurs g. , de vaucouleurs a. , corwin h.g . , buta r.j . , paturel g. , fouque p. , 1991
, third reference catalogue of bright galaxies , volume iii .
new york : springer di nella h. , couch w.j .
, parker q.a .
, paturel g. , 1997 , mnras , 287 , 472 donghia e. , mapelli m. , moore b. , 2008 , mnras , 389 , 1275 dopita m. , et al . , 2010 , ap&ss , 327 , 245 doyle m.t . , et al , 2005 , mnras , 361 , 34 draine b.t .
et al . , 2007 , apj .
, 663 , 866 drew j.e .
2014 , mnras , 440 , 2036 driver s.p . , 1999 , apj , 526 , l69 drory n. , bender r. , feulner g. , hopp , u. , maraston c. , snigula j. , hill , g.j . , 2004 , apj , 608 , 742 engelbracht c. , et al . , 2005 , apj , 628 , l29 engelbracht c. , et al . , 2008 , apj , 678 , 804 few j.m.a .
, madore b.f .
, 1986 , mnras , 222 , 673 fosbury r.a.e .
, hawarden t.g , 1977 , mnras , 178 , 473 freeman k.c . , de vaucouleurs g. , 1974 , apj , 194 , 569 frew d.j . ,
parker q.a .
, 2010 , pasa , 27 , 129 frew d.j .
, bojii i.s . ,
parker q.a . , 2013 , mnras , 431 , 2 frew d.j . , bojii i.s . ,
parker q.a .
, pierce m.j .
, gunawardhana m.l.p . ,
reid w.a . , 2014a , mnras , 440 , 1080 frew d.j .
et al . , 2014b , mnras , 440 , 1345 gaustad j.e . , mccullough p.r . , rosing w. , van buren d. , 2001 , pasp , 113 , 1326 gerber r.a . ,
lamb s.a . ,
balsara d.s . , 1992 ,
apjl , 399 , l51 gil de paz a. , madore b.f .
, 2003 , apjs , 147 , 29 graham j. , 1974 , observatory , 94 , 290 gunawardhana m.l.p .
. , 2013 , mnras , 433 , 2764 hambly n.c .
et al . , 2001a , mnras , 326 , 1279 henriques b. , maraston c. , monaco p. , fontanot f. , menci n. , de lucia g. , tonini c. , 2011 , mnras , 415 , 357 higdon j.l . , higdon s.j.u . ,
rand r.j . , 2011 , apj , 739 , 97 hinshaw g. et al . , 2013 ,
apjs , 208 , 19 howarth i. , 1983 , mnras , 203 , 301 ishihara d. et al . , 2010 ,
a&a , 514 , 1 james p.a .
, 2004 , a&a , 414 , 23 jarrett t.h . ,
chester t. , cutri r. , schneider s.e .
, huchra j.p . , 2003 , aj , 125 , 525 jarrett t.h .
et al . , 2012 ,
aj , 144 , 68 kankare e. et al . , 2014 ,
a&a , 572 , a75 kennicutt r.c . , 1984 ,
apj , 287 , 116 kennicutt r.c . , bresolin f. , bomans d.j . , bothun g.d . ,
thompson i.b .
, 1995 , aj , 109 , 594 kennicutt r.c . ,
, funes s.j . , jos g. , sakai s. , akiyama s. , 2008 , apjs , 178 , 247 kennicutt r.c . , evans n.j . , 2012 , ara&a , 50 , 531 koribalski b.s .
et al . , 2004 ,
aj , 1 , 16 lauberts a. , 1982 , eso / uppsala survey of the eso(b ) atlas . garching : european southern observatory laustsen s. , madsen c. , west r.m . , 1987 , exploring the southern sky .
garching : european southern observatory lee , j.c .
2011 , apjs , 192 , 33 li a. , draine b.t . , 2001 ,
apj , 554 , 778 lindsay e.m . , shapley h. , 1960 , observatory , 80 , 223 longmore a.j .
et al . , 1982 ,
mnras , 200 , 325 madore b.f . , nelson e. , petrillo k. , 2009 , apjs , 181 , 572 marston a.p . ,
appleton p.n . , 1995 ,
aj , 109 , 1002 masci f. , 2013 , astrophysics source code library , 2010 masci f.j . , fowler j.w .
, 2009 , astronomical data analysis software and systems xviii , 411 , 67 mcdonald i. , zijlstra a.a . , boyer m.l . , 2012 , mnras , 427 , 343 melbourne j. , boyer m.l . , 2013 , apj , 764 , 30 meyer m.j .
et al . , 2004 ,
mnras , 350 , 1195 murphy t. , mauch t. , green a. , hunstead r.w . , piestrzynska b. , kels a.p .
, sztajer p. , 2007 ,
mnras , 382 , 382 parker q.a . ,
bland - hawthorn j. , 1998 , pasa , 15 , 33 parker q.a .
et al . , 2005 , mnras , 362 , 689 parker q.a .
et al . , 2006 , mnras , 372 , 79 pilyugin l.s . , vilchez j.m . ,
thuan , t.x . , 2010 , apj 720 , 1738 planck collaboration : ade p.a.r .
et al . , 2014 , a&a , 571 , 28 relao m. , kennicutt r.c .
, eldridge j.j . , lee j.c . ,
verley s. , 2012 , mnras , 423 , 2933 rodgers a.w . , harding p. , bloxham g. , bessell m.s .
, 1998 , pasp , 100 , 626 romano r. , mayya y.d .
, vorobyov e.i .
, 2008 , aj , 136 , 1259 ryder s.d . , walsh w. , malin d. , 1999 , pasa , 16 , 84 sabin l. et al . , 2013 , mnras , 431 , 279 schiminovich d. , wyder t.k .
, martin d. , 2007 , apjs , 173 , 315 schlafly e.f .
, finkbeiner d.p . , 2011 , apj , 737 , 103 schroder a. , kraan - korteweg r.c . , henning p.a . , 2009 ,
a&a , 505 , 1049 skrutskie m.f .
et al . , 2006 ,
aj , 131 , 1163 sersic j.l . , 1974 ,
, 28 , 365 smith r.t . , 1941 ,
pasp , 53 , 187 smith r. , lane r.r . , conn b.c . ,
fellhauer m. , 2012 , mnras , 423 , 543 stark d.v . , kannappan s.j . ,
wei l.h . ,
baker a.j .
, leroy a.k . , eckert k.d . ,
vogel s.n . , 2013 , apj , 769 , 82 staveley - smith l. , kim s. , calabretta m.r . ,
haynes r.f .
, kesteven m.j . , 2003 , mnras , 339 , 87 strauss m.a . , huchra j.p . , david m. , et al . , 1992 , apjs , 92 , 29 taylor e.n .
et al . , 2011 ,
mnras , 418,1587 theys j.c . , spiegel e.a .
, 1976 , apj , 208 , 650 theys j.c . ,
spiegel e.a . , 1977 , apj , 212 , 616 tully r.b . , 1988 , nearby galaxies catalog
. cambridge and new york : cambridge university press van der marel r. , kallivayalil n. , besla g. , 2009 , in proc .
256 , the magellanic system : stars , gas , and galaxies , j.t .
van loon & j.m .
oliveira , eds .
, p 81 viironen k. , delgado - inglada g. , mampaso a. , magrini l. , corradi r.l.m .
, 2007 , mnras , 381 , 1719 vogt f.p.a . , dopita m.a .
, borthakur s. , verdes - montenegro l. , heckman t.m . , yun m.s .
, chambers k.c . , 2015 , mnras in press ( arxiv:1504.03337 ) wright a.e . , griffith m.r . , burke b.f . , ekers r.d . , 1994 ,
apj . suppl .
91 , 111 woudt p.a . ,
kraan - korteweg r.c . ,
fairall a.p . , 1999 ,
a&a , 352 , 39 woudt p.a , kraan - korteweg r.c . , 2001 , a&a , 380 , 441 wright e.l .
et al . , 2010 ,
aj , 140 , 1868 yamamura i. et al . , 2009 ,
ser . , 418 , 3 zwicky f. , 1941 , in theodore von krmn anniversary volume : contributions to applied mechanics and related subjects .
pasadena : california institute of technology . | we report the discovery of the closest collisional ring galaxy to the milky way .
such rare systems occur due to `` bulls - eye '' encounters between two reasonably matched galaxies . the recessional velocity of about 840kms@xmath0 is low enough that it was detected in the aao / ukst survey for galactic h@xmath1 emission .
the distance is only 10.0mpc and the main galaxy shows a full ring of star forming knots , 6.1kpc in diameter surrounding a quiescent disk .
the smaller assumed bullet " galaxy also shows vigorous star formation .
the spectacular nature of the object had been overlooked because of its location in the galactic plane and proximity to a bright star and even though it is the 60@xmath2 brightest galaxy in the hi parkes all sky survey ( hipass ) hi survey .
the overall system has a physical size of @xmath315kpc , a total mass of @xmath4m@xmath5 ( stars + hi ) , a metallicity of [ o / h]@xmath6 , and a star formation rate of 0.2 - 0.5m@xmath5yr@xmath0 , making it a magellanic - type system .
collisional ring galaxies therefore extend to much lower galaxy masses than commonly assumed .
we derive a space density for such systems of @xmath7 , an order of magnitude higher than previously estimated .
this suggests kathryn s wheel is the nearest such system .
we present discovery images , ctio 4-m telescope narrow - band follow - up images and spectroscopy for selected emission components . given its proximity and modest extinction along the line of sight , this spectacular system provides an ideal target for future high spatial resolution studies of such systems and for direct detection of its stellar populations .
galaxies : groups : general galaxies : interactions galaxies : kinematics and dynamics galaxies : star formation galaxies : individual ( eso179 - 13 ) |
in the last decade transport properties of mesoscopic superconductor / normal metal ( s / n ) structures were intensively studied ( see for example the review articles @xcite and references therein ) . it was established that in these nano - structures , i.e. in structures whose dimensions are less than the phase coherence length @xmath1 and the inelastic scattering length @xmath2 , the conductance changes when decreasing the temperature below the critical temperature @xmath3 of the superconducting transition and this variation may be both positive ( @xmath4 ) and negative ( @xmath5 ) @xcite . the increase or decrease of the conductance
@xmath6 depends , in particular , on the interface resistances and is determined by a competition between two contributions caused by the proximity effect .
one of them is due to the suppression of the density - of - state ( dos ) and leads to a decrease of the conductance .
the second one results in increasing @xmath6 and is similar to the maki - thompson term @xcite .
these studies apparently were stimulated by the theoretical work @xcite in which a weak - localization correction to the conductance of a s / n / s mesoscopic structure was calculated and an oscillatory dependence of this correction on the phase difference @xmath7 between the superconductors s was predicted .
these oscillations have been indeed observed @xcite but their amplitude turned out to be two orders of magnitude larger than the predicted one . in order to overcome this discrepancy between theory and experiment
one should take into account the proximity effect .
it was shown that the latter leads to much larger amplitudes of the conductance oscillations than the weak - localization corrections @xcite .
the proximity effect manifests itself also in other interesting peculiarities of the transport properties of s / n structures .
one of them is an interesting dependence of the josephson current @xmath8 on an additional dissipative current @xmath9 through a n wire in a 4-terminal s / n / s structure similar to the one shown in the inset of fig.[fig_geometry_triplet ] . according to theoretical predictions
the current @xmath8 changes sign if the current @xmath9 is large enough @xcite .
this behavior of the josephson current ( @xmath10-contact ) was later experimentally confirmed@xcite .
another interesting effect is a non - monotonic temperature @xmath11 ( or voltage @xmath12 ) dependence of the correction to the conductance @xmath13 @xcite . when decreasing the temperature , @xmath13 increases , reaches a maximum and with further decrease of the temperature drops to zero .
this behavior has been explained theoretically for the case of a short s / n contact ( @xmath14 ) @xciteand a long s / n structure ( @xmath15 ) @xcite , here @xmath16 is the thouless energy , @xmath17 is the diffusion constant , @xmath18 is the length of the n wire ( film ) .
the reason for this behavior is the competition between two contributions to the conductance mentioned above .
it is interesting to note that low - energy states play an important role in transport properties of s / n structures .
the reason is that the condensate penetrates into the n wire over a length @xmath19 which may be much larger than the thermodynamic correlation length @xmath20 provided the characteristic energy @xmath21 therefore , in the limit of low ( compared to @xmath11 ) thouless energy @xmath22 the phase coherence is maintained over distances @xmath23 of the order of the length @xmath18 of the n wire .
these long - range phase - coherent effects have been predicted in @xcite and observed on a 4-terminal s / n / s structure @xcite . in ref.@xcite
the conductance oscillations related to the phase coherence were observed in a s / n / s mesoscopic structure
. it was established for the first time that these oscillations survive in a temperature range where the josephson coupling between the superconductors becomes negligible .
the authors of ref.@xcite observed an increase in the josephson critical current when an additional dissipative current was injected into the n region in a 4-terminal s / n / s structure .
this current leads to long - range effects affecting the critical current . recently ,
similar investigations have been carried out on mesoscopic f / s structures in which ferromagnets ( f ) are used instead of normal ( nonmagnetic ) metals .
it is well known that in the dirty limit , when the relaxation momentum time @xmath24 is very small ( @xmath25 , where @xmath26 is the exchange energy ) , the condensate penetrates into a ferromagnet over a length @xmath27 .
the latter is extremely short ( 5 - 50 ) for strong ferromagnets like fe or ni .
therefore one might expect that the influence of the proximity effect on transport properties of such structures should be negligibly small .
experiments carried out recently on f / s structures showed however that the conductance variation @xmath13 is quite visible ( varying from about 1 to 10% ) when the temperature decreases below @xmath3 @xcite .
it is worth mentioning that the conduction variation was both positive and negative . in some experiments
the variation @xmath13 was related to a variation of the interface conductance ( resistance ) @xcite , whereas in others @xcite to the conduction variation of the ferromagnetic wire @xmath28 .
the theory also predicts both an increase @xcite and decrease@xcite of the conductance . in refs .
@xcite a ballistic contact was analyzed .
it was shown that at @xmath29 the contact conductance @xmath30 is twice as large as its conductance @xmath31 in the normal state ( above @xmath3 ) , which agrees with a theoretical prediction for a n / s ballistic contact , and drops to zero at @xmath32 where @xmath33 is the fermi energy .
the conductance of a diffusive point contact @xmath30 has been calculated by golubov @xcite who showed that @xmath30 is always smaller than the conductance @xmath31 in the normal state . in the case of a mixed conductivity mechanism ( partly diffusive and partly ballistic )
@xmath34 has been calculated in ref . @xcite and it may be both larger or smaller than the conductance in the normal state @xmath31 . in order to obtain the resistance of the system shown in fig.[fig_geometry_triplet ] one should add the interface resistances @xmath35 and the resistance of the f wire @xmath36 .
the conductance of the f wire @xmath37 under the assumption that @xmath35 is sufficiently small has been calculated in refs .
@xcite . as in ref.@xcite ,
the proximity effect was neglected .
it turned out that the conductance of the f wire in the normal state was @xmath38 and was equal to @xmath39 in the superconducting state ( these formulas are valid for a f wire shorter than the spin relaxation length , see @xcite ) .
thus , the conductance of the f wire decreases below @xmath3 .
the mechanism responsible for this behavior is in this case purely kinetic , since the form of the distribution function depends only on the boundary conditions at the interfaces : if the interface transparencies are perfect the distribution function is continuous in the normal state , while in the superconducting state the boundary condition is changed providing the spinless current through the f / s interface . these two different boundary conditions for the distribution function lead to the different values of the conductance above and below @xmath3 . in the present paper
we analyze the conductivity of a f / s structure in the dirty limit when the condition @xmath25 is satisfied . in the next section we study the kinetic mechanism of the conductance variation in a f / s mesoscopic structure neglecting the proximity effect .
the diffusion coefficients and the interface transparencies are assumed to be different for each spin direction .
it will be shown that in this case the conductance variation is always negative , _
i.e. _ the conductance decreases with decreasing temperature below @xmath3 . in the third section we take into account the proximity effect ( the condensate penetration into the ferromagnetic wire ) and
present the results for the conductance on the basis of exact calculations assuming equal diffusion constants and interface transparencies for spin up and down .
we will see that for a strong exchange field @xmath40 , the resistance variation @xmath41 is positive at any temperature and interface resistance .
finally , in section 4 we calculate the conductance variation assuming that the magnetization near the f / s interface has a local inhomogeneity . it will be shown that not only a singlet component arises in the f wire in this case but also a triplet one which penetrates into the f wire over a long distance of the order of @xmath42 , where @xmath43 .
the penetration of the triplet component leads to a positive @xmath13 which may be comparable with the conductance variation observed in n / s structures .
the singlet component exists only in a very short region near the f / s interface and leads to a negligible contribution to the conductance .
in this section , we calculate the conductance of the system shown in fig.[fig_geometry1 ] . we neglect influence of the proximity effect on the conductance assuming that the penetration length of the condensate is very small in comparison with the length of the f wire .
the penetration length of the condensate is equal to @xmath27 in the dirty limit and to the mean free path @xmath44 in the
clean limit ( to be more exact , in the limit of a strong ferromagnet , when the condition @xmath45 is satisfied ) .
the conduction variation is related in this case to different forms of the distribution function in the normal and superconducting states .
the form of the distribution function is determined by boundary conditions at the interfaces f / n ( @xmath46 ) or f / s ( @xmath47 ) . in the normal state
the spin current is finite , whereas in the superconducting one it is equal to zero . in the superconducting state at low temperatures ,
charge is transferred via andreev reflections , i.e. an incident electron is reflected from the f / s boundary and moving back as a hole with the opposite spin direction ( in the s region the charge is carried by singlet cooper pairs with zero total spin ) . in order to find the charge or spin
current we need to calculate the distribution function @xmath48 obeying the diffusion equation @xmath49 here @xmath50 is the distribution function of electrons for a given spin direction @xmath51 , @xmath52 is the distribution function of holes , @xmath53 is the diffusion coefficient which depends on the spin index @xmath54 @xmath55 is the inelastic collision integral .
the inelastic collision integral is of the order of @xmath56 , where @xmath57 is the inelastic scattering time .
we ignore spin relaxation processes in the f wire assuming that the spin relaxation length exceeds the length @xmath18 .
it will be seen that the conductance in the superconducting state is always larger than in the normal one .
the simplest formulas are obtained in the case of a perfect f / s interface .
therefore , we assume that there is no barrier at the f / s interface and the f/f interface transparency is arbitrary ( f is the ferromagnetic or normal reservoir ) . in the next section we will show that the non - zero reflection coefficient at the f / s interface leads also to an increase of the resistance when temperature is lowered below @xmath58 the boundary conditions at the f/f interface have the same form in the normal and superconducting states ( see appendix ) @xmath59 here @xmath60 @xmath61 is the barrier resistance per unit area for a given spin direction @xmath51 , @xmath62 @xmath63 is the density - of - states ( dos ) .
the function @xmath64/2 $ ] is the distribution function in the f reservoir . at the f / s interface
the boundary conditions are different in the normal and superconducting case .
above @xmath65 they are @xmath66 below @xmath3 they have the form ( see appendix ) @xmath67 @xmath68 physically , eqs.([k2 ] ) and ( [ k3 ] ) mean that the electric potential at the perfect f / n(s ) interface is chosen to be zero .
the potential @xmath69 is expressed in terms of @xmath70 as @xcite @xmath71 $ ] .
the condition ( [ k3 ] ) implies that the spin current is zero .
this fact is a result of the andreev reflections .
it is worth mentioning that , in the model under consideration , the conductance related to the andreev reflections differs from zero only if the amplitude of the condensate function at the f / s interface is not equal to zero .
therefore , strictly speaking , we neglect the proximity effect everywhere in the f wire except the nearest neighborhood of the f / s interface .
the electrical current is given by the formula @xmath72 \label{k5}\ ] ] the spin current @xmath73 is given by the same formula with sign `` - '' instead of `` + '' in the brackets and a factor of @xmath74 in front of the integral ( @xmath75 is the bohr magneton ) .
the problem is reduced to finding a solution for eq.([k1 ] ) with the boundary conditions ( [ k2]-[k4 ] ) .
the inelastic collision integral has a complicated form and finding a solution is in general an unrealistic task .
therefore , we consider limiting cases only . in this case of a long inelastic relaxation length @xmath77 , we can neglect the inelastic collision integral and seek a solution in the form @xmath78 where @xmath79 is a constant that determines the current and , hence , the conductance .
the constants @xmath80 and @xmath81 can be determined from the boundary conditions ( [ k2]-[k4 ] ) and are equal to @xmath82 in the normal case , and @xmath83 in the superconducting case , where @xmath84 is the resistance of the f wire in the normal state . knowing the distribution function we can compute the conductances and the spin current @xmath73 . in the normal state , we obtain @xmath85 \;.
\label{k9}\ ] ] whereas the result for the superconducting state reads @xmath86 as follows from eqs.([k9]-[k10 ] ) , the spin current in the normal state is absent only if the conductances in each spin band @xmath87 and @xmath88 are the same . in the superconducting state
this current is always zero .
the conductances @xmath87 and @xmath88 enter the formulas for @xmath89 and @xmath90 ( [ k9]-[k10 ] ) symmetrically . in the case of a small f/f interface resistance ,
we obtain for @xmath89 and @xmath90 the formulas presented in @xciteand in the introduction .
the conductance in the normal state @xmath89 is always larger than or equal to ( if conductances in each spin channel are the same ) the conductance in superconducting state @xmath90 .
indeed , the difference between @xmath89 and @xmath90 can be written in the form @xmath91 ^{2}/{\cal d}_{f / n}{\cal d}_{f / s},\;. \label{k12}\ ] ] where @xmath92 and @xmath93 are the denominators in the expressions ( [ k9]-[k10 ] ) .
therefore , in the model under consideration ( dirty limit , no condensate penetration , zero f / s interface resistance ) the conductance decreases with decreasing temperature below @xmath3 .
note that the formula for the conductance eq.([k10 ] ) is valid at low temperatures ( @xmath94 ) when the contribution of the states with @xmath95@xmath96 to the conductance can be neglected . at arbitrary temperatures
the conductance can be easily found with the help of eq.([k5 ] ) and has the form @xmath97 where @xmath98 is the conductance of the f / s structure at zero temperature determined by eq.([k10 ] ) .
we took into account that at @xmath99 the boundary conditions for the distribution function are given by eq.(3 ) .
let us consider another limiting case when the length @xmath18 of the f - wire exceeds the energy relaxation length @xmath101 in this case the inelastic scattering term in eq.([eq_1 ] ) dominates .
it turns to zero provided the distribution function has the equilibrium form @xmath102/2\;. \label{k13}\ ] ] where the potential @xmath103 depends on the spin index @xmath51 and is linear as a function of the coordinate @xmath104 : @xmath105 . integrating eq .
( [ k1 ] ) over the energies we obtain a relation between the current in the @xmath51 spin band and @xmath106 @xmath107 let us consider first the normal case . in the f wire
we have @xmath108 from the last two equations we find @xmath106 @xmath109 the electric field and the total current are equal to the sum of @xmath110 and @xmath111 over spin indices respectively .
we easily obtain an expression for the conductance @xmath89 , which is identical to eq .
( [ k9 ] ) . in the superconducting case the current @xmath111
is given by @xmath112 the potentials @xmath113 and @xmath114 are related to each other by eq .
( [ k3 ] ) : @xmath115 .
although the form of the distribution function differs from eqs.([k6]-[k7 ] ) , the conductance @xmath90 below @xmath3 remains unchanged with respect to the previous case ( see eq . ( [ k10 ] ) ) .
thus , regardless of the relationship between the lengths @xmath18 and @xmath116 , the conductance of the structure under consideration is given by eqs .
( [ k9]-[k10 ] ) , _
i.e _ it decreases with decreasing temperature below @xmath3 .
in the preceding section we neglected the proximity effect and assumed that the f / s interface resistance was equal to zero . we have shown that the conductance decreases with decreasing temperature provided the conductivities of the ferromagnetic wire or the f/f interface for spin up and down differ from each other . in the present section
we calculate the conductance for an arbitrary f / s interface resistance taking into account the proximity effect ( the penetration of the superconducting condensate into the f wire ) . in this case
the problem becomes more complicated and , for simplicity , we assume that conductivities for the both spin directions are the same .
an equation for the distribution function can be solved exactly and this allows one to present the normalized zero - bias conductance @xmath117 in the form @xmath118 where @xmath119 are the normalized resistances of the f / s and f/f interfaces respectively ( in the normal state ) , @xmath120 @xmath121 is the retarded green s function in the superconductor , @xmath122 is the quasiparticle normalized conductance at a given energy , @xmath123 is the normalized subgap conductance ( the conductance related to andreev reflections ) , @xmath124 .
the angle brackets mean the averaging over the length : @xmath125 the physical meaning of the denominator is simple : the first term in the angle brackets is the normalized resistance of the f wire in the presence of the condensate , the second and third terms are the resistances of the f/f and f / s interfaces below @xmath3 respectively .
one can see that the andreev
conductance @xmath126 is not zero only if the condensate function at the interface ( from the ferromagnetic side ) @xmath127 differs from zero .
the factor @xmath128 in the numerator arises from the normalization of the conductance . in order to calculate the normalized conductance @xmath129
, we have to find the green functions @xmath130 and @xmath131 in the ferromagnetic wire .
these functions obey the usadel equation which upon the substitution @xmath132 and @xmath133 acquires the well known form @xmath134 this equation is complemented by the boundary conditions @xmath135 @xmath136 the solution for eq.([k19 ] ) can be found analytically in some limiting cases @xcite .
it has an especially simple form in the case of a weak proximity effect when @xmath137 and eq.(19 ) may be linearized ( numerical calculations show that the linearized solution differs from the exact solution by less than 10% for @xmath138 ) . here
we present the results of a numerical solution of the usadel equation eq.([k19 ] ) and of the calculation of the conductance @xmath139 in fig.[fig_plot1 ] the temperature dependence of the conductance variation @xmath140 is presented for a structure with good interface transparencies at various exchange energies @xmath141 ( normalized to the thouless energy @xmath142 ) .
it is seen from this figure that @xmath143 is positive only if @xmath26 is not too large compared to @xmath96 or @xmath3 .
a small positive value @xmath143 is observed near @xmath3 for @xmath144 and @xmath145 for @xmath146 the conductance variation is negative and decreases with decreasing @xmath147 for example , in the case of f / al structure the threshold exchange energy @xmath148 above which the conductance deviation is negative for all temperatures , is equal to @xmath149 @xmath150 , that is , this value is much less than the characteristic magnitudes of @xmath26 for ferromagnets as fe , ni , co etc .
a slight increase of @xmath90 above @xmath89 is related to the condensate penetration into the f wire .
this increase occurs only near @xmath3 because in this temperature range the conductance @xmath90 is close to the conductance @xmath31 in the normal state due to a large contribution of the quasiparticle current .
when the temperature decreases , the quasiparticle contribution also decreases and the weak ( at large @xmath141 ) proximity effect can not overcome this decrease of the conductance .
the contribution to the conductance due to proximity effect is suppressed even stronger if the f / s interface resistance is not small compared to @xmath151 . in fig .
[ fig_plot2 ] we plot the temperature dependence of @xmath143 for the case when the f / s interface resistance ( in the normal state ) is two times larger than the conductance of the f wire .
one can see that the normalized conductance decreases with decreasing temperature and becomes very small at low @xmath147 the reason for this behavior of the conductance is quite clear . at high enough temperatures the main contribution to
the conductivity is due to quasiparticles with energies @xmath152 with lowering the temperature , this contribution decreases .
the contribution caused by andreev reflections ( the term @xmath126 in eq.(19 ) ) at large @xmath141 and not small @xmath153 is very small as the condensate amplitude is small . indeed , in this case one can linearize the usadel equation ( 20 ) and obtain for @xmath154 @xmath155 it follows from this equation that at @xmath156 the amplitude of @xmath157 is small ( we consider low energies @xmath158 therefore the andreev conductance
@xmath126 is also small .
the suppression of the andreev reflections in the model considered follows directly from the usadel equation and the widely used boundary conditions of ref .
it is worth mentioning that the andreev reflections are responsible for a subgap conductance in n / i / s junctions , which was observed in ref.@xcite .
the suppression of the andreev conductance by the exchange field is analogous to the suppression of the subgap conductance by an external magnetic field@xcite .
in this section , we consider again a ferromagnetic wire with a metallic contact with a superconductor . in the previous sections we have shown that the proximity effect in the presence of a strong exchange field @xmath26 could be neglected , since the superconducting condensate penetrates into the f wire only over a short distance ( @xmath159 in the dirty limit ) . at the same time ,
the interface resistance @xmath160 increases when the normal reservoir becomes superconducting . in short ,
the total conductance decreases with decreasing temperature below the superconducting critical temperature @xmath3 .
however , in recent experiments on s / f structures a considerable increase of the conductance below @xmath161 was observed @xcite . the measurements also demonstrate that the entire change of the conductance is due to an increase of the conductivity of the ferromagnetic wire as if the superconducting condensate penetrated into the ferromagnet .
a question arises : how can the condensate function penetrate into the f wire over large distances ?
we understand that the cooper pairs must be destroyed by the strong exchange field because the electrons of the pairs can no longer have opposite spins to answer this question we notice that such a simple argument about the destruction of the condensate is based on the assumption that only the singlet pairing exists in the sample .
this argument can not be correct if a triplet component of the superconducting condensate penetrates the ferromagnet .
an arbitrary exchange field can not destroy the triplet superconducting component since both the electrons of the cooper pair are in the same spin band . in this section ,
we suggest a mechanism of formation of the triplet pairing , which is due to a local inhomogeneity of the magnetization in the vicinity of the s / f interface .
we will show that the penetration length of the triplet component into the ferromagnet is of the order @xmath162 , where the energy @xmath95 is of the order of the temperature @xmath11 or the thouless energy @xmath163 , and therefore the increase of the conductance due to the proximity effect may be comparable with that in a s / n structure . we consider the system shown in fig.[fig_geometry_triplet ] and assume that there is a domain wall in the region @xmath164 described by the angle @xmath165 between the magnetization @xmath166 and the @xmath167-axis . in this region
the magnetization is given by @xmath168 for simplicity we assume that @xmath165 varies linearly according to @xmath169 .
we consider again the diffusive limit corresponding to a short mean free path and to the condition @xmath170 , which allows us to describe the system using the usadel equation@xcite .
we assume that the resistance of the f / s interface is not too small and therefore the condensate amplitude @xmath171 is small .
we use the linearized usadel equation for the retarded matrix ( in spin space ) green function @xmath172 , which has the form ( the index @xmath173 is dropped ) @xmath174 here the matrix @xmath175 is defined as @xmath176 .
we also assume that the diffusion coefficient is the same for the both spin bands .
this equation is supplemented by the boundary condition at the s / f interface , which after linearization takes the form @xcite @xmath177 where @xmath178 is the resistivity of the ferromagnet , @xmath35 is the s / f interface resistance per unit area in the normal state , and @xmath179 .
we have assumed that there are no spin - flip processes at the s / f interface .
the solution of eq .
( [ a2 ] ) in the region @xmath164 can be sought in the form @xmath180 where @xmath181 is an unitary transformation given by @xmath182 .
substituting eq .
( [ rotation2 ] ) into eq .
( [ a2 ] ) we obtain the following equation for @xmath183 @xmath184 where @xmath185 is the anticommutator . in the region @xmath186 the magnetization is homogeneous and @xmath183 satisfies eq .
( [ a5 ] ) with @xmath187 .
we see from eq .
( [ a5 ] ) that the singlet and triplet component of the condensate function are mixed by the rotating exchange field @xmath26 . in the region
@xmath186 these components decouple and they should be found by matching the solutions at @xmath188 ( the function @xmath189 is continuous in the entire f wire ) . the solution of eq .
( [ a5 ] ) can be written in the form @xmath190 here the function @xmath191 is the amplitude of the triplet component , whereas @xmath192 and @xmath193 describe the singlet pairing .
the structure of eq .
( [ a5 ] ) allows to seek these amplitudes in the form @xmath194 the functions @xmath193 and @xmath191 can be written in a similar way .
the coefficients @xmath195 , @xmath196 and @xmath197 obey the algebraic equations @xmath198 where @xmath199 and @xmath200 ( indices @xmath201 were dropped ) .
the eigenvalues @xmath202 are the values at which the determinant of eqs .
( [ 4 ] ) turns to zero . in the case of a strong exchange field @xmath26
, @xmath203 is large ( @xmath204 ) and the eigenvalues @xmath202 are given by @xmath205 the eigenvalues @xmath206 correspond to a sharp decay of the condensate in the ferromagnet , while @xmath207 is associated with the slowly varying part . with the boundary condition , eq .
( [ bound - cond ] ) , we see that in a homogeneous case ( @xmath208 the amplitude of the triplet component @xmath191 is zero .
if @xmath209 the function @xmath191 is coupled to the singlet components @xmath192 and @xmath210 and we consider this case .
if the width @xmath211 is small , the triplet component changes only a little in the region @xmath212 and spreads over a large distance of the order @xmath213 in the region @xmath214 . in the case of a strong exchange field @xmath26
, @xmath215 is very short ( @xmath216 ) , the singlet component decays very fast over the length @xmath215 , and its slowly varying part @xmath217 is @xmath218 . in order to obtain the expression for the triplet component
@xmath219 we use eq .
( [ bound - cond ] ) at @xmath220 and assume that the solution vanishes at @xmath221 then , we find @xmath222 ^{-1}\!\right\ } ^{r(a)}\ ; , \label{5}\ ] ] where @xmath223 , @xmath224 is the amplitude of the singlet component at the s / f interface , @xmath225 , @xmath226 , and @xmath227 .
it is clear from eq.([5 ] ) that , at the interface , the triplet component is of the same order of magnitude as the singlet one .
indeed , for the case @xmath228 we obtain from eq .
( [ 5 ] ) @xmath229 , where @xmath230 is the angle characterizing the rotation of the magnetization . therefore , provided the angle @xmath231 and the s / f interface transparency is not too small , the singlet and triplet components are not small .
they are of the same order in the vicinity of the s / f interface but , while the singlet component decays fast over a short distance ( @xmath232 ) , the triplet one varies smoothly along the ferromagnet turning to zero at the f reservoir ( see fig.[fig_penetration ] ) .
one can see also that the singlet component oscillates , which is well known @xcite .
the penetration of the triplet component into the ferromagnet is similar to the penetration of the superconducting condensate into a normal metal .
the presence of the condensate function ( triplet component ) in the ferromagnet can lead to long - range effects and therefore to a significant change of the conductance of a ferromagnetic wire in a s / f structure ( see inset in fig.[fig_geometry_triplet ] ) when the temperature decreases below @xmath3 .
the normalized conductance variation @xmath233 is given by the expression @xcite : @xmath234 ^{2}\right\rangle \;.
\label{6}\ ] ] here @xmath89 is the conductance in the normal state , @xmath235 denotes the average over the length of the ferromagnetic wire between the f reservoirs , and @xmath236 is given by the expression @xmath237 \;.\ ] ] substituting eqs .
( [ 2 ] , [ 5 ] ) into eq .
( [ 6 ] ) one can determine the temperature dependence @xmath238 shown in fig.[fig_conductance_q ] .
we see that @xmath143 increases with decreasing temperature and saturates at @xmath239 . in order to explain the reentrant behavior of @xmath240 observed in refs .
@xcite , one should take into account other mechanisms as those analyzed in refs .
@xcite and in section 2 . however , this question is beyond the scope of the present analysis .
it is also interesting to note that a triplet component of the condensate function with the same symmetry ( odd in frequency @xmath241 and even in momentum @xmath242 ) has been suggested long ago by berezinskii @xcite as a possible phase in superfluid @xmath243 ( this , so called `` odd superconductivity '' , was discussed in a subsequent paper @xcite ) .
being symmetric in space , this component is not affected by potential impurities , in contrast to the case analyzed in ref .
@xcite , where the triplet component of the condensate was odd in space .
while this hypothetical condensate function is not realized in @xmath243 ( in @xmath243 it is odd in @xmath242 but not in frequency ) , this odd ( in @xmath241 ) triplet component does exist in the system considered here , although under special conditions described above . as it was mentioned in the introduction ,
experimental data are still controversial .
it has been established in an experiment @xcite that the conductance of the ferromagnet does not change below @xmath3 and all changes in @xmath143 are due to changes of the s / f interface resistance @xmath35 .
however , in other experiments @xmath35 was negligibly small @xcite .
the mechanism suggested in our work may explain the long - range effects observed in the experiments @xcite . at the same time , the result of the experiment @xcite is not necessarily at odds with our findings .
the inhomogeneity of the magnetic moment at the interface , which is the crucial ingredient of our theory , is not a phenomenon under control in these experiments .
one can easily imagine that such inhomogeneity existed in the structures studied in refs .
@xcite but was absent in those of ref .
the magnetic inhomogeneity near the interface may have different origins .
anyway , a more careful study of the possibility of a rotating magnetic moment should be performed to clarify this question .
we have analyzed the conductance variation @xmath244 of the f / s mesoscopic structure in the dirty limit when the condition @xmath245 is satisfied .
first , neglecting the proximity effect we have shown that below @xmath3 the conductance variation @xmath246 is negative and its magnitude increases with increasing the difference between the conductances for spin up and down .
the change of the conductance is related to changes in the boundary conditions at the f / s interface . below @xmath3 one of the boundary conditions
requires the spin current to be zero .
the formulas for the conductance remain valid even if inelastic scattering ( not spin flip ) processes are taken into account .
we also studied how the proximity effect affects the conductance for an arbitrary transparency of the f / s interface .
the account for the condensate penetration ( the proximity effect ) leads to an increase of the conductance near @xmath3 .
this is possible , however , only if the exchange energy @xmath26 is not too large : the parameter @xmath247 should not be large , where @xmath141 is the exchange energy in units of the thouless energy @xmath248 and @xmath153 is the ratio of the f / s interface resistance in the normal state to the resistance of the f wire .
if the parameter @xmath247 is large , the conductance @xmath90 decreases with decreasing temperature and becomes very small at low @xmath147 this behavior is related to a strong suppression of the andreev reflection processes that determine the conductance at low @xmath11 by the exchange field .
it is important to note that we neglected the inverse proximity effect , that is , suppression of superconductivity in s. in the case of large @xmath141 one has to take into account this effect and calculate a deviation of the green s functions in the superconductor from their bulk values .
it turns out that in the case of a large exchange energy @xmath249 the energy gap in the superconductor may be suppressed and the dos @xmath250 is not zero even at @xmath95@xmath251 . in this case
the quasiparticle conductance is not zero and the charge transfer is possible through the f / s interface at low energies @xmath252 the conversion of the quasiparticle current into the current of cooper pairs is realized in the superconductor over the coherence length @xmath253 ( at low energies ) . at last ,
we have calculated the conductance of a f / s structure in which the magnetization is non - homogeneous near the f / s interface .
it was shown that , in this case , a triplet component arises alongside with the singlet one .
this triplet component in another systems ( he@xmath254 , high @xmath3 superconductors etc ) was studied in several papers @xcite ; it is odd in matsubara frequencies ( the so called odd superconductivity ) and even in momentum space ( in the main approximation in the parameter ( @xmath255 ) ) .
the triplet component penetrates into the f wire over a large distance of the order @xmath256 and leads to a positive conductance variation .
the singlet component penetrates over much shorter length of the order @xmath257 and leads to a negligible contribution to the conductance .
it would be interesting to realize experimentally the situation that we analyzed in the last section and to observe the triplet component not in an exotic system but in an ordinary f / s structure .
here we show how the boundary conditions for the distribution functions @xmath258 can be derived from matching conditions for the quasiclassical green functions , where @xmath259 is the distribution function of electrons and @xmath260 is the distribution functions of holes ( having in mind a superconductor , it is better to speak about electron- and hole - like excitations ) .
we use the boundary conditions for the quasiclassical matrix green functions @xmath261 the matrix elements of which consist of the retarded ( advanced ) green s functions @xmath262 and the keldysh green s function @xmath263 the last matrix is expressed in terms of the matrix distribution function @xmath264 @xmath265 , where the matrix @xmath266 can be represented in the form : @xmath267 the first function @xmath268 determines the order parameter in the superconductor and is equal @xmath269 the second function @xmath270 determines the electrical or spin current and equals @xmath271 the matching condition at the f(n)/s interface for the 4@xmath2724 matrix green function has the form @xcite where @xmath274 is the f / s interface resistance ( in the normal state ) normalized to the resistance of the f wire @xmath275 the physical meaning of eq.(a1 ) is simple . if we take the keldysh ( the element ( 12 ) ) of eq.(a1 ) , multiply it by @xmath276 and calculate the trace , we obtain the current at a given energy @xmath95 .
if we integrate over energies , we obtain on the left side the usual expression for the current where @xmath278 is the dos in the normal state . on the right hand side
we obtain the current at a given energy which is well known in the tunnel hamiltonian method .
we take the keldysh component of eq .
( a1 ) , multiply it by @xmath276 and @xmath279 and calculate the trace .
we obtain thus the equations the symbol \{ ... }@xmath289 stands for @xmath290 . in the case under consideration
@xmath291 @xmath292 and @xmath293 @xmath294so that @xmath295 .
we are interested in the states with subgap energies ( low temperatures ) @xmath296 for these states one has @xmath297 and @xmath298 therefore we obtain that @xmath299 and @xmath300 but @xmath301 one can show that at any non - zero @xmath153 the functions @xmath302 and @xmath303 are not zero and we find from eqs.(a3-a4 ) v. i. falko , a. f. volkov , and c. j. lambert , phys .
b * 60 * , 15394 ( 1999 ) ; v. i. falko , a. f. volkov , and c. j. lambert , jetp lett . *
69 * , 532 ( 1999 ) ; f.taddei , j.h.jefferson . and c.j.lambert , phys.rev.lett .
* 82 * , 4938 ( 1999 ) .
fig.3 : the dependence of the conductance variation @xmath310 , for different values of the exchange field @xmath311 .
here @xmath312 , @xmath313 and @xmath315 .
the solid , dashed and dotted lines correspond to @xmath316 , @xmath317 and @xmath318 respectively .
fig.5 : spatial dependence of the singlet ( dashed line ) and the triplet ( solid line ) components of @xmath171 in the f wire for different values of @xmath319 .
here @xmath320 , @xmath321 and @xmath322 .
@xmath323 is the thouless energy . | we calculate the conductance of a superconductor / ferromagnet ( s / f ) mesoscopic structure in the dirty limit .
first we assume that the ferromagnet exhibits a homogeneous magnetization and consider the case that the penetration of the condensate into the f wire is negligible and the case in which the proximity effect is taken into account .
it is shown that if the exchange field is large enough , the conductance below the critical temperature @xmath0 , is always smaller than the conductance in the normal state .
at last , we calculate the conductance for a f / s structure with a local inhomogeneity of the magnetization in the ferromagnet .
we demonstrate that a triplet component of the condensate is induced in the f wire.this leads to a increase of the conductance below @xmath0
. 0.5 cm 15.5 cm 0.5 cm 0.5 cm |
recently there has been renewed interest in studying the distribution of quark and gluon fields in the three - quark static - baryon potential .
while early studies were inconclusive @xcite , improved computing resources and analysis techniques now make it possible to study this system in a quantitative manner @xcite . in particular , it is possible to directly compute the gluon flux distribution @xcite using lattice qcd techniques similar to those pioneered in mesonic static - quark systems @xcite . like okiharu and woloshyn @xcite
our first interest is to test the static - quark source - shape dependence of the observed flux distribution , as represented by correlations between the quark positions and the action or topological charge density of the gauge fields . to this end , we choose three different ways of connecting the gauge link paths required to create gauge - invariant wilson loops . in the first case , quarks are connected along a t - shape path , while in the second case an l - shape is considered .
finally symmetric link paths approximating a y - shape such that the quarks approximate an equilateral triangle are considered .
the latter is particularly interesting as the probability of observing @xmath1-shape flux - tubes are maximized in this equidistant case . because the signal decreases in an exponential fashion with the size of the loop , it is essential to use a method to enhance the signal .
we use strict three - dimensional ape smearing @xcite to enhance the overlap of our spatial - link paths with the ground - state static - quark baryon potential .
time - oriented links remain untouched to preserve the correct static quark potential at all separations .
excellent signal to noise is achieved via a high - statistics approach based on translational symmetry of the four - dimensional lattice volume and rotational symmetries of the lattice as described in detail below .
this approach contrasts previous investigations using gauge fixing followed by projection to smooth the links and resolve a signal in the flux distribution @xcite .
to study the flux distribution in baryons on the lattice , one begins with the standard approach of connecting static quark propagators by spatial - link paths in a gauge invariant manner .
ape - smeared spatial - link paths propagate the quarks from a common origin to their spatial positions .
the smearing procedure replaces a spatial link , @xmath2 , with a sum of @xmath3 times the original link plus @xmath4 times its four spatially oriented staples , followed by projection back to @xmath5 .
we select the unitary matrix @xmath6 which maximizes @xmath7 , where @xmath8 is the smeared link , by iterating over the three diagonal @xmath9 subgroups of @xmath5 repeatedly .
we repeat the combined procedure of smearing and projection @xmath10 times , with @xmath11 .
untouched links in the time direction propagate the spatially separated quarks through euclidean time . for sufficient time evolution the ground state is isolated .
finally smeared - link spatial paths propagate the quarks back to the common spatial origin .
the three - quark wilson loop is defined as : @xmath12 where @xmath13 is a staple made of path - ordered link variables @xmath14 and @xmath15 is the path along a given staple . in this study , we consider two - dimensional spatial - link paths .
figures 1 and 2 give the projection of t and l shape paths in the @xmath16 plane .
figure 3 shows the idea behind the y - shape . in all cases
the three quarks are created at the origin , o ( white bubble ) , then are propagated to the positions q1 , q2 or q3 ( black circle ) before being propagated through time and finally back to a sink at the same spatial location as the source ( o ) .
-plane.,width=113,height=75 ] -plane.,width=113,height=75 ] -plane.,width=113,height=75 ] for the y - shape we create elementary diagonal `` links '' in the form of boxes as shown in fig .
4 . the @xmath17 and @xmath18 boxes are the average of the two path - ordered link variables going from one corner to the diagonally opposite one . taking both of these paths
better maintains the symmetry of the ground state potential and therefore provides improved overlap with the ground state .
we also consider @xmath19 boxes which are the averages of the possible paths connecting two opposite corners using 1x1 and 1x2 boxes .
we will create further link paths in the future for use in bigger loops as necessary .
hence a diagonal staple is , in fact , an average of several `` squared path '' staples connecting the same end points .
the quark co - ordinates considered for y - shape paths are summarized in table [ tab : coord ] .
we note that the origin @xmath20 is not at the centre of these coordinates .
rather the coordinates are selected to place the quarks at approximately equal distances from each other . .
@xmath21 coordinates for the three quarks considered for y - shape paths .
the separation of the two quarks along a cartesian ( cart . ) direction are compared to the separation of those from the third quark . [ cols="^,^,^,^,^ " , ]
in this investigation we characterize the gluon field by the action density @xmath22 observed at spatial coordinate @xmath23 at euclidean time @xmath24 measured relative to the origin of the three - quark wilson loop .
we calculate the action density using the highly - improved @xmath25 three - loop improved lattice field - strength tensor @xcite on four - sweep ape - smeared gauge links . defining the quark positions as @xmath26 , @xmath27 and @xmath28 relative to the origin of the three - quark wilson loop , and denoting the euclidean time extent of the loop by @xmath29 , we evaluate the following correlation function @xmath30 where @xmath31 denotes averaging over configurations and lattice symmetries as described below .
this formula correlates the quark positions via the three - quark wilson loop with the gauge - field action in a gauge invariant manner . for fixed quark positions and euclidean time
, @xmath32 is a scalar field in three dimensions . for values of @xmath23
well away from the quark positions @xmath33 , there are no correlations and @xmath34 .
this measure has the advantage of being positive definite , eliminating any sign ambiguity on whether vacuum field fluctuations are enhanced or suppressed in the presence of static quarks .
we find that @xmath32 is generally less than 1 , signaling the expulsion of vacuum fluctuations from the interior of heavy - quark hadrons .
for this work we consider 200 quenched qcd gauge - fields using the @xmath35-mean - field improved luscher - weisz plaquette plus rectangle gauge action @xcite on @xmath36 lattices at @xmath37 , providing a lattice spacing of @xmath38 fm as set by the string tension . to improve the statistics of the simulation we use various symmetries of the lattice .
first , we make use of translational invariance by computing the correlation on every node of the lattice , averaging the results over the four - volume . to further improve the statistics , we use reflection symmetries . through reflection on the plane @xmath39
we can double the number of t and y - shaped wilson loops . by using reflections on both the plane @xmath39 and @xmath40
we can quadruple the number of l - shaped loops .
we finally use @xmath41 rotational symmetry about the @xmath42-axis to double the number of wilson loops .
this means we are using both the @xmath16 and @xmath43 planes as the planes containing the quarks .
for this exploratory study we present images for @xmath44 and note that qualitatively similar results are observed for @xmath45 .
we plan to examine this and other similarly conservative alternatives on our anticipated larger lattice volumes providing further statistics improvement .
figure [ ortho ] plots the correlation function of eq .
( [ correl ] ) , @xmath46 , for an ortho - slice in the plane of the three - quark system .
the colours of the ortho - slice , denoting the scalar values of @xmath46 , are rendered as a surface plot below .
away from the quark positions @xmath47 , but within the three - quark system @xmath48 indicating the suppression of qcd vacuum field fluctuations from the region inside heavy - quark hadrons . here
the quarks are approximately 0.75 fm from the centre of the distribution and are about 1.25 fm from each other .
figure [ tubes ] renders the spatial points of @xmath46 where the gluon action density is largely suppressed .
tube - like structures are revealed in y - shape as opposed to @xmath1 shape in accord with expectations from precision static quark - potential analyses .
we note that the centre of the flux tube is equidistant from the quark positions and not centered over the origin of the wilson loop .
this gives us confidence that sensitivity to the source is minimal .
however it is important to examine this issue in greater detail and this will be the subject of a forthcoming publication .
h. ichie , v. bornyakov , t. streuer and g. schierholz , nucl .
* b * ( proc . suppl . )
* 119 * ( 2003 ) 751 , hep - lat/0212036 .
f. okiharu and r. m. woloshyn , nucl .
* b * ( proc . suppl . ) * 129 * ( 2004 ) 745 , hep - lat/0310007 . s. o. bilson - thompson , d. b. leinweber and a. g. williams , annals phys .
* 304 * , 1 ( 2003 ) [ hep - lat/0203008 ] .
m. luscher and p. weisz , commun .
phys . * 97 * , 59 ( 1985 ) [ erratum - ibid . *
98 * , 433 ( 1985 ) ] . | methods for revealing the distribution of gluon fields within the three - quark static - baryon potential are presented . in particular , we outline methods for studying the sensitivity of the source on the emerging vacuum response for the three - quark system . at the same time , we explore the possibility of revealing gluon - field distributions in three - quark systems in qcd without the use of gauge - dependent smoothing techniques .
renderings of flux tubes from a preliminary high - statistics study on a @xmath0 lattice are presented . |
in this paper we consider the solution of complex symmetric shifted linear systems of the form @xmath1 where @xmath2 are nonsingular @xmath3-by-@xmath3 complex symmetric sparse matrices , i.e. , @xmath4 , with scalar shifts @xmath5 , @xmath6 is the @xmath3-by-@xmath3 identity matrix , and @xmath7 are complex vectors of length @xmath3 . the above systems
arise in large - scale electronic structure simulations @xcite , and there is a strong need for the fast solution of the systems . since the given shifted linear systems ( [ cs_sls ] ) are a set of sparse linear systems
, it is natural to use krylov subspace methods , and moreover since the coefficient matrices are complex symmetric , one of the simplest ways to solve the shifted linear systems is applying ( preconditioned ) krylov subspace methods for solving complex symmetric linear systems such as the cocg method @xcite , the cocr method @xcite , and the qmr_sym method @xcite to all of the shifted linear systems ( [ cs_sls ] ) . on the other hand , denoting the @xmath8-dimensional krylov subspace with respect to @xmath9 and @xmath10 as @xmath11 , we observe that @xmath12 this implies that once basis vectors are generated for one of the krylov subspaces @xmath13 , these basis vectors can be used to solve all the shifted linear systems .
in other words , there is no need to generate all krylov subspaces @xmath14 , and thus computational costs involving the basis generation , e.g. , matrix - vector multiplications , are saved . here
we give a concrete example : if we apply the conjugate orthogonal conjugate gradient ( cocg ) method to all the linear systems ( [ cs_sls ] ) , then bases for @xmath14 are generated for @xmath15 .
on the other hand , if we apply the cocg method to just one of the shifted linear systems ( [ cs_sls ] ) ( referred to as the `` @xmath16 '' ) , then the krylov basis vectors are generated from the seed system and these vectors are used to solve the rest of the shifted linear systems .
based on the observation ( [ ks ] ) , the shifted cocg method @xcite has been recently proposed for solving complex symmetric shifted linear systems .
the feature of the shifted cocg method is that the method performs cocg on a seed system and makes it possible to complete cocg for all shifted linear systems without further matrix - vector multiplications .
the feature is completely different from some of the other well - known shifted linear solvers such as the shifted bicgstab(@xmath17 ) method @xcite or the restarted shifted gmres method @xcite since these perform bicgstab(@xmath17 ) ( or gmres ) on a seed system but a different method on the shifted linear systems in order to keep the residuals colinear .
the feature of the shifted cocg method plays a very important role in the seed switching technique @xcite that avoids a minor problem of the shifted cocg method : one requires the choice of a seed system and an unsuitable choice may lead to the undesirable result that some shifted linear systems remain unsolved .
there is another approach to solving the shifted linear systems ( [ cs_sls ] ) . that is the use of krylov subspace methods for non - hermitian shifted linear systems such as the shifted bicgstab(@xmath17 ) method @xcite , the shifted ( tf)qmr method @xcite , the restarted shifted fom method @xcite , and the restarted shifted gmres method @xcite , see also , e.g. , @xcite .
we readily see that the relation ( [ ks ] ) holds not only for complex symmetric matrices but also for non - hermitian matrices , and these methods are based on the use of this shift - invariant relation .
therefore , this can be a good approach .
however , since these methods do not exploit the property of complex symmetric matrices , their computational costs can be more expensive than that of the shifted cocg method .
in this paper we consider the shifted qmr_sym method that is a special case of the qmr method for non - hermitian shifted linear systems @xcite and clarify the most time consuming part of it for a large number of shifted linear systems .
then , in order to reduce the cost , we propose a weighted quasi - minimal residual ( wqmr ) approach and propose a specific weight .
we experimentally show the practical efficiency of the resulting algorithm , referred to as shifted qmr_sym(@xmath0 ) , when the number of shifted linear systems is large enough .
the present paper is organized as follows : in the next section , shifted qmr_sym is described in order to specify the most time consuming part for a large number of shifted linear systems . in section 3 ,
we propose a wqmr approach with a specific weight for reducing the cost of the most time consuming part .
the resulting algorithm , shifted qmr_sym(@xmath0 ) , and its properties are given . in section 4 ,
some results of numerical examples from electronic structure simulations are shown to see the performance of the shifted qmr_sym(@xmath0 ) method .
finally , some concluding remarks are made in section 5 . throughout this paper ,
unless otherwise stated , all vectors and matrices are assumed to be complex .
@xmath18 , @xmath19 , @xmath20 denote the complex conjugate , transpose , and hermitian transpose matrix of the matrix @xmath21 .
@xmath22 denotes the @xmath23-norm written as @xmath24 , where @xmath23 is hermitian positive definite .
in this section , the shifted qmr_sym method and its properties for solving complex symmetric shifted linear systems are introduced . the qmr method for shifted linear systems
was first formulated in @xcite for the case of a general non - hermitian matrix .
therefore , by simplifying the non - hermitian lanczos process @xcite , as is known from other papers such as @xcite , a shifted simplified qmr method , shifted qmr_sym , is readily obtained for the case of a complex symmetric matrix . its algorithm is given next .
+ + @xmath25 & & { \hspace{-7mm}}{\mbox{\boldmath $ x$}}_{0}^{(\ell)}={\mbox{\boldmath $ p$}}_{-1}^{(\ell)}= { \mbox{\boldmath $ p$}}_{0}^{(\ell)}=\mathbf{0 } , \ { \mbox{\boldmath $ v$}}_{1}={\mbox{\boldmath $ b$}}/({\mbox{\boldmath $ b$}}^{\textrm{\scriptsize{t}}}{\mbox{\boldmath $ b$}})^{1/2},\ g_1^{(\ell)}=({\mbox{\boldmath $ b$}}^{\textrm{\scriptsize{t}}}{\mbox{\boldmath $ b$}})^{1/2},\\ & & { \hspace{-7mm}}{\textrm}{{\bf for}}\ n=1,2 , \ldots\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\quad { \textrm}{(the complex symmetric lanczos process)}\nonumber\\ & & { \hspace{-7mm}}\quad \alpha_n = { \mbox{\boldmath $ v$}}_n^{\textrm{\scriptsize{t}}}a{\mbox{\boldmath $ v$}}_n , \nonumber \\ & & { \hspace{-7mm}}\quad \widetilde{{\mbox{\boldmath $ v$}}}_{n+1}= a{\mbox{\boldmath $ v$}}_n-\alpha_n{\mbox{\boldmath $ v$}}_n-\beta_{n-1}{\mbox{\boldmath $ v$}}_{n-1 } , \nonumber \\ & & { \hspace{-7mm}}\quad \beta_{n } = ( \widetilde{{\mbox{\boldmath $ v$}}}_{n+1}^{\textrm{\scriptsize{t}}}\widetilde{{\mbox{\boldmath $ v$}}}_{n+1})^{1/2 } , \nonumber \\ & & { \hspace{-7mm}}\quad { \mbox{\boldmath $ v$}}_{n+1 } = \widetilde{{\mbox{\boldmath $ v$}}}_{n+1}/\beta_{n } , \nonumber \\ & & { \hspace{-7mm}}\quad t_{n-1,n}^{(\ell)}=\beta_{n-1},\ t_{n , n}^{(\ell ) } = \alpha_{n}+\sigma_{\ell},\ t_{n+1,n}^{(\ell)}=\beta_{n } , \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{(solve least squares problems by givens rotations ) } \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{{\bf for}}\ \ell=1,2 , \ldots , m\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\qquad { \textrm}{{\bf if}}\ \|{\mbox{\boldmath $ r$}}_{n}^{(\ell)}\|_2/\|{\mbox{\boldmath $ b$}}\|_2 > \epsilon,\ { \textrm}{{\bf then}}\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{{\bf for}}\ i=\max\{1,n-2\ } , \ldots , n-1\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\qquad \qquad \pmatrix { t_{i , n}^{(\ell ) } \cr t_{i+1,n}^{(\ell ) } } = \pmatrix { c_{i}^{(\ell ) } & s_{i}^{(\ell ) } \cr -\overline{s}_{i}^{(\ell ) } & c_{i}^{(\ell ) } } \pmatrix { t_{i , n}^{(\ell ) } \cr t_{i+1,n}^{(\ell ) } } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{{\bf end } } \nonumber\\ & & { \hspace{-7mm}}\qquad \quad c_n^{(\ell)}=\frac{|t_{n , n}^{(\ell)}|}{\sqrt{|t_{n , n}^{(\ell)}|^2+|t_{n+1,n}^{(\ell)}|^2 } } , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad \overline{s}_n^{(\ell)}=\frac{t_{n+1,n}^{(\ell)}}{t_{n , n}^{(\ell)}}c_n^{(\ell ) } , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad t_{n , n}^{(\ell ) } = c_n^{(\ell)}t_{n , n}^{(\ell)}+s_n^{(\ell)}t_{n+1,n}^{(\ell ) } , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad t_{n+1,n}^{(\ell)}=0 , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad \pmatrix { g_{n}^{(\ell ) } \cr g_{n+1}^{(\ell ) } } = \pmatrix { c_{n}^{(\ell ) } & s_{n}^{(\ell ) } \cr -\overline{s}_{n}^{(\ell ) } & c_{n}^{(\ell ) } } \pmatrix { g_{n}^{(\ell ) } \cr 0 } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{(update approximate solutions $ { \mbox{\boldmath $ x$}}_n^{(\ell)}$)}\nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \mbox{\boldmath $ p$}}_{n}^{(\ell)}= { \mbox{\boldmath $ v$}}_{n}-(t_{n-2,n}^{(\ell)}/t_{n-2,n-2}^{(\ell)}){\mbox{\boldmath $ p$}}_{n-2}^{(\ell)}- ( t_{n-1,n}^{(\ell)}/t_{n-1,n-1}^{(\ell)}){\mbox{\boldmath $ p$}}_{n-1}^{(\ell ) } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \mbox{\boldmath $ x$}}_{n}^{(\ell)}={\mbox{\boldmath $ x$}}_{n-1}^{(\ell)}+(g_n^{(\ell)}/t_{n , n}^{(\ell)}){\mbox{\boldmath
$ p$}}_n^{(\ell ) } , \nonumber \\ & & { \hspace{-7mm}}\quad \quad { \textrm}{{\bf end if } } \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{{\bf end } } \nonumber \\&&{\hspace{-7mm}}\qquad { \textrm}{{\bf if}}\ \|{\mbox{\boldmath $ r$}}_{n}^{(\ell)}\|_2/\|{\mbox{\boldmath $ b$}}\|_2\le \epsilon \ { \textrm}{for all}\ \ell,\ { \textrm}{{\bf then } exit}.\\ & & { \hspace{-7mm}}{\textrm}{{\bf end } } \nonumber \end{aligned}\ ] ] algorithm 2.1 can be regarded as a natural combination of the results given in @xcite . in order to know that the numerical solution is accurate enough , one may need to compute the residual 2-norms . in that case
, the following computation may be useful .
the 2-norms of the @xmath8th residuals of the approximate solutions @xmath26 of the shifted qmr_sym method are given by @xmath27 where @xmath28 and @xmath29 proposition 2.1 is a result known to hold for the qmr method @xcite .
therefore , it also holds for the above specialized variant .
the rest of this section describes some special properties of the shifted qmr_sym method .
let @xmath30 be real symmetric , @xmath31 be complex shifts , and @xmath32 .
then the shifted qmr_sym method ( algorithm 2.1 ) enjoys the following properties : * all matrix - vector multiplications can be done in real arithmetic ; * an approximate solution at @xmath8th iteration step for each @xmath17 has minimal residual 2-norms , i.e. , @xmath26 s are generated such that @xmath33 over @xmath34 ; * @xmath35 the above properties are known results since the properties have been proved for each individual shift .
see @xcite for details .
the properties of proposition 2.2 may be very useful for large - scale electronic structure simulations @xcite and a projection approach for eigenvalue problems @xcite since there are complex symmetric shifted linear systems to be solved efficiently under the assumption of proposition 2.2 .
in this section we consider complex symmetric shifted linear systems with a large number of shifts
. for such systems , say @xmath36 , the most time consuming part of algorithm 2.1 can be generating approximate solutions since the cost for the recurrences is @xmath37 per iteration step . in order to reduce the cost , in this section we propose a weighted quasi - minimal residual approach with a specific weight . in the next subsection
we discuss the details of the approach . in subsection 3.2
we give a specific weight to achieve the reduction of the cost . before the formulation of a weighted quasi - minimal residual ( wqmr ) approach ,
let us recall the complex symmetric lanczos process ( see , e.g. , algorithm 2.1 in @xcite ) .
+ + @xmath38 the matrix form of algorithm 3.1 is known as follows : let @xmath39 and @xmath40 be the @xmath41 and @xmath42 tridiagonal matrices whose entries are recurrence coefficients of the complex symmetric lanczos process , which are given by @xmath43 and let @xmath44 be the @xmath45 matrix with the lanczos vectors as columns , i.e. , @xmath46 . then from algorithm 3.1
, it follows that @xmath47 where @xmath48 .
now we are ready to describe the wqmr approach .
let @xmath26 be approximate solutions at the @xmath8th iteration step for the systems ( [ cs_sls ] ) , which are given by @xmath49 where @xmath50 .
then , from the definition of residual vectors @xmath51 , the update formulas ( [ sol ] ) , and the matrix form of the complex symmetric lanczos process ( [ comp_sym_lanczos_mat ] ) , we readily obtain @xmath52 here , @xmath53 is the first unit vector written by @xmath53=@xmath54 and @xmath55 .
it is natural to determine @xmath56 such that all residual 2-norms @xmath57 are minimized .
however , this choice for @xmath56 is impractical due to large computational costs .
hence , an alternative approach is given , i.e. , the vectors @xmath56 are determined by solving the following weighted least squares problems : @xmath58 where @xmath59 is an @xmath60-by-@xmath60 nonsingular matrix .
thus @xmath61 can be used as a weight since it is hermitian positive definite .
one of the simplest choices for @xmath59 is the identity matrix . in this case , from @xmath62 we have @xmath63 the vector that is minimized is called quasi - residual .
algorithm 2.1 is obtained by solving ( [ wlsp_2norm ] ) using givens rotations , see , e.g. , @xcite .
a slightly generalized choice proposed in @xcite is @xmath64 with @xmath65 for all @xmath66 .
then , it follows that @xmath67 of various possible choices for @xmath68 , a natural one is @xmath69 since @xmath70 contains then the diagonal entries of the upper triangular matrix @xmath71 that is obtained by the @xmath72 factorization of @xmath73 . if we choose @xmath74 , where @xmath75 , then from ( [ qmr_residual ] ) and ( [ wlsp ] ) we have @xmath76 by solving the above weighted least squares problems , all residual 2-norms are minimized .
hence @xmath77 is a rational choice . for each individual shift ,
the resulting algorithm is the same as that in ( * ? ? ?
* algorithm 3.2 ) . in the previous subsection
, we have described the wqmr approach and mentioned that the choice of the weight @xmath61 with @xmath62 leads to the shifted qmr_sym method ( algorithm 2.1 ) . under the assumption of proposition 2.2 ,
the shifted qmr_sym method is ideal in the sense of faber - manteuffel s theorem @xcite since it enjoys minimal residual property and requires not long - term but short - term recurrences for updating approximate solutions , and thus one may think that there is no need to choose other possible weights .
however , we will show in this subsection that even under the assumption of proposition 2.2 there is a practical weight for the wqmr approach .
the motivation for the choice of the weight mainly comes from the freedom of the number @xmath78 of complex symmetric shifted linear systems .
now we consider the computational costs of algorithm 2.1 for a large number of complex symmetric shifted linear systems , i.e. , @xmath79 . for the case
@xmath79 we readily see from algorithm 2.1 that computing the recurrences for updating approximate solutions is the most time - consuming part due to a cost of @xmath37 per iteration step .
hence we will now consider a weight to reduce the computational cost for the recurrences of @xmath26 . to achieve this
we propose the following choice : @xmath80 where @xmath81 is an @xmath8-by-@xmath8 upper bidiagonal matrix of the form @xmath82 and @xmath83 is lower triangular and will be specified below .
next we derive recurrence formulas for updating the approximate solutions @xmath26 . from ( [ wlsp ] ) with the choice @xmath84 of ( [ bidiagonal_weight ] ) it follows that @xmath85 from the above least squares problems we readily see that @xmath86 . hence it follows from ( [ sol ] ) and using @xmath87 that we have the following coupled two - term recurrence relations : @xmath88 the cost per iteration is now @xmath89 . substituting @xmath90 into the above recurrences , we have the even more efficient recurrence formulas .
@xmath91 by this reformulation , the cost becomes @xmath92 .
next , let us consider a choice for @xmath93 satisfying ( [ bidiagonal_weight ] ) .
let @xmath94 be an @xmath8-by-@xmath8 matrix for the form @xmath95 and let @xmath96 be tridiagonal .
then , by determining @xmath97 such that the @xmath98 entry of the matrix @xmath99 is zero , we can fulfill ( [ bidiagonal_weight ] ) in the following way : @xmath100 from the above we see that @xmath101 , and thus @xmath93 and @xmath102 are related by @xmath103 where @xmath104 from the above we readily see that the matrices @xmath93 are lower triangular with all 1 s on the diagonals , and this property will be used for the proof of proposition 3.1 given later .
the shifted qmr_sym method with the weight @xmath105 is referred to as shifted qmr_sym(@xmath0 ) .
its algorithm is given next .
+ + + @xmath25 & & { \hspace{-7mm}}{\mbox{\boldmath $ x$}}_{0}^{(\ell)}={\mbox{\boldmath $ p$}}_{-1}^{(\ell)}= { \mbox{\boldmath $ p$}}_{0}^{(\ell)}=\mathbf{0 } , \ { \mbox{\boldmath $ v$}}_{1}={\mbox{\boldmath $ b$}}/({\mbox{\boldmath $ b$}}^{\textrm{\scriptsize{t}}}{\mbox{\boldmath $ b$}})^{1/2},\ \widetilde{g}_1^{(\ell)}=({\mbox{\boldmath $ b$}}^{\textrm{\scriptsize{t}}}{\mbox{\boldmath $ b$}})^{1/2},\\ & & { \hspace{-7mm}}{\textrm}{{\bf for}}\ n=1,2 , \ldots\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\quad { \textrm}{(the complex symmetric lanczos process)}\nonumber\\ & & { \hspace{-7mm}}\quad \alpha_n = { \mbox{\boldmath $ v$}}_n^{\textrm{\scriptsize{t}}}a{\mbox{\boldmath $ v$}}_n , \nonumber \\ & & { \hspace{-7mm}}\quad \widetilde{{\mbox{\boldmath $ v$}}}_{n+1}= a{\mbox{\boldmath $ v$}}_n-\alpha_n{\mbox{\boldmath $ v$}}_n-\beta_{n-1}{\mbox{\boldmath $ v$}}_{n-1 } , \nonumber \\ & & { \hspace{-7mm}}\quad \beta_{n } = ( \widetilde{{\mbox{\boldmath $ v$}}}_{n+1}^{\textrm{\scriptsize{t}}}\widetilde{{\mbox{\boldmath $ v$}}}_{n+1})^{1/2 } , \nonumber \\ & & { \hspace{-7mm}}\quad { \mbox{\boldmath $ v$}}_{n+1 } =
\widetilde{{\mbox{\boldmath $ v$}}}_{n+1}/\beta_{n } , \nonumber \\ & & { \hspace{-7mm}}\quad t_{n-1,n}^{(\ell)}=\beta_{n-1},\ t_{n , n}^{(\ell ) } = \alpha_{n}+\sigma_{\ell},\ t_{n+1,n}^{(\ell)}=\beta_{n } , \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{(solve weighted least squares problems ) } \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{{\bf for}}\ \ell=1,2 , \ldots , m\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\qquad { \textrm}{{\bf if}}\ \|{\mbox{\boldmath $ r$}}_{n}^{(\ell)}\|_2/\|{\mbox{\boldmath $ b$}}\|_2 > \epsilon,\ { \textrm}{{\bf then}}\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{{\bf for}}\ i=\max\{1,n-1\ } , \ldots , n-1\ { \textrm}{{\bf do:}}\\ & & { \hspace{-7mm}}\qquad \qquad t_{i+1,n}^{(\ell ) } = f_{i}^{(\ell ) } t_{i , n}^{(\ell ) } + t_{i+1,n}^{(\ell ) } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{{\bf end } } \nonumber\\ & & { \hspace{-7mm}}\qquad \quad f_n^{(\ell)}=-\frac{t_{n+1,n}^{(\ell)}}{t_{n , n}^{(\ell ) } } , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad t_{n+1,n}^{(\ell)}=0 , \nonumber \\ & & { \hspace{-7mm}}\qquad \quad \widetilde{g}_{n+1}^{(\ell ) } = f_{n}^{(\ell)}\widetilde{g}_{n}^{(\ell ) } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \textrm}{(update approximate solutions $ { \mbox{\boldmath $ x$}}_n^{(\ell)}$)}\nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \mbox{\boldmath $ p$}}_{n}^{(\ell)}= { \mbox{\boldmath $ v$}}_{n}-(t_{n-1,n}^{(\ell)}/t_{n-1,n-1}^{(\ell)}){\mbox{\boldmath $ p$}}_{n-1}^{(\ell ) } , \nonumber\\ & & { \hspace{-7mm}}\qquad \quad { \mbox{\boldmath $ x$}}_{n}^{(\ell)}={\mbox{\boldmath $ x$}}_{n-1}^{(\ell)}+(\widetilde{g}_n^{(\ell)}/t_{n , n}^{(\ell)}){\mbox{\boldmath
$ p$}}_n^{(\ell ) } , \nonumber \\ & & { \hspace{-7mm}}\quad \quad { \textrm}{{\bf end if } } \nonumber \\ & & { \hspace{-7mm}}\quad { \textrm}{{\bf end } } \nonumber \\&&{\hspace{-7mm}}\qquad { \textrm}{{\bf if}}\ \|{\mbox{\boldmath $ r$}}_{n}^{(\ell)}\|_2 / \|{\mbox{\boldmath $ b$}}\|_2\le \epsilon \
{ \textrm}{for all}\ \ell,\ { \textrm}{{\bf then } exit}.\\ & & { \hspace{-7mm}}{\textrm}{{\bf end } } \nonumber \end{aligned}\ ] ] similar to proposition 2.1 , there is an efficient way to evaluate residual 2-norms as follows : the @xmath8th residual 2-norms of the approximate solutions @xmath26 for the shifted qmr_sym(@xmath0 ) method are given by @xmath106 _ proof_. the proof is similar to that of proposition 2.1 .
it follows from ( [ qmr_residual ] ) , ( [ bidiagonal_weight ] ) , ( [ wlsp_b ] ) , and recalling @xmath86 that we have @xmath107 from ( [ f ] ) and ( [ fn ] ) @xmath93 is a lower triangular matrix with all 1 s on the diagonals .
thus , @xmath108 is also a lower triangular matrix with all 1 s on the diagonals .
it follows that @xmath109 , and thus we have @xmath110 , which concludes the proof .
when we solve a large number of shifted systems , the computational cost of the residual 2-norms for the shifted qmr_sym(@xmath0 ) method is much cheaper than the that for the shifted qmr_sym method since the former cost is of order @xmath3 and the latter is of order @xmath111 per iteration step . observing algorithms of the shifted qmr_sym and the shifted qmr_sym(@xmath0 ) methods
, we see that the work for the weighted least squares problems and for updating approximate solutions in the shifted qmr_sym(@xmath0 ) method is lower than that in the shifted qmr_sym method . for the case
@xmath36 , the most time - consuming part is updating the approximate solutions . in this case , the shifted qmr_sym(@xmath0 ) method may be more efficient than the shifted qmr_sym method since the shifted qmr_sym(@xmath0 ) method requires @xmath92 operations per iteration step for the update while the shifted qmr_sym method needs @xmath37 operations
. this difference will be clearer when we use the results of propositions 2.1 and 3.1 as stopping criterion .
on the other hand , in terms of number of iterations , under a certain assumption , the convergence of the shifted qmr_sym(@xmath0 ) method is worse than that of the shifted qmr_sym method but not worse than that of the shifted cocg method , as described by the following proposition : under the assumption of proposition 2.2 , the shifted qmr_sy + m(@xmath0 ) method ( algorithm 3.2 ) enjoys the following properties : * all matrix - vector multiplications can be done in real arithmetic ; * if breakdown does not occur and each matrix @xmath112is nonsingular , then @xmath113 , where the superscripts sq(@xmath0 ) , scocg , and sq are short for shifted qmr_sym(@xmath0 ) , shifted cocg , and shifted qmr_sym , respectively ; * @xmath114 _ proof_. the proof of ( @xmath6 ) is the same as that of proposition 2.2 , and is based on the fact that under the assumption the complex symmetric lanczos process generates real basis vectors .
next , we give a proof of ( @xmath115 ) .
the @xmath8th residuals of the shifted cocg method @xcite belong to @xmath116 .
hence , each @xmath117 can be written as @xmath118 , where @xmath44 is the same matrix as in ( [ comp_sym_lanczos_mat ] ) since each @xmath117 is orthogonal to each subspace @xmath119\{=@xmath120 , i.e. , @xmath121 , we have @xmath122 , and thus it follows from ( [ comp_sym_lanczos_mat ] ) that we have the relation @xmath123 , where @xmath124 .
since the shifted qmr_sym(@xmath0 ) method has the form ( [ qmr_residual ] ) , it is sufficient to show that @xmath125 . from ( [ wlsp_b ] ) and ( [ relation_l ] ) it follows that @xmath126l_{n+1}^{(\ell)}[{\mbox{\boldmath $ e$}}_1^{\textrm{\scriptsize{t}}}| 0]^{\textrm{\scriptsize{t}}}=g_1(b_n^{(\ell)})^{-1}l_{n}^{(\ell)}{\mbox{\boldmath $ e$}}_1=g_1\{(l_{n}^{(\ell)})^{-1}b_n^{(\ell)}\}^{-1 } { \mbox{\boldmath $ e$}}_1.$ ] since from ( [ fn ] ) and ( [ relation_l ] ) we can readily confirm the relation @xmath127
, we have @xmath128 which is the same as @xmath129 the inequality in ( @xmath115 ) follows from proposition 2.2 since under the given assumption the shifted qmr_sym method enjoys the minimal residual property .
finally , we give a proof of ( @xmath130 ) .
if follows from the proof of ( @xmath6 ) that @xmath131 for all @xmath66 .
thus from proposition 3.1 we have @xmath132 in the property ( @xmath133 ) of proposition 3.2 , breakdown may occur due to the choice ( [ f ] ) of the weighted least squares problems . from proposition 3.2
we see that in terms of the number of iteration steps the shifted qmr_sym(@xmath0 ) method never converges faster than the shifted qmr_sym method but it converges at the same iteration step as the shifted cocg method does .
since the efficiency of the shifted cocg method has been shown already and for the case @xmath36 the computational cost of the shifted qmr_sym@xmath134 method is much less than that of the shifted qmr_sym method , the shifted qmr_sym@xmath134 method can also be useful .
this will be supported by some numerical examples in the next section .
in this section , we report on some numerical examples with the shifted cocg method , the shifted qmr_sym method ( algorithm 2.1 ) , and the shifted qmr_sym(@xmath0 ) method ( algorithm 3.2 ) .
we evaluate these methods in terms of computational time .
all tests were performed on a workstation with a 2.6ghz amd opteron(tm ) processor 252 using double precision arithmetic .
codes were written in fortran 77 and compiled with g77 -@xmath1353 . in all cases
the stopping criterion was set as @xmath136 .
+ + 3.1 .
example 1 + + the first problem comes from the electronic structure calculation of a bulk si(001 ) with 512 atoms in @xcite , which is written as follows : @xmath137 where @xmath138 , @xmath139 is a symmetric matrix with 139264 entries , @xmath140 , and @xmath141 .
since the shifted cocg method requires the choice of a seed system , we have chosen the optimal seed ( @xmath142 ) in this problem ; otherwise some linear systems will remain unsolved . figure 4.1 shows the number of iterations of each method to solve the @xmath17th shifted linear systems .
for example , in fig .
4.1 , the number of iterations for the shifted cocg method at @xmath143 is 150 , which means the shifted cocg method required 150 iterations to obtain the ( approximate ) solutions of the 600th shifted linear system , i.e. , @xmath144 . )
method versus the serial number of the shifted linear systems . ] of the relative residual 2-norms versus the number of iterations of the shifted cocg method , the shifted qmr_sym method , and the shifted qmr_sym(@xmath0 ) method for the shifted linear system with @xmath145 . ] from fig .
4.1 we make three observations : first , the three methods required almost the same number of iterations at each @xmath17 ; second , in terms of number of iterations , the shifted qmr_sym method often converged slightly faster than the other two methods .
this phenomenon is closely related to proposition 2.2 , which will become clearer later ; third , for each method the required number of iterations depends highly on the shift parameters @xmath146 .
this result may come from varying eigenvalues of the coefficient matrices @xmath147 since if we choose @xmath146 close to an eigenvalue of @xmath148 , then @xmath147 is close to singular .
conversely , from the shape of fig .
4.1 one may obtain a partial distribution of eigenvalues of @xmath148 .
one of the residual 2-norm histories for the three methods is given in fig .
4.2 . from fig .
4.2 we see that the relative residual 2-norm of the shifted qmr_sym method decreases monotonically and at every iteration step the norm is less than those of the other two methods .
hence we can say that the property @xmath149 of proposition 2.2 is experimentally supported by this history .
we make another observation in the histories of the shifted cocg method and the shifted qmr_sym(@xmath0 ) method : during the first about fifty iterations , the two methods show the same histories . after that
their histories varies gradually .
hence we see that the property @xmath149 of proposition 3.2 is also experimentally supported by these histories .
each computational time of the three methods is given in fig .
4.3 , where the horizontal axis denotes the number @xmath78 of shifted linear systems that are solved .
for example , in fig .
4.3 , the computational time of the shifted cocg method at @xmath150 is about 0.76 [ sec . ] , which means that it required about 0.76 [ sec . ] to solve the shifted linear systems : @xmath151 . from fig .
4.3 we see that as the number @xmath78 grows larger , the shifted qmr_sym method required more cpu time than the other two methods . on the other hand , the shifted qmr_sym(@xmath0 )
method required almost the same cpu time as the shifted cocg method .
this phenomena can be attributed to the computational costs of updating approximate solutions for each method since there are three facts : first , we know from fig .
4.1 that the three methods required almost the same iterations ; second , the shifted qmr_sym(@xmath0 ) method requires almost the same computational costs as the shifted cocg method , while the shifted qmr_sym method tends to require more computational costs per iteration than the other two methods ; third , for large @xmath78 , updating approximate solutions is one of the most time - consuming parts . the second and the third facts have been discussed already in the previous two sections . in fig .
4.3 we can see little about the properties of the three methods for small @xmath17 .
we therefore show the ratio of each computational time to the computational time of the shifted cocg method in fig .
we see from fig .
4.4 that in terms of the ratio of cpu times the shifted qmr_sym method and the shifted qmr_sym(@xmath0 ) method converged much faster than the shifted cocg method when the number of shifted linear systems is small , say , @xmath152 .
this can be explained in the following way : for small @xmath78 , updating approximate solutions does not affect the cpu time so much .
other operations such as matrix - vector multiplications are now the most time - consuming parts since the three methods required almost the same number of iterations , see fig . 4.1 . from proposition 2.2
@xmath153 and proposition 3.2 @xmath153 we know that in this case the shifted qmr_sym method and the shifted qmr_sym(@xmath0 ) method require real matrix - real vector multiplications . on the other hand ,
the shifted cocg method requires real matrix - complex vector multiplications .
moreover , dot products and vector additions of the complex symmetric lanczos process used in the shifted qmr_sym method and the shifted qmr_sym(@xmath0 ) method can be done in real arithmetic .
hence , the two methods converged much faster than the shifted cocg method .
+ + 3.2 .
example 2 + + the second problem comes from the electronic structure calculation of bulk fcc cu with 1568 atoms in @xcite and is given as follows : @xmath154 where @xmath155 , @xmath156 is a symmetric matrix with 3924704 entries , @xmath140 , and @xmath157 .
each computational time of the three methods for solving the @xmath78 shifted linear systems is shown in fig .
4.5 , where horizontal and vertical axes are the same as those used in fig 4.3 .
the ratio of each computational time to the computational time of the shifted cocg method is shown in fig .
4.6 . from figs .
4.5 , 4.6 we see that the size of this matrix is about 7 times larger than before , and the three methods show similar tendencies to those seen in the previous example .
in this paper , the shifted qmr_sym method was described as a specialization of the qmr method for general non - hermitian shifted linear systems @xcite .
the method has an advantage over the shifted cocg method in that there is no need to choose a suitable seed system .
on the other hand , we have found that for a large number of shifted linear systems , the most time - consuming part of the shifted qmr_sym method is updating approximate solutions , and this cost is higher than that of the shifted cocg method .
we therefore have proposed the weighted quasi - minimal residual approach with a specific weight for reducing the computational cost for updating approximate solutions .
the resulting method , shifted qmr_sym(@xmath0 ) , also does not require to choose a suitable seed system , which is an advantage over the shifted cocg method . from numerical experiments
we have learned that shifted qmr_sym and qmr_sym(@xmath0 ) are competitive in comparison to the shifted cocg method . in particular , qmr_sym(@xmath0 ) can be the method of choice for solving complex symmetric shifted linear systems with a large number of shifts that arise from large - scale electronic structure theory . in future work , numerical tests for general complex
symmetric shifted linear systems will be done to evaluate the performance of the method .
we wish to express our gratitude to roland w. freund ( uc davis ) for his fruitful comments during the conference at harrachov in 2007 .
we are grateful to an anonymous referee for useful comments that substantially enhanced the quality of this paper .
finally , we would like to thank martin h. gutknecht ( eth zurich ) for careful reading of the manuscript .
+ + , _ a numerical method for calculating the green s function arising from electronic structure theory _ , in : frontiers of computational science , eds .
y. kaneda , h. kawamura and m. sasai , springer - verlag , berlin / heidelberg , 2007 , pp .
189 - 195 . | we consider the solution of complex symmetric shifted linear systems .
such systems arise in large - scale electronic structure simulations and there is a strong need for the fast solution of the systems . with the aim of solving the systems efficiently , we consider a special case of the qmr method for non - hermitian shifted linear systems and propose its weighted quasi - minimal residual approach . a numerical algorithm , referred to as shifted qmr_sym(@xmath0 ) ,
is given by the choice of a particularly cost - effective weight .
numerical examples are presented to show the performance of the shifted qmr_sym(@xmath0 ) method . |
seyfert galaxies are relatively nearby ( @xmath2 0.1 ) , moderate - luminosity ( @xmath3 @xmath4 10@xmath5 10@xmath6 erg s@xmath7 ) active galaxies . the large - scale morphologies of these galaxies typically resemble those of normal ( i.e. , inactive ) spiral galaxies , although there are some examples of peculiar or interacting systems ( de robertis , hayhoe , & yee 1998 ) .
optical and uv spectra of their nuclei are characterized by strong atomic emission lines .
based on the widths of these lines , seyfert galaxies are generally divided into two types ( khachikian & weedman 1974 ) : type 1s possess broad permitted lines with full - width at half - maximum ( fwhm ) typically @xmath8 1000 km s@xmath7 and narrow forbidden lines with fwhm @xmath9 500 km s@xmath7 , while the spectra of type 2s show narrow permitted and forbidden lines . those seyferts that show both broad and narrow components in their permitted lines are classified as intermediate types , ranging from 1.2 to 1.9 ( osterbrock , 1977 ; 1981 ) .
osterbrock & pogge ( 1985 ) discovered a class of active galactic nuclei ( agn ) with relatively narrow ( fwhm @xmath10 2000 km s@xmath7 ) permitted lines , like seyfert 2 galaxies , but with emission - ratios that indicate the presence of a high - density region , like the broad - line region ( blr ) in seyfert 1 galaxies .
they dubbed these objects `` narrow - line seyfert 1s '' ( nls1s ) .
as discussed by mathur ( 2000a , 2000b ) , nls1s possess a number of interesting spectral properties .
they show strong fe ii and weak [ o iii ] @xmath115007 emission relative to h@xmath1 , which puts them at one extreme of eigenvector 1 of boroson & green ( 1992 ) , which was determined from principal component analysis of a large sample of low - redshift quasars and seyfert galaxies .
compared to broad - line seyfert 1s ( bls1s ) , the x - ray continua of nls1s are characterized by steep slopes and rapid variability , particularly in the soft ( @xmath12 1 kev ) band ( boller , brandt , & fink 1996 ; brandt , mathur , & elvis 1997 ) .
wills et al .
( 1999 ) found unsually high n v @xmath111240/ c iv @xmath111550 flux ratios in nls1s , which may be indicative of high nitrogen abundances , as would result from metal enrichment due to a recent burst of star formation ( see also shemmer & netzer 2002 ) .
the mid - ir brightness of a number of nls1s provides further evidence of recent or current star formation ( moran , halpern , & helfand 1996 ) .
there are several models to explain the unusual properties of nls1s , and in particular the narrowness of their line profiles .
wandel & peterson ( 1999 ) suggested that the blrs may be relatively farther away from the central continuum source in nls1s , due to over - ionization of the gas close to the active nucleus .
however , based on estimates of the blr sizes from reverberation mapping , peterson et al .
( 2000 ) argue this can not generally be the case .
another possibility is that the blr is flattened and viewed roughly face - on , with narrow line profiles as a result , although peterson et al
. found no compelling evidence for such a preferred viewing angle unless the narrow emission - line regions are mis - aligned with the accretion - disk / black hole systems in these objects .
the current widely - accepted paradigm is that , while agn in general are powered by accretion of material onto a supermassive central black hole , nls1s possess black holes of relatively modest mass ( @xmath12 10@xmath13 m@xmath14 ) , that are accreting matter at or above their eddington limits ( pounds , done , & osborne 1995 ) .
this view is supported by recent observational results that indicate that nls1s possess significantly smaller black - hole mass to bulge mass ratios than their broad - line counterparts ( mathur , kuraszkiewicz , & czerny 2001 ; wandel 2002 ) . in this scenario ,
the narrow line widths are simply due to clouds in motion around the small - mass black hole , while the steep soft x - ray continuum is the high - energy tail of the `` big blue bump '' , which is presumably emission from the accretion disk that peaks in the extreme ultraviolet ( euv ) .
nls1s could therefore be analogous to the `` strong soft '' states of galactic black holes at high accretion rates ( pounds et al .
1995 ) . due to the high accretion rates , the disks in nls1s are likely much hotter than those in bls1s , and hence the emission
is peaked at higher energies .
wang & netzer ( 2003 ) showed that high - accretion rates result in an extremely thin disk which emits a double - peaked continuum , with a peak in the euv / soft x - ray due to the disk itself , and a high energy peak due to a hot corona . based on this model , they suggest that most of the 54 nls1s in the vron - cetty , vron , & gonalves ( 2001 ) sample are super - eddington accretors . assuming that nls1s are
high - rate accretors , and considering the other evidence of extreme activity , mathur ( 2000a , 2000b ) suggested that nls1s are in an early phase of activity and may be the low - redshift analogs of high-@xmath15 qsos ( however , see constantin & shields 2003 ) . in order to fuel seyfert galaxies and other agn
, material must be transported from the host galaxy into the inner nucleus , requiring some means by which the material loses angular momentum .
one possibility is interactions or mergers with other nearby galaxies ( toomre & toomre 1972 ; adams 1977 ) .
however recent studies indicate that seyfert galaxies do not have more close companions than do normal galaxies , and , furthermore , a substantial fraction of seyfert galaxies show no evidence for recent mergers ( de robertis , yee , & hayhoe 1998 , and references therein ) . another process for fueling agn that has received considerable
attention is gas inflow along a stellar bar ( simkin , su , & schwarz 1980 ) .
theoretical studies show that the gas can lose its angular momentum after encountering the gravitational potential of a bar and be transported inward ( shlosman , begelman , & frank 1989 ) .
however , most observational studies in this area have found similar fractions of bars for seyferts and normal galaxies ( heckman 1980 ; simkin et al .
1980 ; ho , filippenko , & sargent 1997 ; mulchaey , regan , & kundu 1997 ; mulchaey & regan 1997 ) , although knappen , shlosman , & peletier ( 2000 ) found a slightly higher percentage of ir bars in seyferts ( 80% ) compared to normal galaxies ( 60% ) .
interestingly , the percentages of seyfert or normal galaxies that have bars vary substantially among these studies , from @xmath030% to @xmath080% .
these discrepancies are probably due to the sample selection ( e.g. , luminous vs. average spirals ) , wave band ( ir observations tend to reveal more bars than optical images , mulchaey & regan 1997 ) , and identification criteria ( e.g. , whether or not to include weak bars ) . nevertheless , well - defined samples that use consistent criteria nearly always find equal percentages of bars in seyfert and normal galaxies ( with the exception noted above ) . since there is strong evidence that most normal galaxies contain inactive supermassive black holes ( kormendy & richstone 1995 ) , the above finding indicates that while bars may be important for driving gas into the inner regions , there are other factors that contribute to the presence of nuclear activity .
dynamical models indicate that a bar potential can drive material to within @xmath01 kpc from the agn , at which point the transport will be halted at the inner lindblad resonance and the infalling gas will form a disk ( shlosman , et al .
if global nonaxisymmetrical instabilities are present , a secondary , gaseous bar can develop at small distances from the nucleus ( @xmath91 kpc ) , and provide a means to drive the gas in towards the active nucleus ( shlosman , et al .
1989 ; heller & shlosman 1994 ) .
in fact , maiolino et al .
( 2000 ) found evidence for gas motion along a secondary bar in the nucleus of the seyfert 2 circinus galaxy .
however , it appears that secondary gas bars are fairly rare among seyfert galaxies .
the most efficient way to detect these bars is through the extinction or reddening caused by their embedded dust . _
hubble space telescope _
( _ hst _ ) ir and optical images indicate that only 10 20% of seyfert galaxies show nuclear ( at distances @xmath91 kpc from the nucleus ) dust bars ( regan & mulchaey 1999 ; martini & pogge 1999 ; pogge & martini 2002 ) . instead
, the _ hst _ images reveal that most ( @xmath080% ) seyfert galaxies show nuclear dust spirals ( regan & mulchaey 1999 ; pogge & martini 2002 ; martini et al .
although these structures are common to both barred and unbarred seyferts , all of the `` grand design '' nuclear spirals ( so - called by their resemblance to the large scale spiral structure of the same name ) are found in barred galaxies ( martini et al .
the grand design spirals appear to connect to dust lanes on the leading edge of the large - scale stellar bars , suggesting a connection to shock fronts in the bars . however , martini et al .
( 2003 ) find that dust spirals in seyfert galaxies are not statistically more numerous than those in normal galaxies .
thus , it is unclear exactly how the dust spirals fuel the agn and what mechanism(s ) actually control the onset of nuclear activity .
if the high accretion - rate paradigm for nls1s is correct , it suggests that the fueling of the agn is more efficient in these seyferts than in their bls1 counterparts .
if so , perhaps there is a connection between large - scale properties and the high accretion rates in nls1s .
krongold , dultzin - hacyan , and marziani ( 2001 ) examined host galaxies and environments , and found no statistical difference between the frequency of companion galaxies in nls1s and bls1s .
another possibility is a difference in their large - scale morphologies . in this paper , we provide evidence that , stellar bars are much more common in nls1s than bls1s .
these large - scale bars , which typically begin at @xmath01 kpc from the nucleus and extend to 5 10 kpc , represent an efficient means for transporting large amounts of gas to the inner regions , which can presumably support the high accretion rates in the nuclei of nls1s .
our sample contains primarily the _ hst _ broad - band images of seyfert 1 galaxies obtained by malkan , gorjian , & tam ( 1998 ) .
this is a uniform sample of 91 seyfert 1 galaxies at z @xmath12 0.035 observed with the wide field planetary camera 2 ( wfpc2 ) through the f606w filter ( centered near 6000 ) .
single exposures of duration 500 s were obtained with nearly all the seyfert nuclei located on the planetary camera ( pc ) chip , yielding a spatial resolution of 01 .
the number of nls1s in this sample is small ( 13 ) ; to increase that number , we searched the _ hst _ archives for nls1s in the vron - cetty et al .
( 2001 ) sample that have wfpc2 broad - band optical images .
we found 6 additional nls1s , which extend to higher redshifts ( up to z @xmath4 0.084 ) ; the effects of including them to make a larger , more heterogeneous sample ( in terms of redshift ) will be discussed .
we retrieved all of the wfpc2 images from the _ hst _ archives , to display a larger field of view ( 37@xmath16 x 37@xmath16 , the projected size of the pc ) than the images published ( mostly 9@xmath16 x 9@xmath16 ) in malkan et al .
the malkan et al .
sample consists of single images of each seyfert galaxy , and we removed only point - like cosmic - ray hits using automated routines .
the remaining galaxies had multiple images , and we removed both point - like and glancing hits through intercomparison of the images . in practice , residual or remaining cosmic ray hits , saturation of the nuclei , or other defects were easily recognized in these images and had little effect on our assessment of the large - scale morphology of the host galaxies . to ensure objectivity , we scrambled the order of the _ hst _ images and the three coauthors independently classified their morphologies without knowledge of seyfert 1 subclass ( nls1 or bls1 ) or the name of the galaxy .
we classified the galaxies into one of 6 major groups : s ( spiral ) , sb ( spiral with a noticeable bar ) , e ( elliptical ) , i ( irregular ) , p ( point - souce , only the active nucleus is visible ) , or ? ( uncertain ) ( note that we did not attempt to assign hubble subclasses ) .
for each galaxy , at least two of the three independent classifications agreed ( in most cases all three agreed ) , and we adopted the majority view .
a large majority of the galaxies in the sample are at redshifts z @xmath4 0.01 0.035 , which yield projected distances of 7 25 kpc across the pc for h@xmath17 @xmath4 75 km s@xmath7 mpc@xmath7 . in the malkan sample , there are 11 out of 91 galaxies at smaller redshifts , extending down to z @xmath4 0.002 ( 1.4 kpc across the pc ) . for these galaxies in particular , we also examined the full wfpc2 images to assess their large - scale morphologies ( each wf chip has a field of view of 13 x 13 ) .
the six additional galaxies from vron - cetty et al .
( 2001 ) have redshifts up to 0.084 , corresponding to 60 kpc across the pc . in table 1 , we give the results of our classifications .
the galaxies are first listed in the same order as in malkan et al .
( 1998 ) , and are followed by the six additional nls1 galaxies from vron - cetty et al .
we also list the redshifts , our morphological classifications ( `` ckg '' ) , and those based on the same images from malkan et al .
( `` mgt '' ) , which include hubble subclasses .
the classes of seyfert nuclei ( bls1 or nls1 ) are based on the listings in the catalog of vron - cetty & vron ( 2001 ) ; the bls1s include seyferts from types 1 to 1.9 . in the last two columns , we give measurements of the full - width at half - maximum ( fwhm ) of the broad component of h@xmath1 ( or in a few cases , h@xmath18 ) , and the references that these measurements were obtained from .
note that 35 seyfert galaxies in the sample have previously been classified as nls1s or bls1s on the basis of their optical spectra , but measurements of the fwhms of the broad lines have not been published .
we note that our general morphological classes agree well with those of malkan et al . of the 91 galaxies , there are only 11 discrepancies , and of the 74 galaxies that we both identified as spirals , there is disagreement on the presence or absence of a bar in only 6 galaxies ( 5 bls1s , 1 nls1 ) . to illustrate our classifications , we show the _ hst _ images of the nls1s in our sample in figures 1 and 2 .
ngc 4051 is not shown , since at its very low redshift ( z @xmath4 0.0023 ) the pc only spans 1.6 x 1.6 kpc .
examination of the palomar sky survey and other ground - based images ( tully et al .
1996 ) shows a strong bar that can be seen extending into the larger wfpc2 frame .
we consider the 92 seyfert galaxies in table 1 with definite morphological classifications ( i.e. , ignoring the `` p '' and `` ? '' classes ) . from this group , we have classified 91% as spirals , and of the spirals , we have classified 33% of the spirals as barred .
the latter is consistent with percentages from studies that have identified strong bars in optical images of both active and normal spiral galaxies ( mulchaey et al .
1997 ; malkan et al .
1998 ; see also 1 ) . however , it is obvious from table 1 that most of the nls1 galaxies are barred spirals , and most of the bls1 galaxies are normal spirals .
quantitatively , if we consider the 84 spirals in this sample , then 65% ( 11/17 ) of the nls1 spirals have bars , whereas only 25% ( 17/67 ) of the bls1 spirals have bars .
if we consider only the malkan et al .
sample , then 64% ( 7/11 ) of the nls1 spirals have bars , which is essentially identical to the result for the full sample .
if we use malkan et al.s ( 1998 ) classifications , then 30% ( 20/67 ) of the bls1 spirals show bars and 55% ( 6/11 ) of the nls1 spirals show bars . to further characterize the trend of bars in nls1 galaxies , figure 3 shows histograms of the fraction of spirals that have bars as a function of the fwhm ( h@xmath1 ) on a log scale .
the shaded regions show a clear trend of decreasing fraction of bars with increasing fwhm for the entire sample .
the trend continues through the three bins associated with nls1s ( fwhm @xmath10 2000 km s@xmath7 ) to the first bin for the bls1s .
the fraction levels out thereafter , and is essentially constant for all bls1s regardless of fhwm .
interestingly , the four nls1 spirals in the full sample with the smallest fwhm ( 600 1000 km s@xmath7 ) all show bars .
the hatched region in figure 3 shows that the same trend occurs for the more uniform malkan et al .
subsample ; there is only a slightly lower fraction of bars in nls1s with widths in the range 1000 2000 km s@xmath7 compared to the full sample .
it is clear from table 1 that the number of nls1s in each bin is quite small ( as low as 3 ) for this subsample , and a larger , more uniform sample of images for both bls1s and nls1s would be helpful for testing this trend .
our analysis of _ hst_/wfpc2 images reveals that stellar bars are more common in nls1s , particularly those with very narrow h@xmath1 profiles , compared to bls1s . this suggests a link between the high fueling rates that can be provided by the bars and the high accretion rates commonly associated with nls1s .
these large - scale stellar bars often extend down to 1 kpc from the nuclei of these galaxies . inside a radius of @xmath01 kpc ,
nuclear dust spirals , and in a few cases dust bars , are likely to be directly responsible for transporting fuel to the active nucleus .
however , the exact connection between large - scale bars , inner dust spirals and bars , and accretion rates in the nucleus is unclear .
we outline one possible scenario below . based on a variety of evidence ,
it has been suggested that nls1s are in an early stage of activity ( mathur 2000a , 2000b ) .
if so , the presence of bars suggests the following evolutionary sequence for galaxies that become active .
1 ) initially , the galaxy is a normal spiral , albeit with a small black hole mass m@xmath19 ( @xmath10 10@xmath13 m@xmath14 ) . although studies of nearby inactive galaxies find few with m@xmath19 this small ( magorrian et al .
1998 ; merritt & ferrarese 2001 ; kormendy & gebhardt 2001 ) , wandel ( 2002 ) suggested that the sample derived from stellar dynamics is biased towards detecting higher mass m@xmath19 , since detecting smaller mass sources requires higher spatial resolution , which is only possible in the most nearby galaxies .
hence there may be a large , undetected population of small m@xmath19 in inactive galaxies .
2 ) due to either a tidal disruption or simply an asymmetrical distribution of mass in the inner disk of the galaxy , a stellar bar forms within a few x 10@xmath20 yrs ( combes & elmegreen 1993 ) . the presence of a bar can efficiently drive gas into the inner nucleus ( friedli & benz 1993 ) , which can be detected in the form of inner dust spirals or bars .
3 ) at this point , the galaxy starts the nls1 phase , powered by a small central black hole accreting at close to its eddington limit . within 10@xmath20 yrs , the bar can trigger star formation ( hunt & malkan 1999 , and references therein ) , which suggests that the evidence for active star formation in nls1s may also be related to the presence of a bar .
dynamical models indicate that once @xmath0 5% of the mass of the disk is redistributed into the inner nucleus , the bar will be destroyed ( freidli & benz 1993 ; norman et al .
although models predict that this will take a few x 10@xmath21 yrs , they do not typically include the effects of asymmterical mass distribution in the inner nucleus , which may shorten the lifetime of the bar ( norman et al .
4 ) at this stage , m@xmath19 will have increased sufficiently ( several to ten times its initial mass ) such that the galaxy appears as a bls1 . since bls1s are substantially sub - eddington , the fueling rate must also have decreased significantly .
there are a number of possible explanations for this , including the weakening of the bar , the diminished reserve of fuel in the inner disk , or the effect of the energized agn itself , which can drive mass outflow ( see crenshaw , kraemer , & george 2003 ) .
interestingly , hunt & malkan ( 1999 ) find evidence that seyferts show outer stellar rings more often than inactive galaxies .
outer rings take @xmath0 several x 10@xmath21 yrs to form , and are likely to outlast the stellar bars .
hence , the rings may be a remnant of the earlier , bar - driven phase of what are now bls1s .
testing the above scenario , and determining the physical connection between large - scale bars and fueling of active nuclei , could be accomplished with more extensive _
images of the nuclear regions in nls1s . the fueling flow can be traced via deep optical and ir images of the dust lanes , and possibly narrow - band images of the low - ionization emission .
narrow - band observations of the high - ionization outflowing gas could be used to determine the geometry of infall versus outflow . finally , uv images could be used to look for evidence of circumnuclear starbursts and
explore their connection to agn activity .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this research has also made use of nasa s astrophysics data system abstract service .
some of the data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
lcllccc eso215-g14 & 0.019 & s & sa & bls1 & & + eso323-g77 & 0.015 & s & sa / b & bls1 & 2500 & w92 + eso354-g4 & 0.033 & s & sa & bls1 & & + eso362-g18 & 0.013 & s & sa & bls1 & 4000 & w92 + eso438-g9 & 0.024 & sb & sbc / d & bls1 & 5000 & k83 + f51 & 0.014 & s & sa & bls1 & 2700 & w92 + f1146 & 0.032 & s & sb & bls1 & 4300 & w92 + heao-1 - 0307 - 730 & 0.028 & sb & s(b)a & bls1 & 2900 & w92 + heao1143 - 181 & 0.033 & i & i & bls1 & 2400 & w92 + heao2106 - 098 & 0.027 & p & p & bls1 & 3835 & r00 + ic1816 & 0.017 & sb & sba / b & bls1 & & + ic4218 & 0.019 & s & sa & bls1 & & + ic4329a & 0.016 & s & sa & bls1 & 4800 & w92 + ir1319 - 164 & 0.017 & s & sb & bls1 & & + ir1333 - 340 & 0.008 & s & so & bls1 & 2400 & w92 + mcg6 - 26 - 12 & 0.032 & sb & sb0 & nls1 & 1145 & v01 + mcg8 - 11 - 11 & 0.020 & s & sb0 & bls1 & 3630 & o82 + mark6 & 0.019 & s & s0 & bls1 & & + mark10 & 0.030 & s & sa / b & bls1 & 2400 & o77 + mark40 & 0.020 & ? & s0 & bls1 & 2000 & o77 + mark42 & 0.024 & sb & sba & nls1 & 865 & v01 + mark50 & 0.023 & s & s0 & bls1 & & + mark79 & 0.022 & sb & sbc & bls1 & 4950 & o82 + mark279 & 0.031 & s & sa & bls1 & 6860 & o82 + mark290 & 0.029 & s & e & bls1 & 2550 & o82 + mark334 & 0.022 & s & i & bls1 & & + mark335 & 0.025 & p & ? & nls1 & 1350 & o82 + mark352 & 0.015 & e & e & bls1 & 3800 & o77 + mark359 & 0.017 & sb & sbb / c & nls1 & 900 &
v01 + mark372 & 0.031 & s & sa & bls1 & 5500 & g91 + mark382 & 0.034 & sb & sba & nls1 & 1280 & v01 + mark423 & 0.032 & s & sb & bls1 & 9000 & o81 + mark471 & 0.034 & sb & sbc & bls1 & & + mark493 & 0.031 & sb & s(b)a & nls1 & 740 & v01 + mark516 & 0.028 & s & sc & bls1 & 4000 & o81 + mark530 & 0.029 & s & sa & bls1 & 6560 & k00 + mark543 & 0.026 & s & sc & bls1 & & + mark590 & 0.027 & s & sa & bls1 & 2680 & s90 + mark595 & 0.028 & s & sa & bls1 & 2360 & s90 + mark609 & 0.032 & s & sa / b & bls1 & & + mark699 & 0.034 & e & e & nls1 & 840 & o81 + mark704 & 0.029 & sb & sba & bls1 & 5500 & s90 + mark744 & 0.010 & s & sb & bls1 & & + mark766 & 0.012 & sb & sbc & nls1 & 1630 & v01 + mark817 & 0.033 & sb & sbc & bls1 & 4300 & o82 + mark833 & 0.039 & i & i & bls1 & & + mark871 & 0.034 & s & sb & bls1 & 3690 & m03 + mark885 & 0.026 & sb & sbb & bls1 & & + mark896 & 0.027 & s & sc & nls1 & 1135 & v01 + mark915 & 0.025 & s & sa & bls1 & & + mark1040 & 0.016 & s & sb & nls1 & 1830 & o82 + mark1044 & 0.016 & s & sa & nls1 & 1010 & v01 + mark1126 & 0.010 & s & sb & bls1 & & + mark1218 & 0.028 & sb & sba & bls1 & & + mark1330 & 0.009 & s & sb / c & bls1 & & + mark1376 & 0.007 & ? & edge on & bls1 & & + mark1400 & 0.029 & s & sa & bls1 & & + mark1469 & 0.031 & s & sa & bls1 & & + ms1110 + 2210 & 0.030 & e & e & bls1 & & + ngc235 & 0.022 & s & sa / b & bls1 & & + ngc526a & 0.018 & i & e / s0 & bls1 & & + ngc1019 & 0.024 & sb & sbb & bls1 & & + ngc1566 & 0.004 & s & sb & bls1 & 2580 & k91 + ngc2639 & 0.011 & s & sb & bls1 & 3100 & h97 + ngc3227 & 0.003 & ? & ? & bls1 & 3920 & s90 + ngc3516 & 0.009 & s & s0 & bls1 & 4760 & c86 + ngc3783 & 0.009 & sb & i & bls1 & 2980 & s90 + ngc4051 & 0.002 & sb & sb & nls1 & 1120 & v01 + ngc4235 & 0.007 & s & ? & bls1 & 7600 & h97 + ngc4748 & 0.014 & s & sa & nls1 & 1565 & v01 + ngc5252 & 0.022 & s & s0 & bls1 & 2500 & a96 + ngc5548 & 0.017 & s & sa & bls1 & 5610 & s90 + ngc5674 & 0.025 & sb & sbc & bls1 & & + ngc5940 & 0.033 & sb & sbc & bls1 & 5240 & m03 + ngc6104 & 0.028 & sb & sbb & bls1 & & + ngc6212 & 0.030 & s & sb & bls1 & 6050 & h86 + ngc6860 & 0.015 & s & sb & bls1 & 3900 & w92 + ngc7213 & 0.006 & s & sa & bls1 & 3200 & w92 + ngc7314 & 0.006 & s & sd & bls1 & & + ngc7469 & 0.017 & s & sb / c & bls1 & 3460 & c86 + iizw10 & 0.034 & s & ?
& bls1 & 3760 & m03 + pks0518 - 458 & 0.034 & e & e & bls1 & 5020 & m03 + tol1059 + 105 & 0.034 & s & s0 & bls1 & & + tol2327 - 027 & 0.033 & s & sb & bls1 & & + um146 & 0.017 & s & sba & bls1 & & + ugc3223 & 0.018 & s & s(b)b / c & bls1 & 4740 & s90 + was45 & 0.024 & sb & sa & bls1 & & + ugc10683b & 0.031 & sb & sba & bls1 & & + ugc12138 & 0.025 & sb & sba & bls1 & & + um614 & 0.033 & s & s0 & bls1 & & + x0459 + 034 & 0.016 & e & e & bls1 & 4320 & g82 + izw1@xmath22 & 0.061 & sb & - & nls1 & 1400 & o77 + ms0144.2 - 0055@xmath22 & 0.080 & s & - & nls1 & 1100 & v01 + mark705@xmath23 & 0.028 & sb & - & nls1 & 1790 & v01 + ms1217.0 + 0700@xmath22 & 0.080 & s & - & nls1 & 1765 & v01 + ms1519.8 - 0633@xmath22 & 0.084 & sb & - & nls1 & 1115 & v01 + ms2210.2 + 1827@xmath22 & 0.079 & sb & - & nls1 & 690 & v01 + | we present a study of the host - galaxy morphologies of narrow- and broad - line seyfert 1 galaxies ( nls1s and bls1s ) based on broad - band optical images from the _ hubble space telescope _ archives .
we find that large - scale stellar bars , starting at @xmath01 kpc from the nucleus , are much more common in nls1s than bls1s .
furthermore , the fraction of nls1 spirals that have bars increases with decreasing full - width at half - maximum ( fwhm ) of the broad component of h@xmath1 .
these results suggest a link between the large - scale bars , which can support high fueling rates to the inner kpc , and the high mass - accretion rates associated with the supermassive black holes in nls1s . |
the equilibrium thermodynamics of a gauge theory is studied in the non - perturbative domain by lattice simulations of the partition function .
much is now known about the phase transitions in @xmath0 and @xmath1 pure gauge theory and in qcd , including the order , the transition temperature , @xmath3 , entropy density , pressure , specific and latent heats and other such quantities @xcite . also of interest
are the screening masses at finite temperatures .
these are defined in general by the exponential spatial falloff of the correlation of two static sources .
a classification of all masses is provided by the transformation properties of the sources . for glueball - like screening masses ,
such a classification was performed in @xcite where the first measurements of several of these masses were reported .
extensive lattice data is available on screening masses from meson and baryon - like sources @xcite , and on the correlation of polyakov lines .
a gauge invariant transfer matrix formulation is easy to write for the lattice regularised theory . for thermal physics
, it is convenient to think of the transfer matrix in a spatial direction .
the free energy and other bulk thermodynamic quantities involve only the largest eigenvalue of the transfer matrix .
correlation functions and the screening masses involve the ratio of this largest eigenvalue with specific other eigenvalues depending on the transformation properties of the source . thus , the full spectrum of screening masses contains much more information about the theory than bulk thermodynamics can provide .
a quantity of special interest is the debye screening mass , @xmath4 , whose inverse gives the screening length of static choromo - electric fields . in the quantum theory ,
this screening mass plays an important role in regulating some infra - red singularities .
lattice measurements have , in the past , concentrated on measuring a mass from the correlation of polyakov loops , @xmath5 .
it was expected that when the gauge coupling becomes small , @xmath6 .
many years ago nadkarni showed that the relation between @xmath4 and @xmath5 is far from being so simple @xcite .
subsequent lattice work focussed on methods of computing @xmath4 .
reisz and collaborators @xcite wrote down a dimensionally reduced theory which could be used to define @xmath4 non - perturbatively .
a recent paper @xcite used the representations of the symmetries of the transfer matrix to write down operators whose correlations could be used to measure the debye mass and @xmath4 was , in fact , first given in @xcite . ] and gave a general parametrisation of the perturbative and non - perturbative terms for @xmath4 .
these parameters have since been determined in a lattice measurement of the debye screening mass using a dimensionally reduced theory at very high temperatures @xcite .
a similar screening mass is required for the magnetic sector of the nonabelian gauge theory in the plasma , in order to get sensible results in the infrared .
however , the magnetic mass is entirely non - perturbative in nature .
it has been the object of many lattice studies @xcite .
in fact , a recent work @xcite tries to extract @xmath4 and @xmath7 from gauge fixed gluon propagators at finite temperature .
many years ago linde discovered @xcite that the thermal perturbation expansion in non - abelian gauge theories breaks down because the magnetic sector is not amenable to perturbative studies . a recent attempt to understand linde s problem in the region where
the coupling @xmath8 has invoked a sequence of dimensionally reduced effective theories @xcite . at length scales of @xmath9 dimensional reduction yields a three dimensional @xmath10 gauge theory coupled to an adjoint scalar field of mass @xmath11 . at longer scales , @xmath12 , the scalar field can be integrated out , and the leading terms in the effective theory correspond to a pure gauge theory in three dimensions . on the basis of this reduction
it has been argued @xcite that a non - vanishing pole in magnetic gluon propagators is absent , and linde s problem is finessed by confinement in the three dimensional gauge theory .
in fact , dimensional reduction has often been used to explore finite temperature theories @xcite , and it has long been argued that the dimensionally reduced theory is confining .
the spatial string tension is known to be non - vanishing for @xmath13 and scales as @xmath14 @xcite .
this , and other , lattice measurements reveal that the gauge coupling close to @xmath3 are too large to trust perturbation theory . at temperatures of a few @xmath3 , the coupling @xmath15 , and
the length scales @xmath16 , and hence can not be decoupled .
the construction of @xcite is not applicable .
however , the transfer matrix argument assures us that @xmath4 is only the smallest in a hierarchy of screening masses .
we can then use the spectrum of the screening masses to explore the symmetries of the transfer matrix and check whether or not dimensional reduction works ; and if it does , then what is the nature of the reduced theory .
our major result is that the degeneracies of the spectrum of screening masses implies that the symmetry group of the spatial transfer matrix is that of two dimensional rotations implying dimensional reduction for @xmath17 .
the measured non - perturbative spectrum is similar to that in 3-d pure gauge theory @xcite .
the symmetries of the transfer matrix , and the physical consequences are discussed in section [ sc.group ] .
the group theory presented in this section is central to the rest of this paper .
the extraction of screening masses in @xmath18 four - dimensional @xmath1 and @xmath0 pure gauge theories take up the next two sections .
these may be skipped by those readers who are not interested in the details of the lattice simulations .
section [ sc.final ] contains a summary of our lattice results and presents our conclusions on the nature of the dimensionally reduced theory .
several technical details are relegated to appendices .
appendix [ sc.irreps ] gives the loop operators used in our computations . in appendix
[ sc.clebsch ] , we examine the possibility of explaining the mass spectrum in terms of perturbative multigluon states . noise reduction techniques for the lattice simulations and the algorithm for projecting on to the lowest state in every channel
are described in appendices [ sc.fuzz ] and [ sc.vary ] respectively .
for the @xmath19 continuum euclidean theory the symmetry of the transfer matrix is the direct product of the full rotation group @xmath20 and the @xmath21 groups generated by charge conjugation , @xmath22 , and time reversal .
irreps of @xmath20 are labelled by the angular momentum and parity , @xmath23 , and of the full symmetry group by @xmath24 . for the lattice
regularised theory , @xmath20 breaks to the discrete subgroup of the symmetries of the cube , @xmath25 .
the consequent reduction of the irreps of @xmath20 is well - known @xcite . in the euclidean formulation of the ( continuum )
equilibrium @xmath18 theory , the transfer matrix in one of the spatial directions is invariant under symmetries of the orthogonal slice .
such slices are three dimensional two of which are spatial and one is the euclidean time .
the symmetry group is that of a cylinder , @xmath26 .
this @xmath21 factor is generated by @xmath27 .
the non - abelian group @xmath28 contains the abelian subgroup of rotations , @xmath29 , and a 2-d parity , @xmath30 .
the 3-d parity @xmath31 , where @xmath32 is the rotation by @xmath33 in @xmath29 . in the high temperature limit
, the effective theory is expected to undergo dimensional reduction to a 3-d gauge theory .
in such a theory , the transfer matrix must have the @xmath28 symmetry of a 2-d slice .
@xmath28 has two one - dimensional irreps @xmath34 ( the 2-d scalar and pseudo - scalar ) and an infinite tower of two - dimensional irreps @xmath35 . under the reduction of @xmath36 to @xmath28 , the spin @xmath37 irrep breaks as @xmath38 on the lattice the irreps of @xmath39 break further into irreps of the automorphism group of a @xmath40-slice , the tetragonal group @xmath41 .
for the three dimensional lattice theory , the automorphism group of the 2-d slice is @xmath42 , which is isomorphic to @xmath43 .
a recent work on 3-d glueballs used this classification of the states @xcite .
this pattern of symmetry breaking is summarised by @xmath44 four of the @xmath21 factor groups are identical , and generated by the 3-d parity @xmath45 . @xmath28 and
@xmath42 contain the 2-d parity @xmath46 .
in this paper we use the notation of @xcite for the crystallographic point groups and their irreducible representations ( irreps ) .
this subsection contains a discussion of the irreducible representations of the lattice symmetries .
the group @xmath43 is generated by eight elements in five conjugacy classes the identity ( @xmath47 ) , rotations of @xmath48 around the @xmath40-axis ( @xmath49 ) , rotations of @xmath33 around the @xmath40-axis ( @xmath50 ) , around the @xmath51 or @xmath52-axes ( @xmath53 ) and around the directions @xmath54 ( @xmath55 ) .
the five irreps are called @xmath56 , @xmath57 , @xmath58 , @xmath59 and @xmath47 .
the first four are one - dimensional and the last two - dimensional .
the 10 irreps of @xmath60 are obtained by adjoining the character of @xmath45 to the irreps of @xmath43 . under the reduction of @xmath25 to @xmath60 ,
the irreps break up as @xmath61 the notation @xmath62 is potentially confusing since these irreps can belong to both @xmath25 and @xmath60 .
a similar confusion can also be caused by the notation @xmath47 , which stands for the two - dimensional irrep of any group . in the rest of this paper
we will mention the group involved whenever we mention one of these irreps . in the dimensionally reduced theory , the lattice symmetry group is @xmath63 .
instead of calling the irreps by the names of the @xmath43 irreps , we denote them by the symbols @xmath64 , @xmath65 , @xmath66 , @xmath67 and @xmath47 respectively .
the first four are one - dimensional irreps , and the sign which indexes them is the character of the 2-d parity @xmath46 in the irrep .
the breaking of @xmath60 irreps to @xmath42 is as @xmath68 our nomenclature for the irreps of @xmath60 is designed to be a mnemonic for the breaking of the @xmath25 irreps ( eq . [ ohtod4h ] ) , but not for the breaking of @xmath60 to @xmath42 ( eq . [ d4htoc4 ] ) . on the other hand ,
the naming convention followed in @xcite simplifies the latter at the expense of the former , since it proceeds from the isomorphism @xmath69 . to make contact with the notation used elsewhere , note that @xmath45 in @xcite corresponds to our 2-d parity @xmath46 and @xmath70 to the character of the operator @xmath71 , which is called @xmath45 in @xcite .
the @xmath28 irreps break to @xmath42 as follows @xmath72 the @xmath64 corresponds to the scalar and the @xmath65 to the pseudo - scalar irrep of @xmath28 .
states or operators also carry a label @xmath73 for the charge conjugation symmetry .
the charge conjugation operator reverses the direction of traversal of a loop , and hence takes the trace into its complex conjugate .
@xmath74 states correspond to real parts of loops and @xmath75 states to imaginary parts . for gauge groups with only real representations , such as @xmath0 , there are no @xmath75 states . the construction of irreps of @xmath60 from loops is given in appendix [ sc.irreps ] .
6truecm in finite temperature lattice simulations , the symmetry group of the transfer matrix is always @xmath60 .
however , at low temperatures , we should expect to see an effective symmetry group @xmath25 .
at the other end of the temperature scale , if dimensional reduction is to be a good approximation , we should see an approximate @xmath42 symmetry .
a guess at the temperature dependence of the screening masses is shown in figure [ fg.guess ] .
if the lattice is big enough , and the lattice spacing is sufficiently small , then we should see the more extended degeneracy of @xmath28 .
in this case the @xmath64 and @xmath65 of @xmath42 will be non - degenerate , since they correspond to different irreps of @xmath28 , but the @xmath66 and @xmath67 will become degenerate , since they correspond to one irrep of @xmath28 .
differences between first and second order phase transitions should be visible in the @xmath76 channel near @xmath3 .
we have simulated the @xmath1 pure gauge theory with wilson action at three temperatures . with @xmath77
the critical coupling is @xmath78 @xcite .
we have performed a simulation at @xmath79 with @xmath80 . at @xmath81
the coupling is @xmath82 @xcite , and for @xmath83 we use the coupling @xmath84 @xcite .
the simulations were performed with a cabbibo - marinari pseudo heat - bath update , where each update acted on three separate @xmath0 subgroups by five kennedy - pendleton moves @xcite .
the class of loop operators measured is listed in appendix [ sc.irreps ] .
noise reduction involved a fuzzing procedure explained in appendix [ sc.fuzz ] . for each irrep of the symmetry group ,
we projected the measured correlation function to the ground state by a variational technique explained in appendix [ sc.vary ] . successive measurements of correlation functions were separated by about one integrated auto - correlation time measured through the polyakov loop .
one of the techniques for extracting screening masses from correlation functions in the long direction ( with @xmath85 sites ) is to solve the equation @xmath86\over \cosh\left[m(z+1/2)(n_z/2-z)\right ] } \;=\ ; { c(z+1)\over c(z ) } \label{su3.local}\ ] ] for the local mass , @xmath87 , given measurements of the correlation function @xmath88 and @xmath89 .
the assumption that the correlation function is described by a single mass is borne out if there is a range of @xmath40 for which the local mass is constant within errors .
if there is such a plateau then we quote it as our estimate of the screening mass .
in addition we have performed fits to correlation functions in the form @xmath90\over
\cosh(\mu_0 n_z/2 ) } + ( 1-r){\cosh\left[\mu_1(n_z/2-z)\right]\over \cosh(\mu_1 n_z/2 ) } \right\ } , \label{su3.form2}\ ] ] where @xmath91 , @xmath92 and @xmath93 are fit parameters with the constraint @xmath94 .
local mass estimates were deemed acceptable if the fitted value of @xmath92 agreed with it .
when the local masses were too noisy to show a plateau , we took @xmath92 as our estimate of the screening mass .
the goodness of the variational projection to the ground state was checked by observing whether the fitted parameter @xmath91 was close to unity . in minimising @xmath95 to perform the fits we took into account covariances of data @xcite .
since we normalised the variational correlator at separation zero to unity , the error in this point is zero .
however , the intrinsic variability in the zero distance correlator is then redistributed over all the points through the covariance . the maximum distance retained in the fit
was always determined by the criterion that the correlation function at that distance should be more than 1-@xmath96 from zero .
the estimates and errors of every measurable were constructed through a jack - knife procedure .
10truecm .fitted masses in @xmath1 theory for @xmath80 on the @xmath98 lattice using 5000 configurations .
note that @xmath91 depends on the basis set of operators , but @xmath92 does not . for the @xmath99 operator
, a plateau in the local masses is used for the estimate of the screening mass . [ cols="^,^,^,^,^,^,^ " , ] gauge invariant @xmath74 states of zero momentum can be constructed as linear combinations of the composite operators @xmath100 , \label{ball}\ ] ] where the trace is over @xmath10 generators .
correlators of such an operator will decay , in leading order , with the mass @xmath101 , where @xmath102 , @xmath103 and @xmath104 the cyclic property of traces ensures that @xmath105 is symmetric under any operation that flips the polarisation indices and simultaneously changes the sign of @xmath106 .
in addition , if we require that the state be symmetric under the exchange of the gluon fields , then only @xmath107 irreps are allowed .
the representation content of these operators in all parts of the brillouin zone is given in table [ tb.clebsch ] . in order to obtain @xmath108 , @xmath74 irreps we have to go to combinations of four gluon field operators .
colour singlet gauge invariant correlators with @xmath75 start with composite operators of three gluon fields .
the reduction of multi - gluon operators then tells us that * for @xmath74 , the lowest mass observed in any @xmath107 channel would be at most half of the lowest mass observed in any @xmath108 channel , since the former are obtained by two - gluon exchange but the latter require at least four - gluon exchange . * if @xmath109 , then @xmath110 . * if @xmath111 , for @xmath112 , then @xmath113 and the mass difference between the @xmath76 and @xmath99 states is @xmath114 $ ] . here
@xmath115 is the spatial size of the @xmath40-slice . * if @xmath116 , then @xmath117 and the mass difference @xmath118 decreases with lattice size .
the first condition is clearly violated in the spectra we obtain for both @xmath0 and @xmath1 theories . at @xmath2 in both the theories we found @xmath119 . in addition , in the @xmath1 theory @xmath120 .
both these observations violate the first condition .
this is sufficient evidence for a failure of the interpretation of these masses as screening masses for perturbative multigluon states .
if we ignore this , and restrict ourselves to the @xmath107 , @xmath74 states only , then one might be tempted to match the @xmath1 spectrum to the third condition above , since the mass difference @xmath121 is non - zero and independent of the lattice size .
however , from eq .
( [ massreln ] ) , the condition @xmath122 is satisfied for @xmath123 but not for @xmath124 . thus , none of the conditions above hold , and we conclude that the spectrum of screening masses can not be given a perturbative interpretation of multigluon states , and indeed gives information about the nonperturbative spectrum of the effective theory .
a similar statement has been made in ref .
@xcite in the context of the massgap obtained from polyakov loop correlators at moderate temperatures , on the basis of the detailed decay patterns of the correlators seen in lattice studies .
loop correlations are known to be very noisy . in order to increase the signal / background ratio , we used a hybrid of teper s doubled - link fuzzing procedure @xcite and the smearing procedure adopted by the ape collaboration @xcite .
we define fuzzed links at level @xmath125 recursively in terms of those at level @xmath126 by the equation @xmath127 @xmath128 are elements of @xmath10 , and the links for @xmath129 are those generated by the monte carlo procedure . for general @xmath10 ,
this maximisation is most easily accomplished using the `` polar '' decomposition of a general complex matrix to write @xmath130 where @xmath131 is a complex number of unit modulus , @xmath132 is hermitian and @xmath128 is special unitary .
there is a discrete ambiguity in this decomposition , corresponding to the signs of the eigenvalues of @xmath132 . when all the eigenvalues of @xmath132 are chosen to be positive
, @xmath128 maximises the trace in eq .
( [ fuzz ] ) . for @xmath0
the algorithm is simpler since @xmath132 is a multiple of the identity .
the projection then involves only a division of @xmath35 by the square root of its determinant . in a test run with @xmath1 at @xmath80 on a @xmath133 lattice , the procedure in eq .
( [ fuzz ] ) was found to perform better than doubled - link fuzzing .
since the latter technique is known to work well on larger lattices , we conclude that the problem is due to the fact that with small lattices , only a small number of doubled - link fuzzing steps is possible . presumably on larger lattices , where more fuzzing levels can be reached ,
equally good results can be obtained with either fuzzing technique . for the @xmath1 theory we worked with eq .
( [ fuzz ] ) and @xmath134 . for @xmath0 ,
since we use a lattice which has 12 spatial sites , upto three levels of doubled - link fuzzing can be performed for the spatial links .
for finite temperature problems it is perfectly all right if the number of fuzzing steps in the time direction is different from that in other directions .
we perform only one doubled - link fuzzing in the time direction .
we checked that using three levels of doubled - link fuzzing gave a better projection than seven steps of ( c.1 ) , and therefore used the former , for our runs near @xmath3 . for our runs at @xmath135 and @xmath136 ,
we experimented with a combination of doubled - link fuzzing and ( c.1 ) ; using a combination of one doubled - link fuzzing followed by two steps of ( c.1 ) , the whole set being repeated once . since this gave a slight improvement over three steps of doubled - link fuzzing , we used this technique at these higher temperatures . for our runs on
the smaller lattice we used one doubled - link fuzzing followed by 4 steps of eq .
( [ fuzz ] ) .
it is not known a priori which linear combination of loop operators acting on the vacuum generates the state with the lowest mass in a channel with given quantum numbers .
however , such a state will give a correlation function which has the slowest possible decay with increasing separation .
given a basis set of loop operators , we can try to construct a linear combination which satisfies this property of slowest decay .
this is the idea of a widely used variational technique @xcite .
since we have found no discussion in the literature of a numerically stable algorithm for its implementation , we document such a method here .
we construct cross correlations between all the loop operators at our disposal to yield the ( symmetric ) matrix of correlations @xmath137 .
a combination of operators which has large projection to the ground state is obtained by solving the variational problem over @xmath138 @xmath139= \lambda(z_0,z_1),\qquad(z_0<z_1 ) .
\label{vary}\ ] ] if @xmath140 is positive definite , as guaranteed by the reflection positivity of the wilson action , then this extremisation problem reduces to finding the maximum eigenvalue of the system @xmath141 with the corresponding eigenvector , we define the @xmath142 variational correlator @xmath143 we can utilise the freedom of normalising the eigenvector @xmath138 to set the variational correlator to unity at separation @xmath144 .
reflection positivity of the action guarantees that @xmath140 is positive definite .
it can be treated as a metric , and after appropriate scaling , the problem in eq .
( [ vary ] ) can be phrased as the extremisation of a quadratic form over a sphere leading to the usual matrix eigenvalue problem @xcite .
algorithmically , this naive idea can be implemented by transforming both sides of eq .
( [ geig ] ) to the basis where @xmath140 is diagonal , absorbing the diagonal elements into @xmath138 by appropriate rescaling , and then solving the usual eigenvalue problem for this transformed @xmath145 .
however , if some of the eigenvalues of @xmath140 are small , then the extremum problem is ill - conditioned because the solution is sent off to infinity along the nearly flat directions . with finite statistics
the problem may be even worse . due to statistical fluctuations
, the measured correlation matrix may not be positive definite .
it is then better to treat the problematic directions as exactly flat , since this discards the subset of the data which is most corrupted by noise .
the solution is easily specified by going to the basis in which @xmath140 is diagonal and blocking the matrices into the form @xmath146 where the eigenvalues , @xmath147 , of @xmath140 which satisfy the cut condition @xmath148 have been set to zero . here
@xmath149 is the maximum eigenvalue of @xmath140 .
the diagonal sub - matrix @xmath150 is positive definite and has a condition number less than @xmath151 . in this basis , eq .
( [ vary ] ) is equivalent to the set of equations @xmath152 solving the latter for @xmath153 and substituting into the former gives the eigenvalue problem @xmath154 y_1 = \lambda g y_1 , \label{sys2}\ ] ] which is well - defined and numerically well - conditioned as long as @xmath155 is invertible .
notice , however , that eq .
( [ geig ] ) is ill - conditioned only if both @xmath145 and @xmath140 have nearly flat directions ; otherwise their roles may be interchanged by inverting eq .
( [ vary ] ) and converting the maximum problem into that of finding a minimum .
if this can not be done , then we must take care of the case that @xmath155 is not invertible .
the @xmath156 matrix @xmath157 is a map from @xmath158 to @xmath159 ( _ i.e. _ , @xmath160 has @xmath161 real components and @xmath153 has @xmath162 ) .
its range is the subset of @xmath159 to which the whole of @xmath158 is mapped .
if @xmath155 is singular , then its null - space is contained in the complement of the range of @xmath157 . as a result , the second of eq .
( [ sys1 ] ) can be solved by a left - multiplication by any matrix which coincides with the inverse of @xmath155 in the complement of its null - space ( _ i.e. _ , in the range of @xmath157 ) .
therefore , in eq .
( [ sys2 ] ) @xmath163 can be replaced by a pseudo - inverse @xmath164 @xmath165 is an orthogonal matrix , @xmath166 is diagonal , and the small components of @xmath166 are set to zero in the pseudo - inverse @xmath167 . with this definition of @xmath168 ,
the generalised eigenvalue problem is completely well - defined and eq .
( [ sys2 ] ) may be solved by the naive algorithm described earlier .
a very nice numerical illustration is provided by our data on the @xmath169 matrix of the @xmath76 correlation function at @xmath2 for the @xmath1 gauge theory ( section [ sc.su3hi ] ) .
the distribution of @xmath170 is shown in figure [ fg.eigs ] .
the ten small eigenvalues clustered at the end of a huge spectral gap have both positive and negative signs , and are due to noise in the data . because they are so well separated , the cut @xmath171 ( in eq . [ cut ] )
can be chosen to have any value between @xmath172 and @xmath173 .
the results for eigenvalues and eigenvectors are stable in this whole range of choices .
somewhat smaller values , @xmath174 , are also found to be acceptable and are in fact preferred for reasons of numerical stability of the linear algebra routines .
removing the cut destabilises the problem completely .
the spectral distributions are similar in most channels for both @xmath0 and @xmath1 theories .
in a few cases the spectral distribution is gapless but has a long tail . even in such cases , @xmath174@xmath172 give stable results .
99 see for example , e. laermann , _ nucl .
_ , b ( _ proc
. suppl . _ ) , 63a - c ( 1998 ) 114 , and references therein .
b. grossman , s. gupta , f. karsch and u. heller , _ nucl .
_ , b 417 ( 1994 ) 289 . c. detar and j. kogut , _ phys
_ , 59 ( 1987 ) 399 ; _ phys . rev .
_ , d 36 ( 1987 ) 2828 ; + s. gottlieb _ et al .
_ , _ phys .
_ , 59 ( 1987 ) 1881 ; + a. gocksch , p. rossi and u. m. heller , _ phys
_ , b 205 ( 1988 ) 334 ; + k. born _ et al .
_ , _ phys .
_ , 67 ( 1991 ) 302 ; + s. gupta , _ phys . lett .
_ , b 288 ( 1992 ) 171 ; + c. bernard _ et al .
_ , _ phys .
lett . _ , 68 ( 1992 ) 2125 ; + g. boyd , s. gupta and f. karsch , _ nucl . phys .
_ , b 385 ( 1992 ) 481 ; + t. hashimoto , t. nakamura and i. o. stamatescu , _ nucl .
_ , b 400 ( 1993 ) 267 .
s. nadkarni , _ phys .
_ , d 33 ( 1986 ) 3738 ; and d 34 ( 1986 ) 3904 .
a. irbck _ et al .
_ , _ nucl .
_ , b 363 ( 1991 ) 34 ; + t. reisz , _ z. phys . _ , c 53 ( 1992 ) 169 ; + l. krkkinen _ et al .
_ , _ phys .
_ , b 282 ( 1992 ) 121 ; + l. krkkinen , p. lacock , b. petersson and t. reisz , _ nucl .
_ , b 395 ( 1993 ) 733 .
p. lacock and t. reisz , _ nucl .
_ , b ( _ proc . suppl _ ) 30 ( 1993 ) 307 .
p. arnold and l. g. yaffe , _ phys .
_ , d 52 ( 1995 ) 7208
. k. kajantie _ et al .
_ , _ phys .
_ , 79 ( 1997 ) 3130 .
a. billoire , g. lazarides and q. shafi , _ phys .
_ , b 103 ( 1981 ) 450 ; + t. a. degrand and d. toussaint , _ phys .
_ , d 25 ( 1982 ) 526 ; + g. lazarides and s. sarantakos , _ phys .
_ , d 31 ( 1985 ) 389 .
u. m. heller , f. karsch and j. rank , _ phys .
_ , d 57 ( 1998 ) 1438 .
a. d. linde , _ phys .
_ , b 96 ( 1980 ) 289 .
e. braaten , _ phys .
_ , 74 ( 1995 ) 2164 ; + e. braaten and a. nieto , _ phys . rev .
_ , 74 ( 1995 ) 3530 .
ginsparg , _ nucl .
_ , b 170 ( 1980 ) 388 ; + t. applequist and r. d. pisarski , _ phys .
_ , d 23 ( 1981 ) 2305 ; + k. kajantie , k. rummukainen and m. shaposhnikov , _ nucl .
_ , b 407 ( 1993 ) 356 .
g. bali _ et al .
_ , _ phys .
_ , 71 ( 1993 ) 3059 ; + l. krkkinen _ et al .
_ , _ phys .
_ , b 312 ( 1993 ) 173 .
m. teper , preprint hep - lat/9804008 . b. berg and a. billoire , _ nucl .
_ , b 221 ( 1983 ) 109 .
m. hamermesh , _
`` group theory and its applications to physical problems '' _ , addison - wesley , reading , massachusetts , 1962 . m. fukugita , m. okawa and a. ukawa , _ nucl .
_ , b 337 ( 1990 ) 181
. y. iwasaki _ et al .
_ , _ phys .
_ , 67 ( 1991 ) 3343 , and _ phys . rev .
_ , d 46 ( 1992 ) 4657 . h. ding and n. christ , _ phys .
_ , 60 ( 1988 ) 1367 ; + a. d. kennedy , j. kuti , s. meyer and b. j. pendleton , _ phys . rev .
_ , 54 ( 1985 ) 87 ; + s. a. gottlieb _ et al .
_ , _ phys .
_ , 55 ( 1985 ) 1958 .
a. d. kennedy and b. j. pendleton , _ phys .
_ , b 156 ( 1985 ) 393
. s. gottlieb _ et al .
_ , _ phys .
_ , d 38 ( 1988 ) 2245 .
f. r. brown _ et al .
_ , _ phys .
_ , 61 ( 1988 ) 2058 .
p. bacilieri _ et al .
_ , _ phys .
_ , b 220 ( 1989 ) 607
. s. gupta _ et al .
_ , _ nucl .
_ , b 329 ( 1990 ) 263 .
r. v. gavai , f. karsch and b. petersson , _ nucl .
_ , b 322 ( 1989 ) 738
. h. chen , j. sexton , a. vaccarino and d. weingarten , _ nucl .
( proc . suppl . )
_ , b 34 ( 1994 ) 357 .
de forcrand , g. schierholz , h , schneider and m. teper , _ phys .
_ , b 152 ( 1985 ) 107 .
c. michael and m. teper , _ nucl .
_ , b 314 ( 1989 ) 347 .
j. engels , j. fingberg and m. weber , _ nucl .
_ , b 332 ( 1990 ) 737 .
j. fingberg , u. heller and f. karsch , _ nucl .
_ , b 392 ( 1993 ) 493 .
m. creutz , _ phys .
_ d 36 ( 1987 ) 515 . c. michael and m. teper , _ phys .
_ , b 199 ( 1987 ) 95 .
j. engels , f. karsch and h. satz , _ nucl .
_ , b 315 ( 1989 ) 419 .
c. michael and s. j. perantonis , _ j. phys .
_ , g 18 ( 1992 ) 1725
. s. p. booth _
_ , _ nucl .
_ , b 394 ( 1993 ) 509 . j. c. neil and j. zinn - justin , _ nucl .
_ , b 280 ( 1987 ) 355 ; + s. gupta , _ phys .
_ , b 325 ( 1994 ) 418 .
e. braaten and a. nieto , _ phys .
_ , 74 ( 1995 ) 3530 .
+ m. teper , _ phys .
_ , b 183 ( 1986 ) 345 . m. albanese _ et al .
_ , b 192 ( 1987 ) 163 . m. lscher and u. wolff , _ nucl .
_ , b 339 ( 1990 ) 222
. j. h. wilkinson , _ the algebraic eigenvalue problem _ , 1965 , clarendon press , oxford ; + h. rutishauser , _ lectures on numerical mathematics _ , 1990 , birkhuser , boston , usa . | we studied screening masses in the equilibrium thermodynamics of @xmath0 and @xmath1 pure gauge theories on the lattice . at a temperature of @xmath2 we found strong evidence for dimensional reduction in the non - perturbative spectrum of screening masses .
mass ratios in the high temperature @xmath1 theory are consistent with those in the pure gauge theory in three dimensions . at the first order @xmath1 phase transition
we report the first measurement of the true scalar screening mass .
@=11 addtoresetequationsection
@=12
# 1to # 1 * dimensional reduction and screening masses + in pure gauge theories at finite temperature * saumen datta and sourendu gupta + department of theoretical physics , + tata institute of fundamental research , + homi bhabha road , mumbai 400005 , india . |
in its very short life time the hitomi x - ray observatory has revealed extremely valuable information on the nature of the intracluster medium ( icm ) of the perseus cluster of galaxies @xcite .
the analysis of hitomi x - ray observations by @xcite show a relatively quiescent icm .
the hot gas has a line - of - sight velocity dispersion of @xmath0 in the region @xmath1 to @xmath2 from the central nucleus .
this implies that the heating of the icm by dissipation of turbulence energy can not offset radiative cooling .
another heating mechanism is at work in the perseus cluster .
the heating process of the icm in cooling flows in clusters and groups of galaxies and in galaxies operates in a negative feedback cycle ( e.g. , @xcite ) .
the energy from the mass - accreting super massive black hole is carried to the icm by jets ( or collimated wind ) , hence it is termed the jet feedback mechanism ( jfm ) .
several processes have been proposed to transfer the energy from the jets to the icm ( e.g. , @xcite ) , some of which can act together , e.g. , cosmic rays and thermal conduction ( e.g. , @xcite ) .
@xcite argues that sound waves , as inferred from the ripples in perseus @xcite , that are excited by jet - inflated bubbles @xcite are more efficient in transferring energy than weak shocks are . @xcite
apply the sound wave heating mechanism to the perseus cluster following the hitomi observations .
@xcite propose that shocks heat the icm in the virgo cluster , while @xcite argue that shocks excited by periodic jets activity heat the icm of the galaxy group ngc 5813 .
@xcite , on the other hand , found problems in the process of shocks - heating of the icm of ngc 5813 .
more generally , @xcite and @xcite found in their two - dimensional numerical simulations , that mixing of the hot bubble gas with the icm is much more significant than heating by shocks . @xcite and @xcite already discussed heating by mixing of the hot bubble gas with the icm .
however , their simulations were unrealistic as they injected artificial bubbles , and mixing was achieved by the destruction of the bubbles by instabilities .
the new simulations @xcite show the bubbles to survive for a long time , and show that jets induce vortices inside and outside the hot bubbles .
these are the vortices that cause the efficient mixing of the hot bubble gas with the icm .
another heating mechanism is by dissipation of turbulence in the icm ( e.g. , @xcite ) , or heating by turbulence and turbulent - mixing ( e.g. @xcite ) .
based on x - ray observation @xcite argue that dissipation of icm turbulence is the main heating process of the icm in the virgo and perseus cooling flow clusters .
@xcite deduce , based on numerical simulations , that turbulence can not be the main heating source .
@xcite , based on idealized simulations with artificial bubble injection ( namely , not with jet ) , reach a similar conclusion .
cosmic rays can also heat the icm ( e.g. , @xcite ) , including the icm in perseus ( e.g. , @xcite ) .
@xcite argues that mixing of cosmic rays that fill the jet - inflated bubbles with the icm is essential for heating by cosmic rays . in an earlier paper @xcite we conducted three - dimensional hydrodynamical simulations of jet - inflated bubbles in the icm . for the parameters used in those simulations
, we found the mixing by vortices to be the main heating mechanism ( mixing - heating ) .
we found that the mixing process accounts for @xmath3 of the energy transferred from the jets to the icm .
kinetic energy of the icm carries about @xmath4 of the energy that is deposited by the jets to the icm .
only a fraction of the kinetic energy develops to icm turbulence and sound waves .
the other fraction of the kinetic energy is in large scale flows .
shocks play only a minor role in heating the icm . based on our earlier findings , as inferred from the fraction of kinetic energy
, it seems that sound waves also play a minor role .
@xcite performed three - dimensional hydrodynamical simulations for a much longer time , and also found that mixing is the main heating process .
they found though , that heating by shocks is the second important process and turbulent heating contributes a small fraction .
we note that turbulence can play a role in determining the properties of the icm ( e.g. , @xcite ) even when it is not the main heating source . in any case , turbulence in the icm
is expected as a by product of the mixing process that is induced by vortices @xcite .
this expectation from numerical simulations is compatible with findings of turbulence in the icm of some cooling flows ( e.g. , @xcite ) .
following the results of hitomi @xcite we conduct further analysis of our 3d numerical simulations .
although the simulations from @xcite were not aiming at the perseus cluster , the reanalysis of our simulations further strengthens our conclusion that heating by mixing process is the main heating process .
we analyze the 3d hydrodynamical simulations that were presented in our earlier paper @xcite , where more details on the numerical scheme and convergence tests can be found .
we present here only the main features of the numerical code .
we use the pluto code @xcite for the hydrodynamic simulations , in a three - dimensional cartesian grid with adaptive mesh refinement ( amr ) .
the computational grid is in the octant where the three coordinates @xmath5 , @xmath6 and @xmath7 are positive . at the @xmath8 , @xmath9 and @xmath10 planes we apply reflective boundary conditions .
the @xmath7 coordinate is chosen along the initial axis of the jets .
in reality two opposite jets are launched simultaneously , such that the flow here is assumed to be symmetric with respect to the @xmath10 plane , amounting to reflective boundary conditions at @xmath10 .
the base computational grid ( lowest amr level ) spans the cube @xmath11 , with @xmath12 divisions in each direction . up to @xmath13 amr levels
are employed with a refinement ratio of @xmath14 .
thus , the highest resolution is @xmath15 . on the outer boundaries we use an outflow boundary condition . at the boundary @xmath10
we inject into the grid a jet with a half - opening angle of @xmath16 .
the jet material is inserted through a circle @xmath17 on the @xmath10 plane with a radius of @xmath18 . the initial jet velocity in the nominal case , the case we will present here , is @xmath19 , a mach number of about @xmath20 .
the direction of the velocity at each injection point @xmath21 in the circle is @xmath22 , where @xmath23 .
the jet is injected during each active episode , and when the jet is turned off reflective boundary conditions apply in the whole @xmath10 plane .
the power of the two jets together during each on - episode is @xmath24 .
the mass deposition rate is thus @xmath25 .
the jet is periodically turned on for @xmath26 and off for @xmath26 .
the simulation begins with an isothermal box of gas at an initial temperature of @xmath27 with a density profile of ( e.g. , @xcite ) @xmath28 ^ { 3 / 4}},\ ] ] with @xmath29 and @xmath30 .
a gravity field is added to maintain an initial hydrostatic equilibrium , and is kept constant in time .
we include radiative cooling in the simulations , where the tabulated cooling function is taken from the solar - metalicity values of table 6 in @xcite .
vortices that are induced by the jets and the jet - inflated bubbles play a crucial role in the interaction of the jets with the icm @xcite . in particular , they lead to mixing of hot bubble gas with the icm .
this mixing heats the icm in a process termed mixing - heating . to demonstrate vortices , we present only two examples from @xcite ; figures 4 and 5 there .
more details can be found there , as well as more cases in earlier papers ( e.g. , @xcite ; @xcite ) . in figs . [
figure : tracers tr4 ] and [ figure : tracers tr3 ] we present the flow pattern in the meridional plane ( perpendicular to the symmetry plane ) at two times in each figure , together with the concentration of traced gas .
the scale of the flow arrows and the initial location of the tracer are described in the figure captions .
the tracers are artificial flow quantities frozen - in to the flow , and hence represent the spreading with time of gas starting in a certain volume .
a tracer s initial value is set to @xmath31 in a certain volume and @xmath32 elsewhere .
if the traced gas is mixed with the icm or the jet s material , its value drops to @xmath33 . figs .
[ figure : tracers tr4 ] and [ figure : tracers tr3 ] demonstrate the complicated flow pattern with many vortices , that is induced by the jets and the bubbles , and the vigorous mixing that takes place
. the vortices also induce sound waves @xcite and turbulence @xcite .
we now turn to examine the velocity dispersion of this complicated flow .
although we did not aim to study the perseus cluster in the simulations we present here ( conducted in 2015 before the hitomi observations ) , the parameters used here are not too different than those of perseus .
the initial temperature in our simulations is @xmath27 , just a little lower than the icm temperature in perseus @xcite .
the density in the inner region is as observed in perseus , @xmath34 for the electron number density .
the average power of the two jets in the present simulations is @xmath35 , while that of the jets in perseus , as inferred from the cavities @xcite , is @xmath36 .
so we might even overestimate the velocity dispersion in perseus .
we do note that @xcite require that the power in sound waves in their suggested heating mechanism be @xmath37 at r=@xmath38 , not much different than the power of the jets in our simulation .
we follow the evolution of the line of sight root mean square ( rms ) of the velocity of the two tracers and of the gas in the grid .
the rms numerical velocity along the line of sight , termed here numerical velocity dispersion , is given by @xmath39 where @xmath40 stands for one of the two tracers or the grid , and @xmath41 and @xmath42 are the kinetic energy and mass , respectively .
we present the line - of - sight rms velocity for gas at @xmath43 , which is basically the icm , and for gas at @xmath44 which is the shocked jet s material .
the relevant quantity to compare with hitomi is the icm gas . in figs .
[ figure : tracers tr4 ] and [ figure : tracers tr3 ] we show the evolution of the line - of - sight rms velocity for @xmath45 and @xmath46 that were defined in figs .
[ figure : tracers tr4 ] , and in fig .
[ figure : tracers tr3 ] . in fig .
[ figure : tracergrid ] we show the numerical velocity dispersion for the entire gas in the grid , according to its temperature . in calculating the numerical velocity dispersion of the gas in the grid we consider all the gas that is located inside the numerical grid at a given time .
we do not count gas that has already left the grid . as we inject more jet s material
, high velocity gas replaces lower velocity gas that has left the grid .
this is the explanation for the increase in the numerical velocity dispersion in the grid that is seen in the left panel of figure [ figure : tracergrid ] .
had we continued the simulation for a longer time , we would have reach a steady state .
+ + + for these simulation we @xcite found that the main heating of the icm is by mixing .
only @xmath47 of the energy is carried by kinetic energy that is distributed among shock waves , sound waves , and global flow .
we find that the line - of - sight rms velocity is in the range of @xmath48 , compatible with the finding from hitomi observations of a line - of - sight velocity dispersion of @xmath0 @xcite .
we consider this compatibility to strengthen our conclusion that heating - mixing is more significant than heating by turbulence and sound waves .
vortices that cause the heating - mixing , excite also sound waves and turbulence , and hence we expect the presence of these in the icm .
the hitomi x - ray observations of the perseus cluster show that dissipation of turbulent energy can not be the main heating mechanism of the icm .
this result , that was anticipated by some studies , leaves three potential heating mechanisms of the icm in the perseus cluster of galaxies .
these heating processes are the dissipation of sound waves @xcite , cosmic rays ( e.g. , @xcite ) , and mixing - heating , where hot gas from jet - inflated bubbles is mixed with the icm @xcite .
we note that heating by cosmic rays and mixing - heating are not excluded , as @xcite argued that mixing is essential for the heating by cosmic rays model to account for observations .
vortices that are formed in the interaction of the jets with the icm excite both the sound waves @xcite and cause the mixing @xcite . in a previous paper @xcite we found that the main heating mechanism of the icm is mixing - heating .
the kinetic energy deposited to the icm by the jets and the bubbles they inflate is about one quarter of the energy deposited by mixing .
the kinetic energy includes also the global flow of the gas , such that the energy carried by sound waves and turbulence is less than @xmath49 of the thermal energy deposited to the icm by mixing .
then there are still uncertainties about the dissipation rate of sound waves in the icm @xcite .
if the dissipation length is large , then only a fraction of the energy that is carried by the sound waves will be dissipated in the cooling flow region in perseus . on the other hand
, @xcite argued that the dissipation length of sound waves is too short , and the energy will be deposited only in the central region of the cooling flow .
we can summarize our view on the heating process of the icm in perseus , based on our numerical simulations and the new result of hitomi , in the following way .
the inflation of bubbles by jets in the icm is accompanied by shocks that run through the icm and by many vortices inside and outside the jet - inflated bubbles .
the vortices excite sounds waves and turbulence , and lead to mixing of hot post - shock jets material , the hot bubble gas , with the icm .
all four processes , shocks , dissipation of sound waves , dissipation of turbulence , and mixing , contribute to the heating of the icm .
our finding is that mixing - heating contributes the most , more than @xmath50 of the heating , including mixing of cosmic rays .
the rest , less than @xmath47 , and possibly more likely @xmath51 , is contributed by the other three processes that must be present , as they are by products of the inflation of bubbles . for computational resources
we acknowledge the linksceem / cy - tera project , which is co - funded by the european regional development fund and the republic of cyprus through the research promotion foundation . | we compare the velocity dispersion of the intracluster medium ( icm ) of the perseus cluster of galaxies as observed by the hitomi x - ray telescope to our three - dimensional hydrodynamical simulations of jet - inflated bubbles in cluster cooling flows , and conclude that the observations support the mixing - heating mechanism of the icm . in the mixing - heating mechanism
the icm is heated by mixing of hot bubble gas with the icm .
this mixing is caused by vortices that are formed during the inflation process of the bubble .
sound waves and turbulence are also excited by the vortices , but they contribute less than 20 per cents to the heating of the icm .
shocks that are excited by the jets contribute even less .
_ key words : _ galaxies : clusters : intracluster medium galaxies : clusters : individual : perseus galaxies : jets [ firstpage ] |
membrane vesicles are made of a closed bilayer of amphiphilic molecules in aqueous solution , having length scales of @xmath0@xmath1 @xmath2 m .
used as simplified models of biological membranes , they have been one of the most extensively studied systems in soft - matter physics @xcite .
vesicles usually enclose both solvent ( water ) and solute molecules .
such vesicular capsules are ubiquitous in cell functions @xcite and used as microreactors and delivery vehicles in various biomedical and cosmetic applications @xcite .
the hydrophobic core of the bilayer membrane poses a kinetic barrier to the permeation of water and water - soluble molecules into and out of the vesicle .
consequently , the permeability coefficients of various solutes across various membranes span a very wide range of values @xcite .
for example , the permeability coefficients of water , urea , and potassium ions through the membranes used in the current work are of the order , respectively , of @xmath3 , @xmath4 , and @xmath5 @xmath2 m/s @xcite .
this implies that the time scales for permeation of water and the solutes examined here are well separated .
thus , the vesicles can safely be assumed to reside in a semi - permeable regime , in which their volume quickly ( essentially immediately ) re - adjusts through osmosis to a change in the number of enclosed solute molecules . as more solute molecules enter , therefore , the vesicle progressively swells first approaching a spherical shape ( the `` ironing '' stage ) , subsequently inflating as a sphere ( the stretching stage ) , and eventually rupturing ( osmotic lysis ) @xcite . in an earlier theoretical work
we argued that a vesicle in such a semi - permeable regime should reach the end of the ironing stage ( _ i.e. _ , the maximum volume - to - area ratio ) critically , through a continuous transition @xcite .
that theory was restricted to unstretchable membranes . here
we extend the theory to the experimentally relevant case of stretchable membranes , finding a slightly modified but similar form of criticality .
we then confirm the existence of the critical scaling behavior in experiments and utilize it to obtain a reliable measurement of permeability coefficients .
a crossover between two stages of vesicle swelling is well known in micropipette - aspiration experiments @xcite .
the stages are distinguished by a markedly different dependence of surface tension on surface strain in the first stage the dependence is exponential , while in the second it is linear @xcite .
the essential difference between this scenario and the one addressed here lies in the different control parameters . in the former case
the swelling is controlled by a hydrostatic pressure difference , whereas in the latter it is controlled by the number of encapsulated solute molecules . as a result ,
the two crossovers are not equivalent .
as will be presented below , in fact , the monitored swelling of our vesicles , with its two distinct stages , occurs within the linear regime_i.e .
_ , the tension depends linearly on strain throughout the observed transition .
we begin by describing the experimental setup in sec .
[ sec_exp ] and the direct experimental results in sec .
[ sec_results ] .
the analysis of the experimental data requires a revised theory , which is presented in sec .
[ sec_theory ] . in sec .
[ sec_analysis ]
we apply the theory to the experimental results to demonstrate data collapse onto a master curve_i.e .
_ , the law of corresponding states and extract additional information , such as the permeability coefficients and the number of surface bending modes contributing to vesicle thermodynamics .
finally , we discuss the various results and their significance in sec .
[ sec_discuss ] .
d-(+)-glucose , d-(+)-sucrose , glycerol , urea , and ethylene glycol were purchased from fluka ( buchs , switzerland ) .
methanol and chloroform were purchased from kemika ( zagreb , croatia ) .
1-palmitoyl-2-oleoyl-_sn_-glycero-3-phosphocholine ( popc ) was purchased from avanti polar lipids ( alabaster , usa ) .
all the solutions were prepared in double - distilled sterile water . a suspension of popc giant unilamellar vesicles ( guvs ) in 0.1 or 0.2 mol / l 1:1 sucrose / glucose solution was prepared using an electroformation method , described in ref . with some modifications @xcite .
lipids were dissolved in a mixture of chloroform / methanol ( 2:1 , v / v ) to a concentration of 1 mg / ml . a volume of 25 @xmath2l of the lipid solution was spread onto a pair of pt electrodes and dried under reduced pressure ( water aspirator ; @xmath6 mmhg ) for 2 hours .
the electrodes were then placed into an electroformation chamber , which was filled with 0.1 or 0.2 mol / l sucrose . ac current ( 8 v , 10 hz ) was applied , and the voltage and frequency were reduced in steps to the final values of 1 v and 1 hz @xcite .
subsequently , the chamber was first drained into a beaker and then flushed with an equal volume of isomolar glucose solution , thus resulting in a suspension of guvs containing entrapped sucrose in a 1:1 sucrose / glucose solution , which increases the contrast in a phase contrast setup and facilitates vesicle manipulation @xcite .
this procedure yields mostly spherical unilamellar vesicles , with diameters of up to 100 @xmath2 m .
an inverted optical microscope ( nikon diaphot 200 , objective 20/0.40 ph2 dl ) with micro - manipulating equipment ( narishige mmn-1/mmo-202 ) and a cooled ccd camera ( hamamatsu orca - er ; c4742 - 95 - 12erg ) , connected _ via _ an ieee-1394 interface to a pc running hamamatsu wasabi software , was used to obtain phase contrast micrographs . in the streaming mode ,
the camera provides @xmath7 12-bit grayscale images at a rate of 8.9 images / s . in the experiment , an individual spherical guv is selected , fully aspirated into a glass micropipette whose inner diameter exceeds the vesicle s diameter , and transferred from a solution containing solutes of very low membrane permeability ( 1:1 glucose / sucrose ) into an iso - osmolar solution of a more permeable solute ( glycerol , urea , or ethylene glycol ) , where the content of the micropipette is released , and the micropipette is subsequently removed .
vesicle response is recorded using a ccd camera mounted on the microscope .
the radius of the vesicle s cross - section was determined from the recorded series of images using a least - squares procedure @xcite .
once the vesicle is released and gets in contact with the target solution , a transient increase of its cross - sectional radius , @xmath8 , is observed .
we attribute it to a small hypotonicity of the outer solution , causing the vesicle to slightly deflate and change its shape from a sphere into an oblate spheroid @xcite .
gravity should lead to a small deviation of the shape from a perfect oblate spheroid @xcite , which we neglect here .
a more important effect of gravity is the breaking of the problem s rotational symmetry , such that the observed lateral radius , @xmath8 , is always the spheroid s larger radius . on a longer time scale than the transient deflation
, the vesicle inflates due to solute permeation and the accompanying osmosis ( _ i.e. _ , water influx ) .
this causes the vesicle to become more spherical , making the observed @xmath8 decrease .
( see fig .
[ fig_swellingcurve ] . )
this is the ironing stage , where the swelling increases the volume - to - area ratio . after a certain time
, @xmath8 reaches a minimum and starts increasing .
( see again fig .
[ fig_swellingcurve ] . )
this marks the crossover to the stretching stage , where the vesicle continues to swell essentially as an inflating sphere .
when the membrane reaches a critical strain , it ruptures , and the vesicle bursts .
subsequently , the membrane is resealed and another cycle of swelling commences . the repeated burst
swelling cycles were analyzed elsewhere @xcite . in the present work we focus on the swelling that precedes the first burst .
the ironing and stretching stages are clearly distinguished in fig.[fig_swellingcurve ] as the decreasing and increasing parts of the swelling curve , respectively . to have a systematic definition of the time axis for all vesicles we define @xmath9 at the minimum of the curve .
we associate the radius at that minimum , @xmath10 , with the relaxed area of the vesicle , @xmath11 .
( we return to examine this assumption in sec .
[ sec_discuss ] . ) from each such curve we also extract the radius at rupture , @xmath12 .
this yields a direct measurement of the membrane s rupture strain , @xmath13 figures [ fig_swellingcurves](a)(c ) show the swelling curves measured for popc vesicles of broadly distributed sizes due to the permeation of three different solutes : urea , glycerol , and ethylene glycol .
each of these curves has the typical shape shown in fig .
[ fig_swellingcurve ] , yet the polydispersity of the vesicles flattens them once they are displayed together .
this is demonstrated in fig .
[ fig_r1r0 ] , where we have replotted the swelling curves for urea after rescaling the ordinate of each of them by its minimum radius , @xmath10 .
figure [ fig_r1r0 ] shows that such a simple size rescaling does not collapse the data onto a master curve .
a more detailed analysis of the temporal swelling curves requires a theory , which is presented in the next section . in fig .
[ fig_swellingcurves ] it is apparent , nonetheless , that there is no correlation between the transition width ( sharpness of the curve s minimum ) and the vesicle size .
we return to this point in sec .
[ sec_analysis ] . a dataset that can be presented without further analysis is the one for @xmath14 , the critical surface strain at rupture , calculated from the directly measured values of @xmath10 and @xmath12 according to eq .
( [ strain ] ) . since rupture is a result of pore formation , which is a nucleated event @xcite , the critical strain as measured in individual vesicles is a stochastic variable .
figure [ fig_rupture ] shows the histogram of measured rupture strains . from this distribution
we get @xmath15 , _
i.e. _ , about 5 percent , which agrees with known values for the maximum sustainable strain of lipid bilayers @xcite .
models of vesicles usually focus on the statistical mechanics of the membrane under the appropriate constraints , the starting point being the helfrich hamiltonian @xcite . in the current work
we choose a different , more general description , based on the thermodynamics of small systems @xcite .
this is because the phenomenon of main interest here the transition between the ironing and stretching stages arises from competition between volume and surface effects , which is independent of the detailed interactions .
since the transition is of a mean - field type @xcite , such a thermodynamic description should be sufficient .
we consider a dilute solution enclosed in a membrane vesicle .
this system is in thermal and pressure contact with a much larger solution having temperature @xmath16 and pressure @xmath17 .
the system is at thermodynamic equilibrium .
thus , its temperature is @xmath16 .
however , since we deal with strong surface effects , the system s equilibrium pressure may deviate from the outer pressure @xmath17 .
the enclosed solution contains @xmath18 solute molecules.in the experiment described above the vesicle actually contains two solutes one that permeates the membrane and another that is trapped inside the vesicle .
analysis for an ideal solution , containing @xmath19 molecules of the former and @xmath20 molecules of the latter , leads to identical results once we take @xmath21 .
although @xmath18 in the experiment changes with time , we assume , because the solute has much lower permeability than water , that the system is at all times in quasi - equilibrium , given the instantaneous number of enclosed solute molecules , @xmath22 . due to the fast water exchange , we treat both the volume @xmath23 and surface area @xmath24 of the vesicle as free , independent thermodynamic variables .
alternatively , one can define a reduced volume , @xmath25 which characterizes the deviation of the vesicle from a perfect spherical shape , and consider @xmath26 and @xmath24 as the thermodynamic variables .
we divide the gibbs free energy of the system into three - dimensional ( volume ) and two - dimensional ( surface ) contributions : @xmath27 taking the enclosed solution to be ideal and dilute , we write its gibbs free energy as @xmath28 + { p_{\rm o}}v , \label{g3d}\ ] ] where @xmath29 is the molecular volume of water . as to the surface
, we distinguish between in - plane ( stretching ) and out - of - plane ( bending ) degrees of freedom , and assume that the vesicle is sufficiently large such that the two are decoupled , _
i.e. _ , contribute additively to the helmholtz free energy , @xmath30 we further assume that the membrane s in - plane free energy is minimum for a certain relaxed area , @xmath31 , and expand it to leading order about that minimum , @xmath32 where @xmath33 is the membrane s stretching modulus . in the bending free energy
we include only the contribution from undulation entropy , which , assuming small fluctuations about a spherical shape , is given by @xcite @xmath34 where @xmath35 is the total number of independent bending modes contributing to the membrane s thermodynamics .
equation ( [ bendentropy ] ) , which describes the suppression of bending fluctuations as the vesicle approaches the maximum volume - to - area ratio ( @xmath36 ) , is the only statistical - mechanical input to the theory .
it is analogous to an equipartition principle ; here each bending mode contributes an equal amount of @xmath37 to the free energy @xcite .
thus , this expression holds quite generally for nearly spherical vesicles , regardless of their detailed properties .
equation ( [ bendentropy ] ) also provides the crucial coupling between @xmath38 and @xmath39 , since @xmath26 depends on both @xmath23 and @xmath24 according to eq .
( [ v ] ) . the number of modes affecting the thermodynamics , @xmath35 , which should be proportional to the number of lipid molecules in the membrane ( _ i.e. _ , to the membrane s relaxed area @xmath31 ) , is _ a priori _ unknown .
further discussion of this number , its dependence on membrane s parameters , and its link to other theoretical treatments of membrane fluctuations , is deferred to sec .
[ sec_discuss ] . here
we prefer to leave it unspecified and extract it from the experiment .
the number of modes is used to define an intensive area , or effective patch size , @xmath40 , having the relaxed value @xmath41 .
one could add , of course , various other contributions to the free energy of the vesicle , such as the membrane s bending energy , surface surface and surface solute interactions ( _ e.g. _ , electrostatic ones ) , gravitational energy , _ etc._. all these , however , do not significantly change when the vesicle becomes spherical and begins to stretch . in other words , in the limit ( to be studied below ) where @xmath42 becomes singular , these contributions are non - singular , and , hence , can not affect the critical behavior in the transition region .
equations ( [ g])([bendentropy ] ) define the gibbs free energy of our finite - size system as a function of @xmath43 and also @xmath23 and @xmath24 .
the equilibrium free energy as a function of @xmath43 alone is obtained by minimizing @xmath44 with respect to the volume and area of the vesicle .
note that by using this procedure we circumvent altogether the subtle and controversial issue of membrane tension .
e.g. _ , refs . and references therein . )
the laplace tension and surface pressure of the membrane emerge naturally from the minimization with respect to @xmath23 and @xmath24 , respectively .
that is also the reason why we have formulated the helmholtz ( rather than gibbs ) free energy of the surface , to avoid specifying any thermodynamic variables pertaining to the surface other than @xmath16 , @xmath24 , and @xmath35 .
another related subtlety is that @xmath45 at fixed @xmath26 ( rather than fixed @xmath23 ) , otherwise there would be another contribution to the stretching modulus from changes in the volume - to - area ratio .
the minimization leads to the following two equations : @xmath46 from which one can calculate @xmath26 and @xmath47 . in eqs.([minimization1 ] ) and ( [ minimization2 ] ) we have defined @xmath48 along with two small parameters , @xmath49 the first related to the vesicle s finite size , and the other to the membrane s finite stretchability . before proceeding , it is instructive to examine eqs.([minimization1 ] ) and ( [ minimization2 ] ) in two limits .
the first is the ordinary thermodynamic limit of infinite system size ( @xmath50 , @xmath51 ) while keeping the stretching modulus finite . in this case
( [ minimization1 ] ) and ( [ minimization2 ] ) become degenerate and yield the expected equilibration of inner and outer pressures , @xmath52 .
the second limit is that of an unstretchable membrane ( @xmath53 , @xmath54 ) . in this case
( [ minimization2 ] ) imposes @xmath55 , and eq .
( [ minimization1 ] ) then yields the behavior that was studied in ref . , with criticality of @xmath26 as @xmath50 .
outside these two limits we get from these equations the relation @xmath56 which demonstrates the actual interplay between the volume - to - area ratio and surface strain .
we see that the vesicle can never attain a perfect spherical shape ( @xmath57 ) , because this would require unphysical strain ( @xmath58 ) .
the deviation of the area from its relaxed value is inversely proportional to the deviation of the shape from a perfect sphere .
this relation becomes more sensitive the smaller the value of @xmath59_i.e .
_ , the vesicle should be strongly swollen , @xmath60 , to get an appreciable strain @xmath61 . to present the detailed behavior of the transition
, we define a control parameter @xmath62 , proportional to the number of enclosed solute molecules , and an order parameter @xmath63 , measuring the deviation from a spherical shape , @xmath64 solving eqs .
( [ minimization1 ] ) and ( [ minimization2 ] ) in the vicinity of the transition , we get @xmath65^{1/2 } \nonumber\\ & & { \tilde{m}}(x ) = \left(\sqrt{1+x^2 } - x \right)/2 .
\label{scaling1}\end{aligned}\ ] ] thus , as @xmath62 increases from negative to positive values , @xmath63 crosses over from appreciable values , @xmath66 , to very small ones , @xmath67 .
the crossover occurs over a range of @xmath62 defined by @xmath68 . in the limit
@xmath69 the point @xmath70 becomes a singular corner where @xmath71 undergoes a discontinuous jump .
the finite size of the vesicle and the stretchability of the membrane both contribute to the broadening of the transition . if we take @xmath72 , eq .
( [ scaling1 ] ) properly coincides with the critical behavior in the case of an unstretchable membrane @xcite . which of the two broadening effects dominates in practice is determined by the ratio @xmath73 where the right - hand side reflects the competition between surface ( numerator ) and volume ( denominator ) effects .
substituting eq .
( [ scaling1 ] ) in eq .
( [ v_a_relation ] ) , we obtain for the surface strain , @xmath74 . yet , since the scaling function satisfies @xmath75^{-1}=4{\tilde{m}}(-x)$ ] , we can rewrite this result as @xmath76 } \delta { \tilde{m}}(-q/\delta ) . \label{scaling_epsilon}\ ] ] thus , up to a prefactor , the strain s behavior is a mirror image of that of @xmath63 .
as @xmath62 increases from negative to positive values , @xmath77 changes from very small values , @xmath78 , to appreciable ones , @xmath79 .
similarly , combining eqs.([minimization1 ] ) , ( [ v_a_relation ] ) , and ( [ scaling1 ] ) , we get for the laplace tension , @xmath80^{-1 } = k\epsilon(q ) , \label{gamma}\ ] ] _ i.e. _ , the surface tension follows the same behavior as the strain , becoming appreciable only for sufficiently large , positive @xmath62 .
equation ( [ gamma ] ) implies that within the transition region the tension is in the so - called linear regime @xcite , where it is proportional to the surface strain .
substituting the rupture strain found above , @xmath81 , and @xmath82 mn / m @xcite , we obtain a rupture tension of order @xmath83 mn / m , which is in line with previously reported values of @xmath84@xmath83 mn / m @xcite .
equation ( [ scaling1 ] ) [ equivalently , ( [ scaling_epsilon ] ) or ( [ gamma ] ) ] describes a law of corresponding states for osmotically swollen , nearly spherical vesicles .
it predicts that , upon rescaling , the swelling ( _ i.e. _ , volume - to - area ratio , strain , or tension ) of such vesicles , regardless of their size , composition , and the nature or concentration of solute , could be collapsed onto a single universal curve defined by the function @xmath85 .
thus , despite the many physical parameters affecting the process , only two parameters actually suffice to completely characterize the osmotic swelling of any nearly spherical vesicle the location of the transition , @xmath86 , and its width , @xmath68
. an interesting consequence of the competition between surface and volume effects is found when we examine the vesicle s swelling beyond the transition , well into the stretching stage . in this case , @xmath87 , eq . ( [ scaling_epsilon ] )
becomes @xmath88 upon replacing the surface strain of a spherical vesicle with its relative volume change , @xmath89 , and the outer pressure of a dilute solution with its concentration , @xmath90 , eq .
( [ epsilon_stretch ] ) reproduces a known result @xcite , which was used in earlier analyses of the stretching stage @xcite . as we have seen above , if @xmath91 , the finite stretching modulus determines the transition width @xmath68 [ eq . ( [ scaling1 ] ) ] . at the same time , according to eq .
( [ epsilon_stretch ] ) , the relatively small value of @xmath33 in this case makes the swelling beyond the transition insensitive to @xmath33 .
being governed by volume effects , the growth in the stretching stage is then simply proportional to the increasing number of enclosed molecules @xcite .
let us first check which of the broadening factors finite size or stretchability is dominant in the experiment .
the stretching modulus of a popc bilayer is @xmath92 mn / m @xcite .
our vesicles have @xmath93@xmath94 @xmath2 m , and the outer concentration is @xmath95@xmath96 m. we get @xmath97@xmath0 .
hence , the transition width is governed by the finite stretching modulus and is expected , therefore , to be independent of vesicle size , @xmath98 as noted above , this implies also that the stretching stage following the transition is insensitive to the value of @xmath33 , with @xmath99 . to further check the theoretical predictions of the preceding section
, we need to relate them to the experimental observables demonstrated in fig .
[ fig_swellingcurve]_i.e .
_ , the temporal change in the principal radius of an oblate spheroidal vesicle , @xmath100 .
we begin with the time axis .
the membrane permeability coefficients of the examined solutes are of order @xmath101 @xmath2 m/s @xcite , the vesicles are of radii @xmath102@xmath94 @xmath2 m , and the swelling process lasts about @xmath103@xmath1 s. we have @xmath104@xmath105 , implying that the concentration of permeating solute inside the vesicle never exceeds a few percent of its outer concentration .
hence , throughout the observed swelling we may assume @xmath106 , leading to @xmath107 thus , the time axis , up to a linear transformation that depends on vesicle size , is equivalent to our control parameter .
next , we should adapt the theoretical scaling relations , eqs.([scaling1 ] ) and ( [ scaling_epsilon ] ) , so that they could be applied to the measured swelling curves , @xmath100 .
this calculation is presented in the appendix . given the experimentally measured dependence , @xmath108 ^ 3 - 1\equiv{g_{\rm exp}}(t)$ ] , we find that the following scale and shift transformation : @xmath109 -x\delta \right.\nonumber\\ & & \left.+ 3(15^{1/3}/4)\delta^{2/3 } \right\ } , \label{collapse}\end{aligned}\ ] ] where @xmath110 and @xmath111 , should collapse the data onto the universal function @xmath112 the rescaling scheme defined in eq .
( [ collapse ] ) , which has been dictated by the nature of the experiment , is not as elegant as the theoretical law of corresponding states , eq .
( [ scaling1 ] ) . nevertheless , the two are equivalent and similarly straightforward .
we have applied the scheme to the data of fig .
[ fig_swellingcurves ] while using the permeability @xmath113 ( which merely scales the horizontal axis ) and the transition width @xmath68 as two fitting parameters .
the results are presented in fig .
[ fig_collapse ] , showing successful data collapse , within the transition region and above it , onto the predicted universal curve , eq .
( [ f ] ) .
theoretically , successful data collapse is expected in the vicinity of the transition .
we obtain data collapse also above the transition , throughout the stretching stage , for technical reasons explained in the appendix . below the transition ( _ i.e. _ , for sufficiently negative @xmath110 ) , the different swelling curves in fig .
[ fig_collapse ] depart from the universal behavior .
the entropy effects responsible for this departure , which set in at sufficiently low surface tension , will be described in sec .
[ sec_discuss ] . in fig .
[ fig_permeability ] we present the fitted values for the popc - membrane permeability coefficients of the three examined solutes .
the values for the permeability of urea [ fig.[fig_permeability](a ) ] are narrowly distributed and concentration - independent , yielding @xmath114 @xmath2 m/s .
this agrees well with the result of @xmath115 @xmath2 m/s , obtained for @xmath0 @xmath2m - radius dopc vesicles by paula _
_ @xcite using dynamic light scattering , despite the different phospholipid and the large difference in the sizes of the studied systems . for glycerol
we get values that significantly increase with glycerol concentration [ fig .
[ fig_permeability](b ) ] : @xmath116 , @xmath117 , and @xmath118 @xmath2 m/s for @xmath119 , @xmath120 , and @xmath96 m , respectively . the last value agrees with the value of @xmath121 @xmath2 m/s , extracted for the same vesicles , at @xmath122 m glycerol , from a different analysis of the stretch burst cycles that follow vesicle rupture @xcite .
a glycerol permeability of @xmath123 @xmath2 m/s was reported for @xmath0 @xmath2m - radius dopc vesicles by paula _
et al . _
@xcite and by dordas and brown @xcite , both at @xmath124 m. these results are not inconsistent with ours , especially given the concentration dependence .
for ethylene glycol we obtain more broadly distributed values that depend on concentration [ fig.[fig_permeability](b ) ] : @xmath125 and @xmath126 @xmath2 m/s for @xmath119 and @xmath96 m , respectively .
the only previously measured permeability for ethylene glycol that we are aware of is @xmath127 @xmath2 m/s , obtained by orbach and finkelstein @xcite using conductivity measurements for a flat egg - pc membrane . this value is larger than ours by an order of magnitude .
we note , though , the very different conditions under which the two experiments were conducted .
all permeability values reported here , including the concentration - dependent ones for glycerol and ethylene glycol , are consistent with those found from the alternative analysis of stretch burst cycles ( unpublished results ) .
we now turn to the second fitting parameter the transition width , @xmath68 .
it is found to lie in the range @xmath128@xmath4 and be uncorrelated with the vesicle size ( fig .
[ fig_delta ] ) , as predicted above .
the small values of @xmath68 establish the self - consistency of our analysis for the crossover between the two swelling stages as a slightly broadened phase transition , and thus explain the success of the resulting data collapse .
the rather broad distribution of these values imply that the transition width is affected by a stochastic variable . of the three parameters appearing in eq .
( [ delta])@xmath16 , @xmath33 , and @xmath129only the effective patch size @xmath129 may be responsible for such stochasticity . in fig .
[ fig_delta](b ) we show the distribution of patch sizes , as arising from eq .
( [ delta ] ) using the measured @xmath68 and @xmath130 mn / m .
the mean patch size is @xmath131 @xmath2m@xmath132 , which is much smaller than the vesicle area , but much larger than the molecular size .
such a patch contains about @xmath133@xmath134 lipids . in the inset of fig .
[ fig_delta](b ) we recast the same data in terms of the number of effective bending modes , @xmath135 .
the number of modes is broadly distributed in the range @xmath134@xmath136 , making its increasing trend with vesicle area hardly discernible .
the experimental results presented above confirm the critical nature of the osmotic swelling of vesicles , as was first suggested in ref .. it is important to note that a similar critical scaling is absent in the case of hydrostatic swelling , as in vesicle aspiration experiments @xcite . to adapt our formulation to such a case
we should merely replace the solute contribution to the gibbs free energy ( the first term in eq . ( [ g3d ] ) ) , with @xmath137 , where @xmath138 is the inner hydrostatic pressure . repeating the minimization procedure presented in sec.[sec_theory ]
, we find that eq .
( [ v_a_relation ] ) remains unchanged , while the swelling order parameter becomes @xmath139 thus , our formalism applies in this case only for sufficiently large inner pressures , @xmath140 ( to ensure a positive @xmath26 ) .
similar to @xmath141 , we may define a swelling control parameter , @xmath142 , yet its value is restricted to @xmath143 .
consequently , eq . ( [ mpin ] ) yields a moderate decrease of @xmath63 from appreciable to small values , @xmath144_i.e .
_ , a gradual approach to a spherical shape .
another way to demonstrate the non - criticality of the hydrostatic - pressure case is to recognize that , once both @xmath17 and @xmath138 are specified , the surface tension of a nearly spherical vesicle is simply given by laplace s law rather than the more elaborate critical expressions of eqs .
( [ scaling1])([gamma ] ) .
this underlines the special thermodynamic behavior of osmotically stressed vesicles , as was indicated before in a more general statistical - mechanical context @xcite .
the key difference between osmotically swollen and directly pressurized vesicles lies in the fact that the pressure difference of the former depends on the volume for a given number of encapsulated solute molecules [ cf .
( [ g3d ] ) ] , whereas the pressure difference of the latter is a given constraint .
this makes the statistical ensembles of @xmath43 and @xmath145 not equivalent @xcite .
we have shown how the data collapse for different vesicle sizes can be utilized to obtain a sensitive measurement of the membrane permeability coefficients of various solutes .
this ability is particularly well demonstrated in the case of urea ( fig .
[ fig_permeability](a ) ) .
overall , the permeability values that we have obtained are in accord with previously published ones .
( see the detailed account in the preceding section and fig .
[ fig_permeability](b ) . ) in addition , we have found an increasing concentration dependence of the permeability for glycerol and ethylene glycol . while we are not aware of an earlier report of such an effect , it is in line with strong evidence for the affinity of polyols to lipid headgroups @xcite and may account for the large range of values reported in the literature for the permeability coefficients of these solutes .
the affinity should make the solute concentration adjacent to the membrane larger than in the bulk solution , leading to an apparent permeability coefficient which is larger than the actual one .
more extensive measurements of the apparent permeability as a function of concentration might allow an extraction of the solute membrane affinity parameters . at higher concentrations
the solute may also disrupt the membrane structure , thus affecting the actual permeability @xcite and possibly other membrane properties @xcite . the transition width , and the corresponding effective membrane patch @xmath129 ,
have been found to be broadly distributed ( fig.[fig_delta ] ) .
the origin of this stochasticity is not yet clear .
it may be that @xmath129 is sensitive to the presence of membrane impurities .
nevertheless , the analysis of transition widths has enabled us to obtain information regarding the number of independent bending modes , @xmath35 , which contribute to the thermodynamics of the vesicle in the transition region . to our knowledge
this variable has not been experimentally accessed before . for the vesicles used here ,
having @xmath146@xmath94 @xmath2 m , we have found numbers @xmath147@xmath136 , which are large but much smaller than the number of lipid molecules in the vesicle ( about @xmath148 ) . to account for these intermediate values of @xmath35
we should link our thermodynamic formulation to the more detailed statistical mechanics of fluctuating vesicles . the definition of @xmath35 in sec .
[ sec_theory ] has been through the surface entropy of a nearly spherical vesicle , eq .
( [ bendentropy ] ) , where each of the @xmath35 modes contributes an equal amount of @xmath37 to the free energy @xcite . examining more closely the partition function of a fluctuating vesicle within helfrich s model @xcite
, we find that these modes , whose contribution is singular in @xmath42 , belong to the small - wavenumber ( tension - dominated ) portion of the fluctuation spectrum .
their number is @xmath149 where @xmath150 is the membrane s bending rigidity .
equations ( [ scaling_epsilon ] ) , ( [ gamma ] ) , and ( [ n ] ) then allow a theoretical estimate of @xmath35 . at the transition we have @xmath151 ( cf .
[ fig_collapse ] ) , which for @xmath92 mn / m @xcite and @xmath152 ( fig . [ fig_delta](a ) ) yields @xmath153 mn / m .
this value of about 1 mn / m is in line with that obtained for nearly spherical vesicles in aspiration experiments @xcite . substituting it , together with @xmath154 @xcite and @xmath155@xmath94 @xmath2 m , in eq .
( [ n ] ) , we get @xmath156 , which is in the range of values that we have found experimentally . equation ( [ n ] ) implies , in fact , that @xmath35 is not constant but increases with the degree of swelling ( _ i.e. _ , with @xmath157 ) . hence , sufficiently far below the transition @xmath157 will be too small , and the theory should fail .
( far beyond the transition , on the other hand , the behavior is dominated by surface strain and unaffected by these entropy considerations . ) combining eqs .
( [ gamma ] ) and ( [ n ] ) , we find that the theory should break down for @xmath158 . with @xmath152
this occurs for @xmath159 , which is consistent with the region where the data collapse begins to fail ( see fig .
[ fig_collapse ] ) .
thus , the assumption of a swelling - independent @xmath35 is the main limitation of the model and should be improved in future studies .
other assumptions are less significant . for example
, the relaxed area of the vesicle , @xmath31 , is actually slightly smaller than the assumed value of @xmath160 , where @xmath161 is the minimum of the experimental @xmath100 curve .
the calculation given in the appendix implies that @xmath162 .
for our broadest transitions this deviation amounts to less than 3% . as shown in ref .
, accounting for additional factors such as a non - ideal solute does not affect the universal behavior in the vicinity of the transition .
we thank oded farago , rony granek , and luka pocivavsek for helpful comments .
eh and hd acknowledge the donors of the american chemical society petroleum research fund for support of this research ( grant no .
46748-ac6 ) .
in this appendix we adapt the theoretical scaling relation obtained in sec .
[ sec_theory ] to the observables in the experimental swelling curves .
the result is the scaling transformation that should be applied to the experimental data to achieve data collapse .
consider an oblate spheroid of principal radii @xmath8 and @xmath163 , and small eccentricity @xmath164^{1/2}\ll 1 $ ] .
the volume of the spheroid is @xmath165 , and its area is @xmath166\}$ ] . from this
we get @xmath167 , leading to @xmath168 , and @xmath169 .
now we have , on the one hand , @xmath170 , and , on the other hand , @xmath171 . equating these two expressions while using the result above for @xmath172 ,
we obtain @xmath173 the correction to this expression is @xmath174 . in the transition region
this correction , as well as @xmath77 , are negligible ( @xmath175 ) compared to the @xmath176 term ( @xmath177 ) . above the transition , at the point where @xmath77 becomes comparable to @xmath176 , we find @xmath178 , whereas @xmath179 .
thus , eq . ( [ scaling2 ] )
contains the leading terms within the transition region as well as above it . using eqs.([scaling1 ] ) , ( [ scaling_epsilon ] ) , and the property @xmath180 , we rewrite eq .
( [ scaling2 ] ) as @xmath181 where we have omitted another term of order @xmath63 and also the ratio @xmath182 . equation ( [ scaling3 ] ) , as required , contains the experimental observable @xmath183 rather than our order parameter @xmath63 .
yet , we need to take one last step before we can apply it to the experimental data .
the expression in eq .
( [ scaling3 ] ) has a minimum at @xmath184 $ ] , which is shifted from the one that we have defined for the experimental curve , @xmath185 $ ] ( cf.fig .
[ fig_swellingcurve ] ) . in other words ,
the actual relaxed radius is slightly shifted from the minimum of the measured swelling curve . to correct for this @xmath186 error , we ought to shift the experimental curves accordingly . if @xmath108 ^ 3 - 1={g_{\rm exp}}(t)$ ]
is the measured dependence , as a function of @xmath62 , then the following scaling and shifting transformation , @xmath187 -x\delta \right.\nonumber\\ & & \left . +
\ 3(15^{1/3}/4)\delta^{2/3 } \right\ } \nonumber\\ & & x = q/\delta,\ \ \ q=(3p / r_0)t,\end{aligned}\ ] ] should collapse the data onto the universal function @xmath188 defined in eq .
( [ scaling3 ] ) .
this is the transformation used in sec.[sec_analysis ] . | as solute molecules permeate into a vesicle due to a concentration difference across its membrane , the vesicle swells through osmosis . the swelling can be divided into two stages : ( a ) an `` ironing '' stage , where the volume - to - area ratio of the vesicle increases without a significant change in its area ; ( b ) a stretching stage , where the vesicle grows while remaining essentially spherical , until it ruptures .
we show that the crossover between these two stages can be represented as a broadened continuous phase transition .
consequently , the swelling curves for different vesicles and different permeating solutes can be rescaled into a single , theoretically predicted , universal curve .
such a data collapse is demonstrated for giant unilamellar popc vesicles , osmotically swollen due to the permeation of urea , glycerol , or ethylene glycol .
we thereby gain a sensitive measurement of the solutes membrane permeability coefficients , finding a concentration - independent coefficient for urea , while those of glycerol and ethylene glycol are found to increase with solute concentration .
in addition , we use the width of the transition , as extracted from the data collapse , to infer the number of independent bending modes that affect the thermodynamics of the vesicle in the transition region . |
rrrccc crrcc helms 0015@xmath60404 & 3.9286&@xmath64.0715 & [email protected] & [email protected] & [email protected]&26.5 & 3.93038&@xmath64.07262 & [email protected] & [email protected] + helms 0041@xmath80410 & 10.3541&@xmath84.1679 & [email protected] & [email protected] & [email protected]&15.8 & & & @xmath935.1 & + l6-xmm 0223@xmath80605 & 35.8056&@xmath86.0860 & [email protected] & [email protected] & [email protected]&69.7 & & & @xmath914.4 & + g09 0918@xmath80039 & 139.6159&@xmath80.6644 & [email protected] & [email protected] & [email protected]&18.9 & & & @xmath917.7 & + g09 0920@xmath60024 & 140.2475&@xmath60.4049 & [email protected] & [email protected] & [email protected]&22.9 & & & @xmath918.6 & + ngp 1313@xmath62924 & 198.4530&@xmath629.4126 & [email protected] & [email protected] & [email protected]&17.2 & & & @xmath942.3 & + ngp 1330@xmath62540 & 202.5866&@xmath625.6749 & [email protected] & [email protected] & [email protected]&18.0 & & & @xmath915.3 & + ngp 1333@xmath62357 & 203.3743&@xmath623.9592 & [email protected] & [email protected] & [email protected]&55.2&203.37514&@xmath623.95909 & [email protected] & [email protected] + ngp 1335@xmath62805 & 203.9409&@xmath628.0986 & [email protected] & [email protected] & [email protected]&27.6&203.94249&@xmath628.09750 & [email protected] & [email protected] + g15 1413@xmath60058 & 213.4580&@xmath60.9725 & [email protected] & [email protected] & [email protected]&16.2&213.45743&@xmath60.97321 & [email protected] & [email protected] + g15 1435@xmath60110 & 218.9043&@xmath61.1682 & [email protected] & [email protected] & [email protected] & 9.2&218.90494&@xmath61.16958 & [email protected] & [email protected] + g15 1450@xmath60026 & 222.6773&@xmath60.4351 & [email protected] & [email protected] & [email protected]&16.3 & & & @xmath918.0 & + l6-fls 1712@xmath66001 & 258.0352&@xmath660.0281 & [email protected] & [email protected] & [email protected]&30.1&258.03111&@xmath660.02722 & [email protected] & [email protected] + & & & & & & & 258.04010 & @xmath660.02625 & [email protected] & [email protected] the first milli - jansky - level submillimeter surveys discovered a population of distant submillimeter - bright galaxies ( smgs ) , namely , unresolved sources with 850 @xmath10 m flux density ( @xmath11 ) greater than 3@xmath85 mjy @xcite .
the smgs selected at wavelengths between 850 @xmath10 m and 1 mm are intense starbursts ( sfr @xmath12 @xmath13 ) at a median redshift of @xmath14 @xcite .
the intense star formation is dust - enshrouded so that the smgs radiate most of their bolometric luminosity in the far - infrared ( ir ) .
the observed molecular and stellar emission indicates that they are massive gas - rich galaxies ( @xmath15@xmath2 ; e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but the typical halo mass of smgs remains uncertain , with estimates ranging from @xmath16 to @xmath17@xmath2 .
two lines of evidence suggest that smgs may inhabit dark matter halos as massive as @xmath1 @xmath2 : ( 1 ) their strong clustering strength estimated from either the angular two - point correlation function ( e.g. , * ? ? ?
* ; * ? ? ?
* ) or the cross - correlation function between smgs and other high - redshift galaxies ( e.g. , * ? ? ?
* ) , and ( 2 ) their high stellar mass and the @xmath18@xmath8@xmath19 relation from abundance matching ( @xmath19 @xmath20 @xmath2 for @xmath18 @xmath21 @xmath2 at @xmath22 ; e.g. , * ? ? ?
however , because source blending due to the large beams of single - dish ( sub)millimeter telescopes may have significantly elevated clustering strength @xcite and the stellar mass estimates remain uncertain within an order of magnitude ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , it is possible that a typical smg may inhabit much less massive halos ( @xmath010@xmath23@xmath2 ) .
smgs are absent in the local universe and it is commonly thought that they have evolved into the massive ellipticals today ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . to understand the evolution of smgs , it is imperative to know how long the observed intense star formation would last . for @xmath17 @xmath2 halos at @xmath24 ,
the average baryonic accretion rate from the mass growth rate of dark matter haloes is @xmath25 @xmath13 @xcite . in such massive halos
, it is expected that most of the baryons will be shock - heated to the virial temperature of the halo ( @xmath26 k ) so that only a small fraction of the accreted gas can actually cool and accrete onto galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
therefore , the ongoing gas accretion is unlikely to sustain the extreme sfrs . without a comparable gas supply rate ,
the sfr would decline with an @xmath27-folding timescale of only @xmath0200 myr ( 2@xmath28/sfr ; e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
at such a rate , the smgs would become red sequence galaxies in only a gyr or 5 @xmath27-folding times ) from @xmath29 gyr@xmath30 for the smgs at the observed epoch to @xmath31 gyr@xmath30 for the red sequence at @xmath5 @xcite . ] .
such a short transitional time of a significant high - redshift star - forming population might help explain the rapid build - up of the massive end of the red sequence at @xmath32 ( e.g. , * ? ? ?
starbursts thus provide an alternative mechanism to the qso - mode feedback ( e.g. * ? ? ?
* ; * ? ? ?
* ) to form red and dead galaxies .
note that both mechanisms still require feedback from radio jets ( i.e. , the maintenance mode ; e.g. , * ? ? ?
* ; * ? ? ?
* ) to prevent the hot gaseous halo from cooling . on the other hand ,
if smgs were in @xmath16 @xmath2 halos , the intense star formation is also unsustainable because the gas accretion rate is only @xmath0110 @xmath13 at @xmath24
. however , could there be enough cool gas in the circumgalactic medium ( cgm ) around smgs to fuel a prolonged starburst phase ( e.g. see the simulation of * ? ? ?
the cgm of co - eval qsos may give us a hint , because they inhabit comparably massive ( @xmath33 ) halos @xcite .
contrary to the expected dominance of virialized x - ray plasma , absorption line spectroscopy of a statistical sample of @xmath5 projected qso pairs reveals the prevalence of cool ( @xmath34 k ) , metal - enriched ( @xmath35 ) , and optically thick ly@xmath3 absorbers ( @xmath36 @xmath37 ) extending to at least the expected virial radius of 160 kpc ( the `` qso probing qso '' [ qpq ] project : @xcite ) .
the high observed covering factor of the cool cgm gas ( @xmath38 ) in @xmath33 halos has been compared to predictions from numerical simulations .
while several studies found that they can not reproduce the high covering factor around qsos ( @xcite , but see @xcite ) , it has been argued that efficient star - formation - driven winds from accreted satellite galaxies that interact with cosmological filaments are required to increase the hi covering factor to the observed level , which is only resolved in the highest resolution cosmological zoom simulations @xcite .
these high - resolution simulations predict that the covering factor is roughly independent of sfr , decreases with redshift , and has relatively large halo - halo variations . in the relevant halo mass range , @xmath39@xmath2 ,
simulations show hi covering factors of @xmath40 with little mass dependence at @xmath5 . given
the estimated halo masses for qsos and smgs , one could expect similar covering factors if these are determined primarily by the interplay between gas infall and star - formation - driven outflows . to test this
, we exploit qso absorption line spectroscopy to probe the cgm of smgs .
we first present the data sets and the method we used to select projected smg@xmath8qso pairs in [ sec : sample ] .
we then describe our followup observations in [ sec : obs ] , including radio interferometer imaging , near - infrared spectroscopy , and optical spectroscopy .
we present our analysis and results in [ sec : result ] , including a comparison between the covering fraction of optically thick gas around smgs and that of @xmath5 qsos .
we summarize the results and conclude in [ sec : summary ] . throughout
we adopt a @xmath41cdm cosmology with @xmath42 , @xmath43 and @xmath44 = 70 km s@xmath30 mpc@xmath30 .
because both high - redshift qsos and smgs have low number counts , we need large samples of both to come up with a sizable sample of projected smg@xmath8qso pairs with small angular separations .
we compiled 464,866 spectroscopically confirmed qsos from various surveys : primarily , the sloan digital sky survey ( sdss ; * ? ? ?
* ) , the 2df qso redshift survey ( 2qz ; * ? ? ?
* ) , the agn and galaxy evolution survey ( ages ; * ? ? ?
* ) , and the mmt hectospec redshift survey of 24 @xmath10 m sources in the spitzer first look survey @xcite . since we select the foreground galaxies that are likely at @xmath45 ( see the next paragraph ) , we keep only the 102,472 qsos at @xmath46 .
the average surface density of these background qsos is @xmath010deg@xmath47 . to select the foreground smgs , we combined source catalogs from a number of wide - area extragalactic surveys carried out by the _
space observatory @xcite : the _ herschel _ multi - tiered extragalactic survey ( hermes , 95 deg@xmath48 ; * ? ? ?
* ; * ? ? ?
* ) , the _ herschel _ astrophysical terahertz large area survey ( h - atlas , 600 deg@xmath48 ; * ? ? ? * ; * ? ? ?
* ) , the _ herschel _ large mode survey ( helms , 301 deg@xmath48 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; clarke et al . in prep .
) , the _ herschel _ stripe 82 survey ( hers , 79 deg@xmath48 ; * ? ?
all of these surveys used spire ( spectral and photometric imaging receiver ; * ? ? ?
* ) to image the sky at 250 , 350 , and 500 @xmath10 m and the combined xid250 m fluxes were all extracted at source positions detected on the 250 @xmath10 m map ( e.g. , * ? ? ?
* ; * ? ? ?
] catalog contain 1,586,047 sources covering a total of 767 deg@xmath48 ( hers and helms fields overlap by 10deg@xmath48 ) . however , most _ herschel _ sources are not smgs ; they are , instead , less luminous dusty star - forming galaxies at lower redshifts ( @xmath49 ; * ? ? ?
* ; * ? ? ?
* ) . to select _ herschel _ sources that are likely _ bona fide _ smgs , we chose only the subsample that satisfies the following criteria : ( 1 ) flux densities peak at 350 @xmath10 m ( @xmath50 and @xmath51 ; i.e. , `` 350@xmath10 m peakers '' ) , ( 2 ) @xmath52 mjy , and ( 3 ) @xmath533@xmath54 detections in all three spire bands .
criterion 1 is essentially a photometric redshift selection because emission from dusts at @xmath55 k would peak at 350 @xmath10 m if redshifted to @xmath14 .
in addition , the blind carbon monoxide ( co @xmath56 ) survey of the brightest 350 @xmath10 m peakers ( @xmath57 mjy ) has shown a strikingly similar redshift distribution as 850 @xmath10m - selected smgs ( @xmath58 ; * ? ? ? * ) , even though most of these bright sources are strongly lensed .
criterion 2 is introduced to ensure that the rayleigh - jeans extrapolation would give @xmath59 mjy , the classic definition of an smg , given a typical power - law slope of 3.5 for a modified blackbody with a frequency - dependent absorption cross section ( @xmath60 ) .
criterion 3 ensures that all of the sources we considered are statistically significant .
this is necessary because the image depth varies substantially from field to field , ranging from @xmath61mjybeam@xmath30 for the large helms and hers fields ( confusion noise included ; * ? ? ?
* ; * ? ? ?
* ) to confusion limited with @xmath62mjybeam@xmath30 for the deeper hermes fields @xcite . of course , our crude color plus flux selection is not completely reliable , but it has weeded out most of the dusty star - forming galaxies that are not smgs .
only 70,823 _ herschel _ sources remained after this selection .
the average surface density of 92deg@xmath47 is five times lower than the observed 870 @xmath10 m source count above @xmath63 mjy ( @xmath0500deg@xmath47 ; * ? ? ?
* ; * ? ? ?
this is not surprising given that almost half of the total _
herschel _
area is only 10 - 20% complete at @xmath64 mjy .
note that this incompleteness is not a concern for our purpose .
we identified 230 projected smg@xmath8qso pairs with angular separations ( @xmath65 ) is measured between the _ herschel _
250@xmath10 m position and the optical position of the qso . ] between 5 @xmath66 36 by cross - matching the qso and smg subsamples described above .
the corresponding impact parameters ( @xmath67 ) are between @xmath68 300 kpc for @xmath69 .
the impact parameter is defined as the transverse proper distance at the redshift of the foreground smg , which equals the angular diameter distance of the smg multiplied by the angular separation on sky ( @xmath70 ) .
these qso sightlines thus probe out to @xmath01.5 virial radii of @xmath17 @xmath2 halos ( @xmath71 kpc at @xmath24 ) .
given the positional uncertainty of the _ herschel _ 250 @xmath10 m sources , the 5 lower limit on the angular separation is imposed as an attempt to avoid far - ir - luminous qsos , i.e. , the qsos are the smgs themselves . through visual inspection of the qso spectra , we further excluded 31 pairs whose qsos exhibit strong broad absorption lines ( bals ; this makes them unsuitable for absorption line work ) , have wrong redshifts , or are misclassified .
therefore , our final sample includes 199 pairs , among which 90/163 qsos have sdss @xmath72 ( bright enough for absorption line spectroscopy ) .
one _ herschel _ source is probed by two qsos at @xmath738.0 and 18.2 .
no qso probes multiple _ herschel _ sources within 36 , but a `` single '' _ herschel _ source may consists of multiple smgs due to the large beam size ; so it is possible that a single qso could probe multiple smgs at different impact parameters ( this is indeed the case for the last source shown in fig . [ fig : detections ] ) .
extensive followup observations are needed to perform the absorption - line analysis of smgs .
identifying the absorption features in the spectrum of the background qso requires a precise redshift for the foreground smg .
but the full - width - at - half - maximum ( fwhm ) angular resolution of _ herschel _ 18/25/36 at 250/350/500@xmath10 m precludes longslit spectroscopic observations with typical slit width of 1 . to obtain more accurate positions and to identify blended sources , we exploit interferometer observations with
the karl g. jansky very large array ( vla ) . furthermore ,
once the positions are determined with sub - arcsec accuracy , we need to determine the spectroscopic redshifts of the smgs with near - ir spectrographs by targeting rest - frame optical lines that suffer less dust extinction than rest - frame uv lines .
finally , a high s / n optical / near - uv spectrum of the background qso is needed to detect the uv absorption lines imprinted by the diffuse medium around the smgs .
below we describe these observations in more details .
far - ir - luminous galaxies like our _ herschel _ sources are expected to be luminous in the radio wavelengths , according to the ir - radio correlation @xcite .
we can thus obtain better positions for the _ herschel _ sources by identifying the radio counterparts with interferometers .
we observed 15 smg@xmath8qso pairs with the vla in the b configuration with the c - band ( 6 ghz ) receivers ( program i d : 15a-266 ) .
the sample was selected randomly from the smg@xmath8qso pairs with qsos brighter than @xmath74 , excluding those with @xmath7530 .
we later realized that two of the vla targets in the helms field are likely spurious detections because of galactic cirrus ( clarke et al . in prep . ) , so we excluded them in this discussion .
table [ tab : vlaphoto ] lists the _ herschel _ positions and photometry of the final vla sample .
the receivers have a total bandwidth of 4ghz at a central frequency of 5.9985ghz .
the targets were selected from six different extragalactic fields . to maximize the observing efficiency , we grouped the targets with their r.a . into five scheduling blocks ( sbs ) of 0.8 to 3.1hours .
a nearby unresolved calibrator was observed every @xmath010 min .
depending on the r.a .
of the targets , 3c48 , 3c286 , or 3c295 was observed for bandpass and flux - density calibration .
the entire program took 8.3 hrs of vla time .
the on - source integration time ranges from 9 to 70 minutes , allocated based on the 6 ghz flux density estimated from fitting the _ herschel _ photometry with the sed template of a well - studied , strongly lensed smg at @xmath76 ( smm j2135 - 0102 , aka .
the eyelash ; * ? ? ? * ) .
the observations were calibrated using the common astronomical software applications ( casa ) package @xcite .
we used the vla pipeline to perform basic flagging and calibration .
additional flagging was performed whenever necessary by inspecting the visibility data .
we used the self - calibration technique with bright sources within the primary beam to reduce the gain errors for four fields : helms0015@xmath60404 , helms0041@xmath80410 , l6-xmm0223@xmath80605 , and ngp1313@xmath62924 . for imaging and deconvolution
, we used the standard casa task clean with `` natural '' weighting to achieve the best sensitivity .
the resulted restoring beams are on average 15@xmath7712 fwhm .
the rms image noise level ranges from 2.1 to 14.1 @xmath10jybeam@xmath30 , with a mean of 6 @xmath10jybeam@xmath30 .
this measured noise is consistent with thermal noise using natural weighting of the visibility data , except for maps contaminated by the sidelobes of the strong sources lying far outside the cleaned image .
we will present the results in [ sec : radiodetections ] , but here is a summary : the vla detected six of the 13 smgs ; for one of those , ngp 1335 + 2805 , the qso at @xmath78 itself is responsible for the ir emission , so we excluded it from subsequent spectroscopic observations . with the vla positions that are accurate to @xmath790.1
, we carried out a redshift survey for the five vla - detected smgs with near - ir spectrographs .
the vla resolved the _ herschel _ source of l6-fls1712@xmath66001 into two sources at similar brightness ( fig .
[ fig : detections ] _ last panel _ ) .
we chose the one closer to the _ herschel _ position as the nominal counterpart and have obtained a deep near - ir spectrum at that location .
it is unclear whether the two sources are physically related because the redshift of the other source remains to be determined .
we observed g151435@xmath60110 , l6-fls1712@xmath66001 , and ngp1333@xmath62357 with the luci-1 spectrograph ( lbt nir - spectroscopic utility with camera and integral - field unit ; * ? ? ? * ) on the large binocular telescope ( lbt ) on 2015 apr 14 .
we used the 200 l / mm @xmath80+@xmath81 grating at @xmath82 @xmath10 m and the n1.8 camera ( 025 per pixel ) to obtain a spectral range between 1.5 and 2.3 @xmath10 m .
the 1-wide 4-long slit is centered on the vla - determined smg position , and it is aligned with the qso in each pair to obtain the qso spectrum simultaneously . the qso spectrum is bright enough to serve as a useful reference for spatial alignment . the spectral resolution ( @xmath83 ) is @xmath0940 in @xmath80-band and @xmath01290 in @xmath81-band .
we obtained 32@xmath77 120 s exposures for each target . between exposures , we dithered along the slit among six dithering positions distributed within 40 .
atmospheric transparency varied dramatically during the night and no telluric star was observed .
so we used the telluric star observation from the previous night for an approximate telluric and flux calibration .
we observed helms0015@xmath60404 and g151413@xmath60058 with the gemini near - infrared spectrograph ( gnirs ; * ? ? ?
* ) in the queue mode ( program ids : gn-2015b - q-46 , gn-2016a - q-41 ) .
we used the cross - dispersing prism with the 31.7 l / mm grating and the short camera to obtain a complete spectral coverage between 0.85 - 2.5 @xmath10 m . with the 068 short slit , the spectral resolution is @xmath84 across all orders .
we obtained 14@xmath77 300 s exposures for helms0015@xmath60404 on 2015 aug 9 , aug 12 , and oct 18 , and 24@xmath77 115 s exposures for g151413@xmath60058 on 2016 may 20 .
we dithered by 3 along the 7-slit between exposures .
data reduction was carried out with a modified version of ` longslit_reduce ` @xcite for luci-1 by fuyan bian @xcite and a modified version of spextool @xcite for gnirs by katelyn allers ( private communication ) .
these idl packages carry out the standard data reduction steps for near - ir spectroscopy : flat - fielding , wavelength calibration , pairwise sky subtraction , residual sky removal , shifting and coadding , spectral extraction , telluric correction , and flux calibration .
although previous optical spectra exist for all of the qsos in our sample , most of them do not have the necessary wavelength coverage or sufficient s / n for absorption line analysis .
hence we obtained new optical spectra for the three background qsos that are associated with the spectroscopically identified smgs .
we observed helms0015@xmath60404 and ngp1333@xmath62357 on 2016 jan 9 and l6-fls1712@xmath66001 on 2015 jun 13 with the low resolution imaging spectrometer ( lris ; * ? ? ?
* ) on the keck i telescope .
we used the 1 longslit and the 560 dichroic for both runs . for the former run
, we used the 600/4000 grism on the blue side ( @xmath85 at 4,000 , fwhm = 300 km s@xmath30 ) and the 600/7500 grating at @xmath86 on the red side to cover from 3,300 to 8,700 . for the latter run
, we used the 400/3400 grism on the blue side ( @xmath87 at 4,000 , fwhm = 500 km s@xmath30 ) and the 400/8500 grating tilted to a central wavelength of @xmath88 on the red side to cover from 3,300 to 10,200 .
the total integration time for each source ranged between 30 min and 40 min .
conditions were non - photometric for both nights .
we reduced the raw data with ` xidl ` , an idl data reduction package for a number of spectrographs written by two of us ( jxp and jfh ) .
the pipeline follows the standard data reduction steps and reduces the blue and red channels separately .
it begins by subtracting a super bias from the raw ccd frames , tracing the slit profiles using flat fields , and deriving the 2d wavelength solution for each slit using the arcs .
then it flat - fields each slit and rejects cosmic - rays , identifies objects automatically in the slit , and builds the bspline super sky model without rectification @xcite . after subtracting the super sky model
, it performs optimal 1d extraction based on the spatial profile of the qso @xcite .
finally , it removes instrument flexure using isolated sky lines , performs heliocentric correction , and does flux calibration .
we targeted the _ herschel _ sources in 13 smg@xmath8qso pairs with the vla and made six clear detections ( 46% detection rate ) .
figure [ fig : detections ] shows the vla detections and table [ tab : vlaphoto ] lists the 6 ghz positions and flux density measurements for the detections and 3@xmath54 upper limits on the non - detections .
the vla revealed five genuine smg@xmath8qso pairs ( helms0015@xmath60404 , ngp1333@xmath62357 , g151413@xmath60058 , g151435@xmath60110 , l6-fls1712@xmath66001 ) and one far - ir - luminous qso ( ngp1335@xmath62805 ) .
the vla - detected smgs have 6 ghz peak flux densities between 11 and 76 @xmath10jy with a mean of @xmath040 @xmath10jy . for these detections
, there is a clear positive correlation between the observed flux densities at 6 ghz and 500 @xmath10 m , as expected from the ir - radio correlation convolved with a redshift distribution .
the 54% non - detection rate is not surprising given the uncertainties in the predicted 6 ghz flux density and the fact that the sensitivity of radio data may deviate substantially from the theoretical prediction because of confusion sources and weather . the offsets between the _ herschel _ positions and the vla positions range between 2.7 and 7.9 .
the maximum offset is comparable to the estimated 2@xmath54 positional uncertainty of _ herschel _ 250 @xmath10m - selected catalogs : @xmath89 for fwhm = 18.1 and snr = 5 @xcite .
this indicates that a fraction of the `` pairs '' that have small angular separations ( @xmath90 ) could be single sources , i.e. , qsos that are far - ir - luminous . in our vla - detected sample ,
ngp1335@xmath62805 , with @xmath91 , is the only such case ( see the third panel of fig . [
fig : detections ] ) .
the vla position agrees with the optical qso position within 01 , which is well within the cross - band astrometry accuracy of @xmath00.3 ( sdss@xmath6vla ; e.g. , * ? ? ?
but on the other hand , the smg in l6-fls 1712@xmath66001 is a separate source from the qso , despite its even smaller 250 @xmath10 m separation ( @xmath92 ) .
we thus chose not to exclude pairs with @xmath93 .
the far - ir - luminous qsos in our sample are interesting on their own right and they will be discussed elsewhere . rccccc ccccc helms 0015@xmath60404 & j001543.29 + 040421.4 & 2.515&13.1 & 2.1&j001542.31 + 040433.5 & 3.256 & 19.0 & 157&@xmath94 & ambiguous + ngp 1333@xmath62357 & j133330.03 + 235732.7 & 2.184&12.6 & 2.3&j133330.36 + 235709.9 & 3.108 & 23.3 & 198&@xmath95 & ambiguous + l6-fls 1712@xmath66001 & j171207.47 + 600138.0 & 2.043&12.6 & 2.6&j171209.00 + 600144.4 & 2.821 & 13.1 & 112&@xmath96 & optically thin estimating the intrinsic properties of
the vla - detected _ herschel _ sources require knowledge of their spectroscopic redshifts .
we obtained deep near - ir spectra for the five vla counterparts whose spectroscopic redshifts are previously unknown .
we detected h@xmath3 and [ nii ] lines from three of the five sources , enabling accurate determination of the spectroscopic redshifts ( fig .
[ fig : smgspec ] and table [ tab : nhi ] ) .
the [ nii]/h@xmath3 line ratios are consistent with the range observed in typical smgs ( @xmath97 [ nii]/h@xmath98 ; @xcite ) . the two remaining sources , g151413 + 0058 and g151435 + 0110 , show near - ir continuum emission without detectable emission lines or stellar absorption features .
it is possible that the emission lines either fall into one of the telluric absorption bands or are simply outside of the spectral range .
our redshift success rate is thus 60% , comparable to previous redshift surveys of smgs in the optical ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . with the photometry from _ herschel _ and the vla and spectroscopic redshifts
, we can estimate the total rest - frame 8@xmath81000 @xmath10 m luminosity ( @xmath99 ) and the ir - to - radio luminosity ratio ( @xmath100 ) .
the results are listed in table [ tab : nhi ] .
we fit the _ herschel _ photometry with a modified blackbody at the spectroscopic redshift , with @xmath101 fixed to 1.5 .
we find that they have @xmath102@xmath103 and sfr@xmath104 = 470 - 1500 @xmath13 .
the sfr is estimated from @xmath99 using the calibration of @xcite for a @xcite initial mass function : @xmath105 the extrapolated 850 @xmath10 m flux densities range between @xmath106 mjy , confirming that they are _ bona fide _ smgs .
the ir - to - radio luminosity / flux ratio is estimated using the following equation @xcite : @xmath107 where @xmath108 is rest - frame integrated 8 - 1000 @xmath10 m flux and @xmath109 is the rest - frame 1.4 ghz flux density .
the former is from the modified blackbody fit to the ir sed , and the latter is converted from the observed 6 ghz flux densities assuming a typical synchrotron slope ( @xmath110 ) .
the three smgs show @xmath111 , consistent with the observed radio - ir correlation for _ herschel _ sources ( @xmath112 ; * ? ? ?
this indicates that our sample is not biased to radio - loud agns despite being radio selected . to compare with previous work on the cgm of other high - redshift galaxies , we need to characterize the covering fraction of optically thick gas clouds around smgs ( i.e. , lyman limit systems [ llss ] with column densities of neutral hydrogen @xmath113
@xmath37 ) .
we search for ly@xmath3 absorbers near the spectroscopic redshifts of the foreground smgs and classify the ly@xmath3 absorbers in the qso spectra following the procedure used in the qpq study ( e.g. , * ? ? ?
the classification is based on the ly@xmath3 equivalent widths ( ews ) and the presence of associated metal transitions .
we first identify hi ly@xmath3 absorption due to the smgs in the qso spectra within @xmath7600 km s@xmath30 of the systemic redshifts of the smgs .
the search window is chosen because the escape velocity is 610 km s@xmath30 at the virial radius ( @xmath114 kpc ) of a @xmath17 @xmath2 dark matter halo at @xmath22 @xcite .
strong ly@xmath3 absorption lines are detected in all three systems ( fig .
[ fig : abs ] ) .
we report their rest - frame ews ( @xmath115 ) , updated angular separations based on the vla positions ( @xmath116 ) , and the impact parameters ( @xmath67 ) in table [ tab : nhi ] . as we have found in our earlier qso absorption line studies , systematic error associated with continuum placement and line blending
generally dominates the statistical error of the qso spectra .
so we estimate the error of @xmath115 assuming 10% error in the continuum placement . from our best - fit gaussians to the absorption profiles ( shown in fig .
[ fig : abs ] ) , we find that all three systems show strong ly@xmath3 absorption lines with @xmath117 .
note that when @xmath118 , the hi ly@xmath3 absorption from the smg may lie within the qso ly@xmath101 forest .
unfortunately , all three pairs fall in this category , so contamination from ly@xmath101 lines from systems in the ly@xmath3 forest may be a concern .
however , it turned out that ly@xmath101 contamination is not a serious issue for these systems . because assuming the detected ly@xmath3 absorption to be ly@xmath101 , we searched for the corresponding ly@xmath3 lines in each spectrum but did not find any .
given the absorption profiles of the ly@xmath3 line , we then searched for the associated metal transitions commonly observed in optically thick absorption systems ( e.g. , * ? ? ? * ) : siii@xmath1191260,1304,1527 , oi@xmath1191302 , cii@xmath1191335 , siiv@xmath1191394,1403 , civ@xmath1191548,1551 , feii@xmath1191608,2383,2600 , alii@xmath1191671 , and mgii@xmath1192796,2804 .
however , none of these metal transitions is convincingly detected in any of our systems . in the following ,
we discuss the systems individually and provide our classifications : * _ helms0015@xmath60404_. this system shows an hi ly@xmath3 absorption at @xmath6500 km s@xmath30 with @xmath120 and _ intrinsic _ fwhm = 430 km s@xmath30 ( fig .
[ fig : abs ] _ top panels _ ) . given this measurement ,
one may have concluded that this is an optically thick absorber with @xmath121 @xmath37 if the absorption is dominated by a single component .
the column density is estimated from the theoretical curve of growth where @xmath122 for @xmath123 @xmath37 , the regime where the relation is insensitive to the @xmath124-parameter @xcite .
there is putative cii absorption at @xmath6400 km s@xmath30 ( @xmath125 ) ; but it is likely a false identification of a ly@xmath3 forest line , because it is misaligned in velocity with the ly@xmath3 absorption and none of the strong metal lines in the `` clean '' region outside of the lyman forests is detected ( e.g. , siii , civ , feii , alii ) . without detections of strong metal lines and/or the damping wings in ly@xmath3 , we can not conclude that the absorber is optically thick based on current data because the high @xmath115 could be due to line blending .
we conservatively classify this system as ambiguous , even though the non - detection of strong metal lines strongly suggests that it is an optically thin absorber , because fewer than 5% of optically thick absorbers have no detection of a metal transition @xcite . * _ ngp1333@xmath62357_.
this system shows an hi ly@xmath3 absorption at @xmath126 km
s@xmath30 with @xmath127 and intrinsic fwhm = 260 km s@xmath30 ( fig .
[ fig : abs ] _ middle panels _ ) .
we observe putative absorption around the expected locations of cii , siiv , siii , and civ . but
all of these lines lie within the ly@xmath3 forest , and their velocity profiles are distinctly different from that of the ly@xmath3 line , indicating that these putative metal absorption are spurious .
the feii and alii lines are outside of the lyman forests , but the expected absorption lines are not detected . like the previous case , we conservatively classify this system as ambiguous . * _ l6-fls1712@xmath66001_. there is a broad but shallow hi ly@xmath3 absorption line at @xmath128 km s@xmath30 with @xmath129 and intrinsic fwhm = 950 km s@xmath30 ( fig .
[ fig : abs ] _ bottom panels _ ) . despite the high ew ,
this system is unlikely optically thick because the line center drops only to 40% of the continuum intensity ( i.e. , it does not go line black ) . like the previous cases
, we observe putative metal absorption inside the lyman forests ( e.g. , oi , cii , siii , and siiv ) but no metal absorption outside of the forests ( e.g. , civ , feii , alii , and mgii2796,2804 lines . ] ) . note that the weak civ absorption is blueshifted by more than 200 km s@xmath30 from the ly@xmath3 line center .
hence , we classify this system as optically thin . in summary ,
we have identified one clearly optically thin case and two ambiguous cases with three qso sightlines covering impact parameters between @xmath130 kpc around smgs at @xmath131 .
the two ambiguous cases are also likely to be optically thin , given the non - detection of any metal transitions .
for background qso sightlines probing the cgm of @xmath132 foreground qsos , one observes hi absorbers optically thick at the lyman limit @xmath13360% of the time .
the high hi covering fraction extends to at least the expected virial radius of @xmath0160 kpc @xcite . in figure
[ fig : covfrac ] , we compare ( 1 ) the smg sample distribution with the qsos from the qpq project , and ( 2 ) the covering fraction of optically thick hi gas around smgs with that around qsos .
the two samples overlap in the plane of foreground redshift vs. impact parameter : the smgs cover the same redshift range as the qsos in the intermediate impact parameter range between 100 and 200 kpc .
we calculate the 1@xmath54 binomial confidence intervals of the optically thick fraction using the quantiles of the beta distribution @xcite . because there is no clearly optically thick absorber among the three systems we analyzed , our 1@xmath54 confidence interval of the covering fraction is [email protected]% for a non - detection in a sample of three . for comparison
, we consider all of the qso sightlines with @xmath130 kpc from the qpq project @xcite and we find the optically thick covering fraction is @xmath134% for 21 clearly optically thick systems among 33 systems . therefore , despite our small sample , the upper bound of our 1@xmath54 confidence interval is 3@xmath54 below the best - estimated covering fraction around qsos at the same ranges of redshifts and impact parameters . note that although our analysis may appear to have limitations based on the classification of two of the smg absorbers as ambiguous , these same limitations also apply to the qpq analysis , and are inherent to any attempt to classify absorbers using low or moderate resolution spectra without coverage of the hi lyman limit ( at 912 in the rest - frame ) . as such ,
we have followed exactly the same procedure for absorber classification as in the qpq studies , enabling a direct comparison to that work .
note , however , there are some minor differences between our analysis and the qpq analysis which are compared in fig .
[ fig : covfrac ] .
first , the bulk of the qpq dataset @xcite utilized background qso spectra with moderate resolution ( @xmath135 ) .
this is a factor of two higher than the resolution we used for the qso spectra probing two of our smgs ( @xmath85 ) , whereas in l6-fls1712@xmath66001 our spectrum has @xmath87 .
but the lower resolution of our spectra does not play a significant role in the resulting classification of our absorbers , since the typical strength of the metal lines seen in the optically thick systems would still be easily detectable at @xmath85 , and the one system that we observed at @xmath87 is unambiguously optically thin , since it does not go line black .
another significant difference , is whereas @xcite restricted analysis to only those foreground qsos with redshifts lying within the ly@xmath3 forest of the background qso , we have included foreground smgs landing in the ly@xmath101 forest , owing to the larger redshift separations between the smgs and the qsos in our sample .
this is unlikely to impact our deduced covering factor in fig .
[ fig : covfrac ] though . for example
, earlier qpq studies @xcite also considered absorbers lying in the ly@xmath101 forest , and the covering factor deduced there was consistent with that derived in @xcite .
furthermore , in all of the three smg@xmath8qso pairs shown in fig .
[ fig : abs ] , multiple metal lines that are strong in optically thick absorbers ( i.e. , siii@xmath1191527 , civ@xmath1361548,1551 , aliii@xmath1191671 , and mgii@xmath1362796,2804 ) land in the clean regions outside of the lyman forest but remain undetected . finally , whereas qpq searched for absorbers within a velocity window in a @xmath71500 km s@xmath30 window owing to the large errors in qso redshifts , we adopted a @xmath7600 km s@xmath30 search window .
we were able to consider a smaller velocity interval because our smg redshifts derived from narrow rest - frame optical lines are much more accurate , and this interval was instead chosen to encompass the typical virial motions in a @xmath17@xmath2 dark matter halo .
the larger velocity interval used in qpq formally implies a higher level of contamination from physically unassociated absorbers .
but the contamination is estimated to be at only @xmath03% level for a @xmath71500 km s@xmath30 window ( * ? ? ?
* see their figure 10 and table 7 ) .
this is too small to be responsible for the difference in the optically thick covering factor that we observe between qsos and smgs .
motivated by the unique properties of smgs and their purported evolutionary link to high - redshift qsos and today s massive ellipticals , we have started a project to use qso absorption line spectroscopy to probe the diffuse cool hi gas in the cgm of smgs .
this work requires a sample of projected smg@xmath8qso pairs , which are extremely rare .
wide - area sub - millimeter surveys are needed to compile such a sample .
thanks to the advent of _ herschel _ , we have identified 163 smg@xmath8qso pairs with bright @xmath137 qsos ( @xmath138 ) and angular separations between 5 and 36 from a suite of wide - area _
herschel _ surveys and spectroscopic qso surveys .
extensive followup observations are required to carry out the absorption line study . to allow slit spectroscopy ,
the first stage is to use an interferometer to pin down the positions of the _ herschel _ sources .
this paper focuses on a subsample of 13 smg@xmath8qso pairs that were observed with the vla in c - band . with an average integration time of 25 min per source , the vla detected sources within the _ herschel _ beam in six fields .
one of the six smg@xmath8qso pair turns out to be a far - ir - luminous qso at @xmath78 , while the _ herschel _ source in another pair turns out to be a pair of smgs .
hence , we effectively identified six smg@xmath8qso pairs from observations of 13 fields ( 46% ) .
the second stage is to measure the spectroscopic redshifts of the smgs .
we observed five of the six vla - identified smg@xmath8qso pairs with near - ir spectroscopy .
we were able to determine the redshifts for three of the five smgs ( 60% ) from the redshifted h@xmath3 and [ nii ] lines .
the remaining two sources show only featureless continuum in the near - ir windows , making it impossible to determine an accurate redshift .
the last stage is to obtain optical spectroscopy of the qsos once it is confirmed that the qsos are in the background of the smgs . because we have selected only the brightest qsos , the success rate of this step is approaching 100% once bad qsos are removed from the sample . because of the low success rates of the first two followup stages ,
only @xmath023% of the initial sample ( i.e. , three smg@xmath8qso pairs out of 13 ) are spectroscopically confirmed and are suitable for final absorption line study .
our main findings are as follows : 1 .
the near - ir spectra of the five vla - detected sources are all detected in hour - long integrations with 8-meter telescopes , although only three of those show emission lines that yielded accurate spectroscopic redshifts .
since we positioned the slits on the vla positions , this result shows that the spatial offset , if any , between the near - ir and the radio counterparts is less than an arcsec ( the slit width ) , consistent with the finding from differential lensing in strongly lensed sources ( e.g. , * ? ? ?
the vla - identified _ herschel _ `` 350 @xmath10mpeakers '' at @xmath131 are similar to smgs selected at longer wavelengths ( i.e. , 850 @xmath10 m to 1 mm ) , in terms of the [ nii]/h@xmath3 ratio ( i.e. , gas metallicity ) , the ir luminosity ( i.e. , sfr ) , and the ir - to - radio luminosity ratio ( i.e. , radio excess due to agn ) .
the vla - identified smgs are optically faint and unbiased to radio - loud agns , so they indeed represent a galaxy population distinct from the optical selected qsos and the lyman break galaxies ( lbgs ) in the same redshift range .
strong hi ly@xmath3 absorption is found in the background qso spectra for all of the three spectroscopically confirmed smg@xmath8qso pairs with impact parameters between @xmath130 kpc .
here we have adopted a much narrower search window ( @xmath7600 km s@xmath30 ) than the qpq study , further reducing the level of contamination from physically unrelated clouds .
however , none of the three absorption line systems seems optically thick at the lyman limit ( i.e. , llss with @xmath113 @xmath37 ) , in contrast to the @xmath060% covering fraction of llss around qsos from the qpq study despite similar data quality , foreground redshifts , and impact parameters .
our comparison thus suggests _ either _ that smgs do not have a substantial neutral gas reservoir in their halos that could potentially fuel a prolonged star formation phase _ or _ that smgs inhabit @xmath139@xmath2 halos so that our sightlines are yet to probe inside their virial radii . if the latter , their halos are comparable to those of lbgs .
@xcite found an optically thick covering fraction of @xmath140% around lbgs at @xmath141 and @xmath142 kpc .
note that this is @xmath02@xmath77 lower than that of coeval qsos and is consistent with the 1@xmath54 confidence interval that we were able to place for the smgs . on the other hand ,
the difference in the optically thick hi covering fraction between smgs and qsos casts doubt on the evolutionary link between the two populations , unless agn outflows can somehow affect the physical state of gas at hundreds of kpc scales within its short lifetime .
our final conclusion is limited by the small sample size . to enable a more robust comparison with previous absorption - line studies , we badly need to increase the effective yield of our survey from the current level of @xmath023% . in a future publication , we will present observations with the atacama large millimeter / submillimeter array ( alma ) to pinpoint the positions of the _ herschel _ sources in the smg@xmath8qso pairs . observing at a wavelength ( 870 @xmath10 m ) much closer to the selection wavelengths ( 250 to 500 @xmath10 m ) , we expect doubling the detection rate with integration times of just several minutes per source
herschel _
sources in our sample are too faint to allow a quick co redshift search with alma ( e.g. , the survey of strongly lensed smgs by * ? ? ?
* ) , but spectrographs on large optical telescopes covering the full optical@xmath6ir range at a moderate spectral resolution ( @xmath143 ) will greatly increase the redshift search range and decrease the areas blinded by strong airglow lines .
we thank f. bian and k. allers for providing the data reduction packages , and d. ludovici for helping with the vla data calibration of l6-xmm0223@xmath80605 .
the national radio astronomy observatory is a facility of the national science foundation ( nsf ) operated under cooperative agreement by associated universities , inc .
support for this work was provided by the nsf through award gssp sospa3 - 016 from the nrao .
acknowledges support from the nsf grant ast-1614326 , the nasa jpl award 1495624 , and funds from the university of iowa . j.x.p . acknowledges support from the nsf grants ast-1010004 , ast-1109452 , ast-1109447 and ast-1412981 .
a.c . acknowledges support from the nsf grant ast-1313319 and the nasa grants
nnx16af39 g and nnx16af38 g . d.k . acknowledges support from the nsf grant ast-1412153 and funds from the university of california san diego .
z.y.z . acknowledges support from the european research council in the form of the advanced investigator programme , 321302 , cosmicism .
herschel_-atlas is a project with _ herschel _ , which is an esa space observatory with science instruments provided by european - led principal investigator consortia and with important participation from nasa .
the h - atlas website is http://www.h - atlas.org/. the us participants acknowledge support from the nasa _
_ science center / jpl .
the lbt is an international collaboration among institutions in the united states , italy and germany .
lbt corporation partners are : the university of arizona on behalf of the arizona university system ; istituto nazionale di astrofisica , italy ; lbt beteiligungsgesellschaft , germany , representing the max - planck society , the astrophysical institute potsdam , and heidelberg university ; the ohio state university , and the research corporation , on behalf of the university of notre dame , university of minnesota and university of virginia .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation . based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the national research council ( canada ) , conicyt ( chile ) , ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) , and ministrio da cincia , tecnologia e inovao ( brazil ) .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain . , j. p. , waters , b. , schiebel , d. , young , w. , & golap , k. 2007 , in astronomical society of the pacific conference series , vol .
376 , astronomical data analysis software and systems xvi , ed .
r. a. shaw , f. hill , & d. j. bell , 127 | we present the first results from an on - going survey to characterize the circumgalactic medium ( cgm ) of the massive high - redshift galaxies detected as submillimeter galaxies ( smgs ) .
we constructed a parent sample of 163 smg - qso pairs with separations less than @xmath036 by cross - matching far - infrared - selected galaxies from _ herschel _ with spectroscopically confirmed qsos .
the _
herschel _ sources were selected to match the properties of smgs .
we determined the sub - arcsecond positions of six _ herschel _ sources with the very large array and obtained secure redshift identification for three of those with near - infrared spectroscopy . the qso sightlines probe transverse proper distances of 112 , 157 , and 198 kpc at foreground redshifts of 2.043 , 2.515 , and 2.184 , respectively , which are comparable to the virial radius of the @xmath1 @xmath2 halos expected to host smgs .
high - quality absorption - line spectroscopy of the qsos reveals systematically strong hi ly@xmath3 absorption around all three smgs , with rest - frame equivalent widths of @xmath4 .
however , none of the three absorbers exhibits compelling evidence for optically thick hi gas or metal absorption , in contrast to the dominance of strong neutral absorbers in the cgm of luminous @xmath5 qsos .
the low covering factor of optically thick hi gas around smgs tentatively indicates that smgs may not have as prominent cool gas reservoirs in their halos as the co - eval qsos and that they may inhabit less massive halos than previously thought . |
the study of the kuiper belt has exploded since the discovery of 1992 qb1 by @xcite .
the brightness distribution of objects with @xmath2 magnitude brighter than @xmath026 is relatively well - established by many surveys , most recently by ( * ? ? ?
* and references therein ) .
the brightness distribution is adequately described by a simple cumulative luminosity function , where @xmath3 and @xmath4 , for objects with magnitude @xmath5 .
there is clear evidence for a break to a shallower slope for fainter objects : the deepest survey , conducted using the advanced camera for surveys on the _ hubble space telescope _
@xcite extended to @xmath6 , and found a factor of @xmath025 fewer objects than would be expected if the same distribution extended into this range .
the size distribution of kuiper belt objects ( kbos ) is believed to reflect a history of _ agglomeration _ during the planetary formation epoch , when relative velocities between particles were low and collisions typically resulted in particles sticking together , followed by _ destructive collisions _ when the relative velocities were increased by dynamical processes after the giant planets formed @xcite .
the slope of the distribution function for larger objects reflects the early phase of agglomeration , while the shallower distribution for smaller objects reflects a subsequent phase of destructive collisions .
the location of the break moves to larger sizes with time , while the distribution for smaller objects is expected to evolve towards a steady state collisional cascade @xcite .
models for the spectrum of small bodies differ between @xcite , who derived a double power - law distribution , and @xcite , whose simulations show more structure , depending on material properties . thus , the size spectrum encodes information about the history of planet formation and dynamics .
however , the size spectrum for small kbos is not constrained by the imaging surveys because the objects of interest are too faint for direct detection using presently available instruments .
these small objects may , however , be detected indirectly when they pass between an observer and a distant star @xcite .
the challenge confronting any survey exploiting this technique is the combination of very low anticipated event rate and short duration of the events ( typically @xmath7 1 second ) .
other groups are attempting similar occultation surveys .
@xcite reported three events in 10 star - hours of photometric data sampled at 45 hz , which they modeled as objects at 15 au , 140 au , and 210 au , respectively , placing the inferred objects outside the kuiper belt .
@xcite reported results of 5 star - hours of data sampled at 40 hz , during which no events were detected .
@xcite reported a surprisingly high rate of possible occultation events in _ rxte _ x - ray observations of sco x-1 , but many of these events have since been attributed to instrumental effects @xcite .
we report here the first results of the taiwanese american occultation survey ( taos ) .
taos differs from the previously reported projects primarily in the extent of the photometric time series , a total of @xmath8 star - hours , and in that data are collected simultaneously with three telescopes .
some compromises have been made in regard to signal - to - noise ( snr ) , which is typically lower than in previously reported surveys , and in cadence , which is 4 hz or 5 hz , in contrast to the higher rates mentioned above .
the substantial increase in exposure more than compensates for the lower cadence and snr , and we are able to probe significant ranges of the model space for small kbos . we have also developed a statistical analysis technique which allows efficient use of the multi - telescope data to detect brief occultation events that would be statistically insignificant if observed with only one telescope .
taos has been collecting scientific data since 2005 .
observations are normally carried out simultaneously with three 50 cm telescopes ( a , b , and d , separated by distances of 6 meters and 60 meters ; this system is described by @xcite ) . over 15 tb of raw images have been taken .
we report here on the first two years of data taken simultaneously with all three telescopes .
the data set comprises 156 _ data runs _ , where a data run is defined as a series of three - telescope observations of a given field for durations of @xmath090 minutes .
thirty data runs with a 4 hz sampling rate were taken before 2005 december 15 , and 126 data runs were collected subsequently with a sampling rate of 5 hz . only fields with ecliptic latitudes @xmath9 were analyzed . over 93% of the data
were collected in fields with @xmath10 , so the results of our analysis are relevant to the sum of the cold and excited kbo populations @xcite .
no data run was included unless each star was sampled more than 10,000 times in each telescope .
the angle from opposition in these data runs is distributed from @xmath11 to @xmath12 .
the number of stars ( with @xmath13 , which typically gives a snr @xmath14 ) monitored in the data runs ranges between 200 and 2000 .
the images were analyzed using an aperture photometry package @xcite devised exclusively for taos images .
lightcurves were produced for each star by assembling the photometric information into a time series .
a star in each data run has a lightcurve from each of the three telescopes .
the data presented in this paper comprises 110,895 _ lightcurve sets _ ( where a lightcurve set is defined as a set of three lightcurves , one for each telescope , for the same star in a single data run ) , containing @xmath15 individual photometric measurements .
the photometric data are not calibrated to a standard system .
changes in atmospheric transparency during a data run produce flux variations ( ) that could undermine our occultation search algorithm .
such low - frequency trends in a lightcurve can be removed by a numerical high - pass filter that preserves information of a brief occultation event , typically with a duration less than 12 data points with the taos sampling rate .
our filter takes a time series of @xmath16 measured at time @xmath17 to produce an intermediate series @xmath18 where @xmath19 is the running average of 33 data points centered on @xmath20 .
the series @xmath21 is then scaled by the local fluctuation , @xmath22 with @xmath23 being the standard deviation of @xmath21 of 151 data points centered at @xmath20 .
both the mean and standard deviation are calculated using _ three - sigma clipping_. this filtering proves effective to remove slow - varying trends in the lightcurve , while preserving high - frequency fluctuations that we aim to detect , as illustrated in .
we now confront the two central challenges in the search for extremely rare occultation events in these data : _ ( 1 ) to search for events simultaneously in three parallel data streams , _ and _ ( 2 ) to determine the statistical significance of any rare events that are found . _
the second is not straightforward because the statistical distribution of our photometric measurements is not known in advance ; approximations based on gaussian statistics are unreliable far from the mean .
this motivates a non - parametric approach
. we thus found it useful to represent each data point by its _
rank _ in the filtered , rescaled lightcurve data @xmath24 . that is , the rank of a data point ranges from @xmath25 ( lowest @xmath26 ) to @xmath27 ( highest @xmath26 ) , for a data run comprising @xmath28 photometric measurements taken with telescope a , b , or d. the _ rank triples _ @xmath29 form the basis of further analysis of the multi - telescope data .
the statistical distribution of these ranks is known exactly , since each rank must occur exactly once in each time series .
thus , the probability that a given rank will occur at time @xmath17 is @xmath30 .
when the photometric data are uncorrelated , the probability that a particular rank triple will occur is simply @xmath31 .
this allows a straightforward test for correlation between the photometric data taken in the three telescopes : the rank triples should be distributed uniformly in a cube with sides of length @xmath28 .
a non - uniform pattern in the cube , on the other hand , indicates correlation .
shows an example of the rank series of one telescope against another ; this indicates that the raw photometric data @xmath32 are strongly correlated , but the filtered data @xmath24 are not .
given that the rank triples are uncorrelated , the ranks can be used to search for possible occultation events , as follows : _ a true occultation event will exhibit anomalous , correlated low ranks in all three telescopes . _
the rank triples thus allow an elegant test for the statistical significance of a candidate event .
consider the quantity @xmath33 since the ranks are uncorrelated ( unless we have an occultation event ) , we can calculate the exact probability density function , this distribution can be approximated by a @xmath34 distribution of the form @xmath35 , where @xmath36 is the number of telescopes . ] for @xmath37 .
this in turn allows us to compute the probability that a given triple of low ranks randomly occurred in an uncorrelated lightcurve set ; this is our measure of statistical significance .
an illustration of the power of this approach is shown in , where a simulated occultation event is readily recovered .
we thus screened all of our series of rank triples for events with low ranks in all three telescopes . in the analysis reported here
, we considered only events for which @xmath38 , which leads to an expected value of @xmath39 false positive events in the entire data set of @xmath40 rank triples . _
_ no statistically significant events emerged from this analysis . _ _ , which did not pass our cut on @xmath37 .
we expect to have @xmath01 false positive event at that significance level or higher . ]
while the example event shown in is readily recovered , objects with smaller diameters or which do not directly cross the line of sight might not be so easily detected .
an efficiency calculation is thus necessary for understanding the detection sensitivity of our data and analysis pipeline to different event parameters , notably the kbo size distribution .
our efficiency test started with implanting synthetic events into observed lightcurves , with the original noise .
each original lightcurve @xmath32 was modified by the implanted occultation , @xmath41 where @xmath42 is the simulated event lightcurve ( with baseline @xmath43 far from the event , @xcite ) , and @xmath19 is the average of the original series over a 33 point rolling window . since we preserved the original noise in the modified lightcurves , the noise where the implanted occultation event takes place for which the flux diminishes would be slightly overestimated , hence our efficiency estimate is conservative .
we assumed spherical kbos at a fixed geocentric distance of @xmath44 au .
( given our sampling rate , varying the kbo distance within the kuiper belt has little effect on our simulated lightcurves . )
the event epoch @xmath45 was chosen randomly and uniformly within the duration @xmath46 of the lightcurve set .
the angular size @xmath47 of each star , necessary for the simulated lightcurve calculation , was estimated using stellar color and apparent magnitude taken from the usno - b @xcite and 2mass @xcite catalogs .
the impact parameter of each event was chosen , again randomly and uniformly , between 0 and @xmath48 , where @xmath49 is the _ event cross section _
@xcite , @xmath50^{{2 \over 3 } } + \theta_*\delta.\ ] ] here @xmath51 is the diameter of the occulting object , and @xmath52 is the _ fresnel scale _ , @xmath53 , where @xmath54 nm is the median wavelength in the taos filter .
the relative velocity between the earth and kbo @xmath55 , necessary for the conversion of the occultation diffraction pattern to a temporally sampled lightcurve , is calculated based on the angle from opposition during each data run . to adequately cover a wide range of parameter space ,
two efficiency runs were completed in which we implanted each lightcurve set with exactly one simulated occultation event . for each event ,
the diameter of the kbo was chosen randomly according to a probability , or weighting factor , @xmath56 . in the first run
, objects of diameters @xmath57 0.4 , 0.5 , 0.6 , 0.7 , 0.8 , and 0.9 km were added with @xmath58 for each diameter . in the second run , objects of diameters @xmath59 1 , 2 , 3 , 5
, 10 and 30 km were added with weights @xmath60 .
the modified lightcurves @xmath61 were reprocessed using the same procedure described in .
the recovered events and the event parameters were then used to calculate the number of expected occultation event in our survey .
that is , we calculated the quantity @xmath62,\ ] ] where the sum is over all lightcurve sets where a simulated event is successfully recovered in the reanalysis ( ) .
essentially @xmath63 is the _ effective solid angle _ of our survey , insofar as taos can be considered equivalent to a survey that is capable of counting every kbo of diameter @xmath51 in a solid angle @xmath64 with 100% efficiency .
the expected number of detected events by kbos with sizes ranging from @xmath65 to @xmath66 can then be written as @xmath67 where @xmath68 is the differential surface number density of kbos .
the integrand of contains two factors : the _ model - dependent _ size distribution @xmath68 , and the _ model - independent _ effective solid angle @xmath63 , which describes the sensitivity of the survey to objects of diameter @xmath51 .
given the observed number of events and the value of @xmath63 resulting from the efficiency calculation , we can place model - dependent limits on the the population of kbos .
based on the absence of detections in this data set , any model with a size distribution such that @xmath69 is inconsistent with our data at the 95% confidence level .
note that there are an infinite number of models that satisfy the above requirement .
we thus make the reasonable choice of a power - law size distribution @xmath70 , where @xmath71 is chosen such that the cumulative size distribution is continuous at 28 km with the results of @xcite .
we integrate from @xmath72 km down to our detection limit of @xmath73 km , and solve with @xmath74 , to find @xmath75 .
our null detection thus eliminates any power law size distribution with @xmath76 at the 95% c.l . , setting a stringent upper limit ( see ) to the number density of kbos .
we have surveyed the sky for occultations by small kbos using the three telescope taos system . we have demonstrated that a dedicated occultation survey using an array of small telescopes , an innovative statistical analysis of multi - telescope data , and a large number of star - hours , can be used as a powerful probe of small objects in the kuiper belt , and we are thus able to place the strongest upper bound to date on the number of kbos with @xmath77 km .
we continue to operate taos , soon with an additional telescope , and will report more sensitive survey results in the future .
work at ncu was supported by the grant nsc 96 - 2112-m-008 - 024-my3 .
work at the cfa was supported in part by the nsf under grant ast-0501681 and by nasa under grant nng04g113 g .
work at asiaa was supported in part by the thematic research program as-88-tp - a02 .
work at ucb was supported by the nsf under grant dms-0405777 .
work at yonsei was supported by the krcf grant to korea astronomy and space science institute .
work at llnl was performed under the auspices of the u.s .
doe in part under contract w-7405-eng-48 and contract de - ac52 - 07na27344 .
work at slac was performed under u.s .
doe contract de - ac02 - 76sf00515
. work at nasa ames was funded by nasa / p.g.&g . | results from the first two years of data from the taiwanese - american occultation survey ( taos ) are presented .
stars have been monitored photometrically at 4 hz or 5 hz to search for occultations by small ( @xmath03 km ) kuiper belt objects ( kbos ) .
no statistically significant events were found , allowing us to present an upper bound to the size distribution of kbos with diameters 0.5 km @xmath1 28 km . |
ebola hemorrhagic fever is a highly infectious and lethal disease named after a river in the democratic republic of the congo ( formerly zaire ) where it was first identified in 1976 @xcite .
twelve outbreaks of ebola have been reported in congo , sudan , gabon , and uganda as of september 14 , 2003 @xcite .
two different strains of the ebola virus ( ebola - zaire and the ebola - sudan ) have been reported in those regions . despite extensive search ,
the reservoir of the ebola virus has not yet been identified @xcite .
ebola is transmitted by physical contact with body fluids , secretions , tissues or semen from infected persons @xcite .
nosocomial transmission ( transmission from patients within hospital settings ) has been typical as patients are often treated by unprepared hospital personnel ( barrier nursing techniques need to be observed ) .
individuals exposed to the virus who become infectious do so after a mean incubation period of @xmath9 days ( @xmath10 days ) @xcite .
ebola is characterized by initial flu - like symptoms which rapidly progress to vomiting , diarrhea , rash , and internal and external bleeding .
infected individuals receive limited care as no specific treatment or vaccine exists .
most infected persons die within @xmath11 days of their initial infection @xcite ( @xmath12 mortality @xcite ) .
+ using a simple seir ( susceptible - exposed - infectious - removed ) epidemic model ( figure [ myfig0 ] ) and data from two well - documented ebola outbreaks ( congo @xmath1 and uganda @xmath2 ) , we estimate the number of secondary cases generated by an index case in the absence of control interventions ( @xmath3 ) . our estimates of @xmath3 are @xmath4 ( sd @xmath5 ) for congo ( 1995 ) and @xmath6 ( sd @xmath7 ) for uganda ( 2000 ) .
we model the course of the outbreaks via an seir epidemic model that includes a smooth transition in the transmission rate after control interventions are put in place .
we also perform an uncertainty analysis on the basic reproductive number @xmath3 to account for its sensitivity to disease - related parameters and analyze the model sensitivity of the final epidemic size to the time at which interventions begin .
we provide a distribution for the final epidemic size . a two - week delay in implementing public health measures results in an approximated doubling of the final epidemic size .
we fit data from ebola hemorrhagic fever outbreaks in congo ( 1995 ) and uganda ( 2000 ) to a simple deterministic ( continuous time ) seir epidemic model ( figure [ myfig0 ] ) .
the least - squares fit of the model provides estimates for the epidemic parameters .
the fitted model can then be used to estimate the basic reproductive number @xmath3 and quantify the impact of intervention measures on the transmission rate of the disease . interpreting the fitted model as an expected value of a markov process
, we use multiple stochastic realizations of the epidemic to estimate a distribution for the final epidemic size .
we also study the sensitivity of the final epidemic size to the timing of interventions and perform an uncertainty analysis on @xmath3 to account for the high variability in disease - related parameters in our model .
individuals are assumed to be in one of the following epidemiological states ( figure [ myfig0 ] ) : susceptibles ( at risk of contracting the disease ) , exposed ( infected but not yet infectious ) , infectives ( capable of transmitting the disease ) , and removed ( those who recover or die from the disease ) .
susceptible individuals in class @xmath13 in contact with the virus enter the exposed class @xmath14 at the per - capita rate @xmath15 , where @xmath16 is transmission rate per person per day , @xmath17 is the total effective population size , and @xmath18 is the probability that a contact is made with a infectious individual ( i.e. uniform mixing is assumed ) . exposed individuals undergo an average incubation period ( assumed asymptomatic and uninfectious ) of @xmath19 days before progressing to the infectious class @xmath20 .
infectious individuals move to the @xmath21-class ( death or recovered ) at the per - capita rate @xmath22 ( see figure [ myfig0 ] ) .
the above transmission process is modeled by the following system of nonlinear ordinary differential equations @xcite : + @xmath23 where @xmath24 , @xmath25 , @xmath26 , and @xmath27 denote the number of susceptible , exposed , infectious , and removed individuals at time @xmath28 ( the dot denotes time derivatives ) .
@xmath29 is not an epidemiological state but serves to keep track of the cumulative number of ebola cases from the time of onset of symptoms .
+ the analogous stochastic model ( continuous time markov chain ) is constructed by considering three events : _ exposure _ , _ infection _ and _ removal_. the transition rates are defined as : + l c l event & effect & transition rate + exposure & ( s , e , i , r ) @xmath30 ( s-1 , e+1 , i , r ) & @xmath31 + infection & ( s , e , i , r ) @xmath30 ( s , e-1 , i+1 , r ) & @xmath32 + removal & ( s , e , i , r ) @xmath30 ( s , e , i-1 , r+1 ) & @xmath33 + + + the event times @xmath34 at which an individual moves from one state to another are modeled as a renewal process with increments distributed exponentially , + @xmath35 where @xmath36 .
+ the final epidemic size is @xmath37 where @xmath38 , and its empirical distribution can be computed via monte carlo simulations @xcite .
+ the intervention strategies to control the spread of ebola include surveillance , placement of suspected cases in quarantine for three weeks ( the maximum estimated length of the incubation period ) , education of hospital personnel and community members on the use of strict barrier nursing techniques ( i.e protective clothing and equipment , patient management ) , and the rapid burial or cremation of patients who die from the disease @xcite . their net effect , in our model ,
is to reduce the transmission rate @xmath16 from @xmath39 to @xmath40 . in practice , the impact of the intervention is not instantaneous . between the time of the onset of the intervention to the time of full compliance , the transmission rate is assumed to decrease gradually from @xmath39 to @xmath41 according to + @xmath42 where @xmath43 is the time at which interventions start and @xmath44 controls the rate of the transition from @xmath39 to @xmath41 .
another interpretation of the parameter @xmath44 can be given in terms of @xmath45 , the time to achieve @xmath46 .
the data for the congo ( 1995 ) and uganda ( 2000 ) ebola hemorrhagic fever outbreaks include the identification dates of the causative agent and data sources .
the reported data are ( @xmath47 , @xmath48 ) , @xmath49 where @xmath47 denotes the @xmath50 reporting time and @xmath48 the cumulative number of infectious cases from the beginning of the outbreak to time @xmath47 . + * congo 1995 . * this outbreak began in the bandundu region , primarily in kikwit , located on the banks of the kwilu river .
the first case ( january 6 ) involved a 42-year old male charcoal worker and farmer who died on january 13 . the ebola virus was not identified as the causative agent until may @xmath51 . at that time , an international team implemented a control plan that involved active surveillance ( identification of cases ) and education programs for infected people and their family members .
family members were visited for up to three weeks ( maximum incubation period ) after their last identified contact with a probable case .
nosocomial transmission occurred in kikwit general hospital but it was halted through the institution of strict barrier nursing techniques that included the use of protective equipment and special isolation wards .
a total of @xmath52 cases of ebola were identified ( @xmath53 case fatality ) .
daily ebola cases by date of symptom onset from march @xmath54 through july @xmath55 are available ( figure [ figdailycases ] ) @xcite . +
* uganda 2000 . * a total of @xmath56 cases ( @xmath57 case fatality ) of ebola were identified in three districts of uganda : gulu , masindi and mbara .
the onset of symptoms for the first reported case was on august @xmath58 , but the cause was not identified as ebola until october @xmath59 by the national institute of virology in johannesburg ( south africa ) .
active surveillance started during the third week of october .
a plan that included the voluntary hospitalization of probable cases was then put in place .
suspected cases were closely followed for up to three weeks .
other control measures included community education ( avoiding crowd gatherings during burials ) and the systematic implementation of protective measures by health care personnel and the use of special isolation wards in hospitals .
weekly ebola cases by date of symptom onset are available from the who ( world health organization ) @xcite ( from august @xmath60 , @xmath2 through january @xmath61 , @xmath62 ) ( figure [ figdailycases ] ) .
+ empirical studies in congo suggest that the incubation period is less than @xmath63 days with a mean of @xmath9 days @xcite and the infectious period is between @xmath64 and @xmath65 days .
the model parameters @xmath66 , @xmath41 , @xmath67 , @xmath44 , @xmath22 ) are fitted to the congo ( 1995 ) and uganda ( 2000 ) ebola outbreak data by _ least squares _
fit to the cumulative number of cases @xmath68 in eqn .
( [ eqn1 ] ) .
we used a computer program ( berkeley madonna , berkeley , ca ) and appropriate initial conditions for the parameters ( @xmath69 , @xmath70 , @xmath71 @xcite , @xmath72 @xcite ) .
the optimization process was repeated @xmath11 times ( each time the program is fed with two different initial conditions for each parameter ) before the `` best fit '' was chosen .
the asymptotic variance - covariance @xmath73 of the least - squares estimate is + @xmath74 + which we estimate by + @xmath75 + where @xmath76 is the total number of observations , @xmath77 and @xmath78 are numerical derivatives of @xmath79 . + for small samples , the confidence intervals based on these variance estimates may not have the nominal coverage probability .
for example , for the case of zaire @xmath1 , the @xmath80 confidence interval for @xmath44 based on asymptomatic normality is ( @xmath81 ) .
it should be obvious that this interval is not `` sharp '' as it covers negative values whereas we know @xmath82 .
the likelihood ratio provides an attractive alternative to build confidence sets ( figure [ figq ] ) .
formally , these sets are of the form + @xmath83 + where @xmath84 is the @xmath85 quantile of an @xmath86 distribution with appropriate degrees of freedom .
parameter estimates are given in table [ tableparameters ] .
the basic reproductive number @xmath3 measures the average number of secondary cases generated by a primary case in a pool of mostly susceptible individuals @xcite and is an estimate of the epidemic growth at the start of an outbreak if everyone is susceptible .
that is , a primary case generates @xmath87 new cases on the average where @xmath39 is the pre - interventions transmission rate and @xmath88 is the mean infectious period .
the effective reproductive number at time @xmath28 , @xmath89 , measures the average number of secondary cases per infectious case @xmath28 time units after the introduction of the initial infections and @xmath90 as the population size is much larger than the resulting size of the outbreak ( table [ tableoutbreaks ] ) .
hence , @xmath91 . in a closed population ,
the effective reproductive number @xmath92 is non - increasing as the size of the susceptible population decreases .
the case @xmath93 is of special interest as it highlights the crossing of the threshold to eventual control of the outbreak .
an intervention is judged successful if it reduces the effective reproductive number to a value less than one . in our model , the post - intevention reproductive number @xmath94 where @xmath41 denotes the post - intervention transmission rate . in general ,
the smaller @xmath41 , the faster an outbreak is extinguished . by the delta method @xcite , the variance of the estimated basic reproductive number
@xmath95 is approximately + @xmath96 a rough estimate of the population size in the bandundu region of congo ( where the epidemic developed ) in 1995 is computed from the population size of the bandundu region in @xmath97 @xcite and annual population growth rates @xcite ( table [ tableoutbreaks ] ) . for the case of uganda ( 2000 ) , we adjusted the population sizes of the districts of gulu , masindi and mbara in @xmath98 and annual population growth rates @xcite ( table [ tableoutbreaks ] ) .
these estimates are an upper bound of the effective population size ( those at risk of becoming infected ) for each region .
estimates of the effective population size are essential when the incidence is modeled with the pseudo mass - action assumption ( @xmath99 ) which implies that transmission grows linearly with the population size and hence the basic reproductive number @xmath100 . in our model
, we use the true mass - action assumption ( @xmath31 ) which makes the model parameters ( homogeneous system of order @xmath54 ) independent of @xmath17 and hence the basic reproductive number can be estimated by @xmath101 @xcite .
in fact , comparisons between the pseudo mass - action and the true mass - action assumptions with experimental data have concluded in favor of the later @xcite .
the model assumption that @xmath17 is constant is not critical as the outbreaks resulted in a small number of cases compared to the size of the population .
log - normal distributions seem to model well the incubation period distributions for a large number of diseases @xcite . here , a log - normal distribution is assumed for the incubation period of ebola in our uncertainty analysis .
log - normal distribution parameters are set from empirical observations ( mean incubation period is @xmath9 and the @xmath102 quantile is @xmath63 days @xcite ) .
the infectious period is assumed to be uniformly distributed in the range ( @xmath103 ) days @xcite . + a formula for the basic reproductive number @xmath3 that depends on the initial per - capita rate of growth @xmath104 in the number of cases ( figure [ figr0uncertainty ] ) , the incubation period ( @xmath19 ) and the infectious period ( @xmath88 ) can be obtained by linearizing equations @xmath105 and @xmath106 of system ( [ eqn1 ] ) around the disease - free equilibrium with @xmath107 .
the corresponding jacobian matrix is given by : + @xmath108 and the characteristic equation is given by : @xmath109 where the early - time and per - capita free growth @xmath104 is essentially the dominant eigenvalue . by solving for @xmath16 in terms of @xmath104 , @xmath67 and @xmath22 , one can obtain the following expression for @xmath3 using the fact that @xmath110 : + @xmath111 + our estimate of the initial rate of growth @xmath104 for the congo 1995 epidemic is @xmath112 day@xmath113 , obtained from the time series @xmath114 , @xmath115 of the cumulative number of cases and assuming exponential growth ( @xmath116 ) .
the distribution of @xmath3 ( figure [ figr0uncertainty ] ) lies in the interquartile range ( iqr ) ( @xmath117 ) with a median of @xmath118 , generated from monte carlo sampling of size @xmath119 from the distributed epidemic parameters ( @xmath19 and @xmath88 ) for fixed @xmath104 @xcite .
we give the median of @xmath3 ( not the mean ) as the resulting distribution of @xmath3 from our uncertainty analysis is skewed to the right .
using our parameter estimates ( table [ tableparameters ] ) , we estimate an @xmath3 of @xmath4 ( sd @xmath5 ) for congo ( 1995 ) and @xmath6 ( sd @xmath7 ) for uganda ( 2000 ) . the effectiveness of interventions is often quantified in terms of the reproductive number @xmath120 after interventions are put in place . for the case of congo @xmath121
( sd @xmath122 ) and @xmath123 ( sd @xmath124 ) for uganda allowing us to conclude that in both cases , the intervention was successful in controlling the epidemic .
furthermore , the time to achieve a transmission rate of @xmath125 ( @xmath126 ) is @xmath127 ( @xmath102 ci ( @xmath128 ) ) days and @xmath129 ( @xmath80 ci ( @xmath130 ) ) days for the cases of congo and uganda respectively after the time at which interventions begin .
+ we use the estimated parameters to simulate the ebola outbreaks in congo ( 1995 ) and uganda ( 2000 ) via monte carlo simulations of the stochastic model of section @xmath131 @xcite .
there is very good agreement between the mean of the stochastic simulations and the reported cases despite the `` wiggle '' captured in the residuals around the time @xmath43 of the start of interventions ( figure [ figmodel2 ] ) .
the empirical distribution of the final epidemic sizes for the cases of congo @xmath1 and uganda @xmath2 are given in figure [ figoutbreaksizedistr ] .
+ the final epidemic size is sensitive to the start time of interventions @xmath43 .
numerical solutions ( deterministic model ) show that the final epidemic size grows exponentially fast with the initial time of interventions ( not surprising as the intial epidemic growth is driven by exponential dynamics ) .
for instance , for the case of congo , our model predicts that there would have been @xmath60 more cases if interventions had started one day later ( figure [ figsensinterv ] ) .
using epidemic - curve data from two major ebola hemorrhagic fever outbreaks @xcite , we have estimated the basic reproductive number ( @xmath3 ) ( table [ tableoutbreaks ] ) .
our estimate of @xmath3 ( median is @xmath118 ) obtained from an uncertainty analysis @xcite by simple random sampling ( figure [ figr0uncertainty ] ) of the parameters @xmath67 and @xmath22 distributed according to empirical data from the zaire ( now the democratic republic of congo ) @xmath0 ebola outbreak @xcite is in agreement with our estimate of @xmath132 from the outbreak in congo @xmath1 ( obtained from least squares fitting of our model ( [ eqn1 ] ) to epidemic curve data ) . + the difference in the basic reproductive numbers @xmath3 between congo and uganda is due to our different estimates for the infectious period ( @xmath88 ) observed in these two places .
their transmission rates @xmath39 are quite similar ( table [ tableparameters ] ) .
our estimate for the infectious period for the case of congo ( @xmath133 days ) is slightly larger than that of uganda ( @xmath134 days ) .
clearly , a larger infectious period increases the likelihood of infecting a susceptible individual and hence increases the basic reproductive number .
the difference in the infectious periods might be due to differences in virus subtypes @xcite .
the congo outbreak was caused by the ebola - zaire virus subtype @xcite while the uganda outbreak was caused by the ebola - sudan virus subtype @xcite .
+ the significant reduction from the basic reproductive number ( @xmath3 ) to the post - intervention reproductive number ( @xmath120 ) in our estimates for congo and uganda shows that the implementation of control measures such as education , contact tracing and quarantine will have a significant effect on lowering the effective reproductive rate of ebola .
furthermore , estimates for the time to achieve @xmath125 have been provided ( table [ tableparameters ] ) .
+ we have explored the sensitivity of the final epidemic size to the starting time of interventions .
the exponential increase of the final epidemic size with the time of start of interventions ( figure [ figsensinterv ] ) supports the idea that the rapid implementation of control measures should be considered as a critical component in any contingency plan against disease outbreaks specially for those like ebola and sars for which no specific treatment or vaccine exists . a two - week delay in implementing public health measures results in an approximated doubling of the final outbreak size .
because the existing control measures cut the transmission rate to less than half , we should seek and support further improvement in the effectiveness of interventions for ebola . a mathematical model that considers basic public health interventions for sars control in toronto supports this conclusion @xcite .
moreover , computer simulations show that small perturbations to the rate @xmath44 at which interventions are put fully in place do not have a significant effect on the final epidemic size .
the rapid identification of an outbreak , of course , remains the strongest determinant of the final outbreak size .
+ field studies of ebola virus are difficult to conduct due to the high risk imposed on the scientific and medical personnel @xcite .
recently , a new vaccine that makes use of an _ adenovirus technology _ has been shown to give cynomolgus macaques protection within @xmath135 weeks of a single jab @xcite .
if the vaccine turns out to be effective in humans , then its value should be tested .
a key question would be `` what are the conditions for a successful target vaccination campaign during an ebola outbreak ? '' to address questions of this type elaborate models need to be developed .
99 centers for disease control ( cdc ) .
ebola hemorrhagic fever . `
( http://www.cdc.gov/ncidod/dvrd/spb/mnpages/dispages/ebola.htm ) ` , accessed on august 24 , 2003 .
centers for disease control ( cdc ) .
ebola hemorrhagic fever : table showing known cases and outbreaks , in chronological order . `
( http://www.cdc.gov/ncidod/dvrd/spb/mnpages/dispages/ebotabl.htm ) ` , accessed on august 24 , 2003 .
world health organization ( who ) .
ebola hemorrhagic fever : disease outbreaks . `
( http://www.who.int/disease-outbreak-news/disease/a98.4.htm ) ` , accessed on october 17 , 2003 .
breman , jg , johnson , km , van der groen , g , et al . a search for for ebola virus in animals in the democratic republic of the congo and cameroon : ecologic , virologic , and serologic surveys 1979 - 1980 .
j inf dis 1999;179:s139 - 47 .
leirs , h , mills , jn , krebs , jw , et al . search for the ebola virus reservoir in kikwit , democratic republic of the congo : reflections on a vertebrate collection .
j inf dis 1999;179:s155 - 63 .
world health organization ( who ) .
ebola hemorrhagic fever . `
( http://www.who.int/inf-fs/en/fact103.html ) ` , accessed on august 24 , 2003 .
breman , jg , piot , p , johnson , km , et al .
the epidemiology of ebola hemorrhagic fever in zaire , 1976 .
proc int colloquium on ebola virus inf held in antwerp , belgium , @xmath136 december 1977 .
birmingham , k and cooney , s. ebola : small , but real progress ( news feature ) .
nature med 2002;8:313 .
anderson , rm and may , rm . infectious diseases of humans .
oxford university press , oxford , 1991 .
brauer , f , castillo - chavez , c. mathematical models in population biology and epidemiology , springer - verlag , new york , 2000 .
renshaw , e. modelling biological populations in space and time .
cambridge university press , cambridge , 1991 .
khan , as , tshioko , fk , heymann , dl , et al .
the reemergence of ebola hemorrhagic fever , democratic republic of the congo , 1995 .
j inf dis 1999;179:s76 - 86 .
world health organization ( who ) .
outbreak of ebola hemorrhagic fever , uganda , august @xmath2 - january @xmath62 .
weekly epidemiological record 2001;76:41 - 48 .
piot , p , bureau , p , breman , g. d. heymann , et al .
clinical aspects of ebola virus infection in yambuku area , zaire , 1976 .
colloquium on ebola virus inf . held in antwerp , belgium , @xmath136 december 1977 .
bickel , p , doksum , ka . mathematical statistics .
holden - day , oakland , california , 1977 .
the world gazetteer .
democratic republic of the congo . `
( http://www.world-gazetteer.com/fr/fr_cd.htm ) ` , accessed on august @xmath137 , 2003 .
un - habitat ( united nations human settlelments programme ) .
republic of congo .
website : ` ( http://www.unhabitat.org/habrdd/conditions/midafrica/zaire.html ) ` , accessed on august 24 , 2003 .
uganda bureau of statistics ( ubos ) .
website : ` ( http://www.ubos.org/ ) ` , accessed on august 24 , 2003 .
castillo - chavez , c , velasco - hernandez , jx and fridman , s. modeling contact structures in biology . in : s.a .
levin ed . , frontiers of theoretical biology , lecture notes in biomathematics 100 , springer - veralg , berlin - heidelberg - new york 1994:454 - 91 .
de jong , mcm , diekmann , o and heesterbeek , h. how does transmission of infection depend on population size ? .
in : mollison , d , ed .
epidemic models : their structure and relation to data , cambridge university press , 1995:84 - 94 .
sartwell , pe .
the distribution of incubation periods and the dynamics of infectious disease .
am j epidem 1966;83:204 - 216 .
blower , sm , dowlatabadi , h. sensitivity and uncertainty analysis of complex models of disease transmission : an hiv model , as an example .
int stat rev 1994;2:229 - 243 .
niikura , m , ikegami , t , saijo , m , et al .
analysis of linear b - cell epitopes of the nucleoprotein of ebola virus that distinguish ebola virus subtypes , clin diagn lab immunol 2003;10:83 - 87 .
chowell , g , fenimore , pw , castillo - garsow , ma , et al . sars outbreaks in ontario , hong kong and singapore : the role of diagnosis and isolation as a control mechanism .
j theor biol 2003;24:1 - 8 .
chowell , g , castillo - chavez , c , fenimore , pw , et al . implications of an uncertainty and sensitivity analysis for sars s basic reproductive number for general public health measures ( submitted ) .
nabel , gj .
surviving ebola virus infection ( news feature ) .
nature 1999;4:373 - 374 .
sullivan , nj , geisbert , tw , geibsert , jb , et al . accelerated vaccination for ebola virus haemorrhagic fever in non - human primates .
nature 2003;424:681 - 684 .
clarke t and knight , j. fast vaccine offers hope in battle with ebola .
nature 2003;424:602 .
.parameter definitions and baseline estimates obtained from the best fit of the model equations ( [ eqn1 ] ) to the epidemic - curve data of the congo @xmath1 and uganda @xmath2 outbreaks ( figure [ figmodel2 ] ) .
the parameters were optimized by a computer program ( berkeley madonna , berkeley , ca ) using a _
least squares _ fitting technique and appropriate initial conditions for the parameters ( @xmath69 , @xmath70 , @xmath71 @xcite , @xmath72 @xcite ) .
the optimization process was repeated @xmath11 times ( each time the program is fed with two different initial conditions for each parameter ) before the `` best fit '' was chosen .
[ cols="<,<,^,^,^,^ " , ] | despite improved control measures , ebola remains a serious public health risk in african regions where recurrent outbreaks have been observed since the initial epidemic in @xmath0 . using epidemic modeling and data from two well - documented ebola outbreaks ( congo @xmath1 and uganda @xmath2 ) , we estimate the number of secondary cases generated by an index case in the absence of control interventions ( @xmath3 ) .
our estimate of @xmath3 is @xmath4 ( sd @xmath5 ) for congo ( 1995 ) and @xmath6 ( sd @xmath7 ) for uganda ( 2000 ) .
we model the course of the outbreaks via an seir ( susceptible - exposed - infectious - removed ) epidemic model that includes a smooth transition in the transmission rate after control interventions are put in place .
we perform an uncertainty analysis of the basic reproductive number @xmath3 to quantify its sensitivity to other disease - related parameters .
we also analyze the sensitivity of the final epidemic size to the time interventions begin and provide a distribution for the final epidemic size .
the control measures implemented during these two outbreaks ( including education and contact tracing followed by quarantine ) reduce the final epidemic size by a factor of @xmath8 relative the final size with a two - week delay in their implementation . |
among the heavy quarkona , the @xmath15 meson being the unique meson with two different heavy quarks in the standard model has aroused great interest since its discovery by the cdf collaboration @xcite . although the direct " hadronic production of the @xmath15 meson has been systematically studied in refs .
@xcite , as a compensation to understand the production mechanism and the @xmath15 meson properties , it would be helpful to study its production indirectly " through @xmath20(@xmath21)-quark , @xmath22-boson , and @xmath23-boson decays , as too many directly produced @xmath15 events shall be cut off by the trigging condition @xcite .
a systematical study on the indirect production of @xmath15 mesons through @xmath20(@xmath21)-quark , @xmath22-boson , and @xmath23-boson decay can be found in refs .
@xcite , refs .
@xcite , and refs .
@xcite , respectively .
meanwhile , it has been found that the higher excited states like @xmath24 and @xmath25 wave states can provide sizable contributions in the @xmath15 meson s indirect production through @xmath26-boson decay in ref .
@xcite ; one should take these higher fock states contributions into account so as to make a better estimation with other indirect production mechanisms . therefore , to present a systematic study on higher fock states indirect production of @xmath15 meson indirect production through top
quark decays under the nonrelativistic quantum chromodynamics framework ( nrqcd ) @xcite is one of the purposes of the present paper . with the lhc running at the center - of - mass energy @xmath13 tev and luminosity raising up to @xmath12 , about @xmath27 @xmath20-quark or @xmath21-quark events per year
will be produced @xcite .
this makes the lhc much better than tevatron , since more @xmath28-quark rare decays can be adopted for precise studies .
a systematic study on the production of the @xmath29 or @xmath15 meson and its excited states via @xmath20-quark or @xmath21-quark decays , e.g. @xmath30\rangle + cw^{+}$ ] or @xmath31\rangle + \bar{c}w^{-}$ ] , can be found in refs .
@xcite , where @xmath32 wave states .
their results show that large number of heavy quarkonium events through top quark decays can be found at lhc(slhc , dlhc , and tlhc , etc .
@xcite ) , so these channels shall be helpful for studying heavy quarkonium properties .
in addition to the production of the two color - singlet @xmath33 wave states @xmath34 and @xmath35 , naive nrqcd scaling rule shows that the production of the four color - singlet @xmath36-wave states @xmath37 and @xmath38 ( with @xmath39 ; @xmath40 ) shall also give sizable contributions in @xmath41\rangle+w^{+}q$ ] , where @xmath8 stands for @xmath9 or @xmath10 quark accordingly .
these higher excited @xmath42\rangle$ ] quarkonium states , may directly or indirectly ( in a cascade way ) decay to its ground state with almost 100% possibility via electromagnetic or hadronic interactions .
so , it would be interesting to study higher fock states contributions so as to make a more sound estimation of the production of the heavy quarkonium through top quark decays , and hence to be a useful reference for experimental studies . in the framework of an effective theory of the nrqcd ,
a doubly heavy meson system is considered as an expansion of various fock states . the relative importance among those infinite ingredients
is evaluated by the velocity scaling rule . in evaluating the production and decay amplitude of the heavy quarkonium
, each factor can be separated into a short - distance factor and a long - distance coefficient .
the short - distance factor can be computed using perturbative quantum chromodynamics ( pqcd ) , and the long - distance factor is associated with quarkonia structure , which is expressed in terms of nonperturbative matrix elements @xmath43 .
the matrix elements @xmath44 can be expressed in terms of the meson s nonrelativistic wave function , or its derivatives , evaluated at the origin under the color - singlet model @xcite , and the wave function is identified with the schr@xmath0dinger wave function calculated in potential models for heavy quarkonium . as a result ,
the rigorously calculated nonrelativistic wave function at the origin is very important in studying heavy quarkonium decay and production . because of the emergence of massive fermion lines in @xmath41\rangle+w^{+}q$ ] , the analytical expression for the squared amplitude becomes too complex and lengthy .
for such complicated processes , one important way out is to deal with it directly at the amplitude level . for this purpose ,
the improved trace technology " suggested and developed by refs .
@xcite shows that the hard scattering amplitude can also be expressed by the dot products of the concerned particle momenta like that of the squared amplitudes . in the present paper
, we shall adopt improved trace technology to derive the analytical expression for all the mentioned fock states , and to be a useful reference , we shall transform its form to be as compact as possible by fully applying symmetries and relations among them .
this paper is organized as follows : in sec .
ii , we show our calculation techniques for the mentioned top quark semi exclusive decays to heavy quarkonium . in order to calculate the production of the excited heavy quarkonium via top
quark decays , we present five qcd - motivated potential models for heavy @xmath45\rangle$ ] quarkonium in sec.iii . in sec.iv , we calculate and tabulate all the values of the schr@xmath0dinger radial wave functions , its first nonvanishing derivative , and its second nonvanishing derivative at zero quark - antiquark separation . furthermore , we also present numerical results and make some discussions on the properties of the heavy quarkonium production through top quark decays .
the final section is reserved for a summary .
we shall deal with some typical top quark semiexclusive processes for the heavy quarkonium production , i.e. , @xmath46\rangle(q_3 ) + w^{+}(q_2 ) + q(q_1)$ ] , where @xmath47 ( @xmath48 ) are the momenta of the corresponding particles .
according to the nrqcd factorization formula @xcite , its total decay widths @xmath49 can be factorized as @xmath50\rangle+ q+w^{+ } ) \langle{\cal o}^h(n ) \rangle,\ ] ] where the nonperturbative matrix element @xmath51 describes the hadronization of a @xmath52 pair into the observable quark state @xmath53 and is proportional to the transition probability of the perturbative state @xmath52 into the bound state @xmath42\rangle$ ] . as for the color - singlet components ,
the matrix elements can be directly related to the schr@xmath0dinger wave functions at the origin for the @xmath33 wave states , the first derivative of the wave functions at the origin for the @xmath36 wave states , or the second derivative of the wave functions at the origin for the @xmath54 wave states @xcite , which can be computed via potential nrqcd ( pnrqcd ) @xcite , lattice qcd @xcite and/or the potential models @xcite .
the short - distance decay widths are given by @xmath55\rangle+q+w^{+})= \frac{1}{2 q^0_{0 } } \overline{\sum } |m|^{2 } d\phi_3,\ ] ] where @xmath56 means that we need to average over the spin and color states of the initial particle and to sum over the color and spin of all the final particles .
in the top quark rest frame , the three - particle phase space can be written as @xmath57 we have done a calculation to simplify the @xmath58 phase space with massive quark / antiqark in the final state in refs . @xcite . to shorten the paper ,
we shall not present it here and the interested reader may turn to these three references for the detailed technology . with the help of the formulas listed in refs .
@xcite , one can not only derive the whole decay widths but also obtain the corresponding differential decay widths that are helpful for experimental studies , such as @xmath59 , @xmath60 , @xmath61 , and @xmath62 , where @xmath63 , @xmath64 , @xmath65 is the angle between @xmath66 and @xmath67 , and @xmath68 is the angle between @xmath66 and @xmath69 .
in particular , the partial decay widths over @xmath70 and @xmath71 can be expressed as @xmath72 where @xmath73 is the mass of the top quark .
the color - singlet nonperturbative matrix element @xmath44 can be related to the schr@xmath0dinger wave function @xmath74 at the origin or the first derivative of the wave function @xmath75 at the origin for @xmath33- and @xmath36-wave quarkonium states . for convenience
, we have adopted the convention of refs .
@xcite for the nonperturbative matrix element .
@xmath76\rangle}(0)|^2,\nonumber\\ \langle{\cal o}^h(np ) \rangle & \simeq & |\psi^\prime_{\mid(b\bar{q})[np]\rangle}(0)|^2.\end{aligned}\ ] ] since the spin - splitting effects are small , we will not distinguish the difference between the wave function parameters for the spin - singlet and spin - triplet states at the same @xmath77th level .
and then our task is to deal with the hard - scattering amplitude for specified processes @xmath78\rangle + cw^{+},~~t\rightarrow |(b\bar{b})[n]\rangle + bw^{+}.\end{aligned}\ ] ] for convenience , we shorten the two processes as @xmath79\rangle(q_3 ) + w^{+}(q_2)+q(q_1)$ ] , where @xmath8 stands for @xmath9 or @xmath10 quark accordingly .
the feynman diagrams of the process are presented in fig .
[ feyn1 ] , where the intermediate gluon should be hard enough to produce a @xmath80 pair or @xmath81 pair , so the amplitude is pqcd calculable .
\rangle(q_3 ) + w^{+}(q_2)+ q(q_1)$ ] , where @xmath8 stands for the @xmath9 and @xmath10 quark in the left and right panels , respectively . @xmath42\rangle$ ] quarkonium stands for a heavy quarkonium fock state.,scaledwidth=45.0% ] these amplitudes can be generally expressed as @xmath82 where @xmath83 stands for the number of feynman diagrams , @xmath84 and @xmath85 are spin states , and @xmath86 and @xmath87 are color indices for the outing @xmath8 quark and the initial top quark , respectively .
the overall factor @xmath88 , here @xmath89 is the cabibbo - kobayashi - maskawa ( ckm ) matrix element .
@xmath90s in the formulas are listed in ref .
@xcite .
as mentioned above , we adopt the improved trace technology to simplify the amplitudes @xmath91 at the amplitude level . in a difference from the helicity amplitude approach @xcite ,
only the coefficients of the basic lorentz structures are numerical at the amplitude level .
however , by using the improved trace technology in refs .
@xcite , one can sequentially obtain the squared amplitudes , and the numerical efficiency can also be greatly improved .
the standard procedures of the improved trace technology for @xmath79\rangle(q_3 ) + w^{+}(q_2)+q(q_1)$ ] have been presented in ref .
nonperturbative matrix elements @xmath44 can be related to the wave function at the origin @xcite . in the rest frame of the @xmath45\rangle$ ] quarkonium , it is convenient to separate the schr@xmath0dinger wave function into radial and angular pieces as @xmath92 where @xmath77 is the principal quantum number , and @xmath93 and @xmath83 are the orbital angular momentum quantum number and its projection . @xmath94 and @xmath95 are the radial wave function and the spherical harmonic function accordingly . further on , the value of the radial wave function , its first nonvanishing derivative or its second nonvanishing derivative at the origin can be obtained as in @xcite @xmath96 where @xmath97 , @xmath98 , and @xmath99 correspond to the radial wave functions @xmath100\rangle}(0)$ ] , @xmath101\rangle}(0)$ ] , and @xmath102\rangle}(0)$ ] at the origin .
the wave function @xmath103\rangle}(0)$ ] , the first derivative of the wave function @xmath104\rangle}(0)$ ] , and the second derivative of the wave function @xmath105\rangle}(0)$ ] at the origin are related to the radial wave function @xmath100\rangle}(0)$ ] , the first derivative of the radial wave function @xmath101\rangle}(0)$ ] , and the second derivative of the radial wave function @xmath102\rangle}(0)$ ] at the origin , accordingly .
@xmath106\rangle}(0)&=&\sqrt{{1}/{4\pi}}r_{|(q\bar{q}')[ns]\rangle}(0),\nonumber\\ \psi'_{|(q\bar{q'})[np]\rangle}(0)&=&\sqrt{{3}/{4\pi}}r'_{|(q\bar{q}')[np]\rangle}(0),\nonumber\\ \psi{''}_{|(q\bar{q'})[nd]\rangle}(0)&=&\sqrt{{5}/{16\pi}}r{''}_{|(q\bar{q}')[nd]\rangle}(0).\end{aligned}\ ] ] next , we will give a brief introduction to the five qcd - motivated potentials that give reasonable accounts of the @xmath107\rangle$ ] , @xmath108\rangle$ ] , and @xmath3\rangle$ ] quarkonium .
\(1 ) buchm@xmath109ller and tye have given the qcd - motivated potential ( b.t .
potential ) with two - loop correction @xcite as @xmath110 and in which @xmath112 , with @xmath113 , is the regge slope .
@xmath114 , @xmath115 , where here @xmath116 is the number of active flavor quarks .
@xmath117 stands for the scale parameters , and @xmath118 is the modified minimal subtraction scheme .
the parameter @xmath119 can be expressed in terms of the string constant @xmath120 . and @xmath121\sin(\frac{q}{\lambda}~r),\ ] ] with @xmath122},\ ] ]
where @xmath123 , and @xmath124 is the euler constant .
@xmath125 is a physical quantity and therefore independent of the choice of gauge and the subtraction scheme . for small values of @xmath126
, it has the form of @xmath127 for large @xmath126 , perturbative qcd implies @xmath128 here @xmath129 is the transfer momentum in the rest frame of the @xmath107\rangle$ ] , @xmath108\rangle$ ] , and @xmath3\rangle$ ] quarkonium .
\(2 ) the qcd - motivated potential with one - loop correction is given by john l. richardson ( j. potential ) @xcite as @xmath130,\end{aligned}\ ] ] where @xmath131.\ ] ] \(3 ) the qcd - motivated potential with two - loop correction is given by k. igi and s. ono ( i.o .
potential ) @xcite as @xmath132,\end{aligned}\ ] ] with @xmath133,\ ] ] in which @xmath134 , @xmath135 , where @xmath136 has the same meaning of the first potential model , and @xmath137 or @xmath138 is the mass of the heavy quark @xmath8 or @xmath139 , accordingly .
\(4 ) the qcd - motivated potential with two loop correction is given by yu - qi chen and yu - ping kuang ( c.k .
potential ) as @xcite @xmath140,\end{aligned}\ ] ] with @xmath141 and @xmath142}{\lambda_{\overline{ms}}~ r},\ ] ] where @xmath143 .
\(5 ) the qcd - motivated coulomb - plus - linear potential ( cor .
potential ) @xcite has the form of @xmath144
for calculating the wave function at the origin of the five potentials @xcite , we adopt the scale parameters @xmath117 as @xmath145=0.386 gev , @xmath146=0.332 gev , @xmath147=0.231 gev , @xmath148=0.0938 gev @xcite .
the quark mass is adopted as the values of the constituent quark mass of the @xmath149\rangle$ ] quarkonium derived in refs .
the quantities @xmath150\rangle}(0)|^2 $ ] , @xmath151\rangle}(0)|^2 $ ] , and @xmath152\rangle}(0)|^2 $ ] are presented in tables [ tabrpa ] , [ tabrpb ] , and [ tabrpc ] for the five potential models . during the following calculation , we adopt the values of wave functions at the origin under the b.t .
potential as the central values for calculations of the decay widths of @xmath153\rangle + qw^{+}$ ] , [ @xmath136=3 is for @xmath108\rangle$ ] quarkonium and @xmath136=4 for @xmath3\rangle$ ] quarkonium ] , since it is noted that the b.t .
model potential has the correction of two - loop short - distance behavior in pqcd @xcite .
the results for the other four potential models , i.e. , the j. model @xcite , the i.o .
model @xcite , the c.k .
model @xcite , and the cor .
model @xcite , will be adopted as an error analysis .
.radial wave functions at the origin and related quantities for @xmath107\rangle$ ] quarkonium . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] [ tabrpf ] in the present paper , we only calculate and discuss the decay widths of @xmath24 and @xmath25 wave of @xmath108\rangle$ ] and @xmath3\rangle$ ] quarkonium via the top - quark decays under the five potential models . yet we believe that the values of the wave functions at the origin of @xmath24 , @xmath25 , and @xmath154 wave of @xmath155 , @xmath156 , @xmath157 in tables [ tabrpa ] , [ tabrpb ] , and [ tabrpc ] under the five potential models are helpful for both theoretical and experimental study .
in the present paper , we have calculated the values of the schr@xmath0dinger radial wave function at the origin of @xmath107\rangle$ ] , @xmath108\rangle$ ] , and @xmath3\rangle$ ] quarkonium for the five potential models , and made a detailed study on the higher excited heavy quarkonium production through top quark semiexclusive decays , i.e. , @xmath158\rangle + cw^{+}$ ] and @xmath159\rangle + bw^{+}$ ] , within the nrqcd framework .
results for @xmath42\rangle$ ] quarkonium fock states , i.e. , @xmath160\rangle$ ] and @xmath161\rangle$ ] , and @xmath162\rangle$ ] and @xmath163\rangle$ ] ( @xmath164 ) have been presented . and to provide the analytical expressions as simply as possible , we have adopted the ` improved trace technology ' developed in refs .
@xcite to derive lorentz- invariant expressions for top quark decay processes at the amplitude level .
such a calculation technology shall be very helpful for dealing with processes with massive spinors .
numerical results show that higher @xmath24 and @xmath25 wave states in addition to the ground @xmath165 wave states can also provide sizable contributions to heavy quarkonium production through top quark decays , so one needs to take the higher @xmath24 and @xmath25 wave states into consideration for a sound estimation .
if all the excited states decay to the ground state @xmath166\rangle$ ] with @xmath167 efficiency , we can obtain the total decay width for @xmath168 quarkonium production through top quark decays as shown by eqs .
( [ bct ] ) and ( [ bbt1 ] ) . at the lhc , due to its high collision energy and high luminosity
, sizable heavy quarkonium events can be produced through top quark decays , i.e. , @xmath169 @xmath170 quarkonium events and @xmath171 @xmath172 bottomonium events per year can be obtained . therefore we need to take these higher excited states into consideration for a sound estimation .
( cdf collaboration ) , phys .
d * 58 * , 112004 ( 1998 ) ; a. abulencia _
( cdf collaboration ) , phys .
. lett . * 96 * , 082002 ( 2006 ) ; a. abulencia _
et al . _ , cdf collaboration , phys .
97 * , 012002 ( 2006 ) .
c.h . chang and y.q .
chen , phys .
d * 48 * , 4086 ( 1993 ) ; c.h .
chang , y.q .
chen , g.p .
han , and h.t .
jiang , phys .
lett.364 b , 78 ( 1995 ) ; c.h .
chang and x.g .
j. c * 38 * , 267 ( 2004 ) ; r.m .
thurman - keup , a.v .
kotwal , m. tecchio , and a.b .
wagner , rev .
phys * 73 * , 267 ( 2001 ) .
chang , c. driouich , p. eerola , and x.g .
wu , comput .
. commun . *
159 * , 192 ( 2004 ) ; c.h .
chang , j.x .
wang , and x.g .
wu , comput .
. commun . *
174 * , 241 ( 2006 ) ; * 175 * , 624 ( 2006 ) ; x.y .
wang and x.g .
wu , comput .
. commun . *
183 * , 442 ( 2006 ) .
chang , nucl .
b * 172 * , 425 ( 1980 ) ; r. baier and r. rueckl , phys .
* 102 * b , 364 ( 1981 ) ; e.l .
berger and d.jones , phys .
d * 23 * , 1521 ( 1981 ) ; h. krasemann , z. phys .
c * 1 * , 189 ( 1979 ) ; g. guberina , j. kuhn , r. peccei , and r. rueckl , nucl .
b * 174 * , 317 ( 1980 ) . | the value of the quarkonium wave function at the origin is an important quantity for studying many physical problems concerning a heavy quarkonium .
this is because it is widely used to evaluate the production and decay amplitudes of the heavy quarkonium within the effective field theory framework , e.g. , the nonrelativistic qcd ( nrqcd ) . in this paper , the values of the schr@xmath0dinger radial wave function or its first nonvanishing derivative at zero quark - antiquark separation , i.e. , @xmath1\rangle$ ] , @xmath2\rangle$ ] , and @xmath3\rangle$ ] quarkonium ,
have been tabulated under five potential models with new parameters for the heavy quarkonium .
moreover , the production of the lower - level fock states @xmath4\rangle$ ] and @xmath5\rangle$ ] , together with the higher excited fock states @xmath6\rangle$ ] and @xmath7\rangle$ ] ( @xmath8 stands for the @xmath9 or @xmath10 quark ; @xmath11 ) through top quark decays has been studied with the new values of heavy quarkonium wave functions at the origin under the framework of nrqcd . at the lhc with the luminosity @xmath12 and the center - of - mass energy @xmath13 tev
, sizable heavy quarkonium events can be produced through top quark decays , i.e. , @xmath14 @xmath15 and @xmath16 , and @xmath17 @xmath18 and @xmath19 events per year can be obtained according to our calculation . + * pacs numbers : * 12.38.bx , 14.65.ha , 14.40.nd , 14.40.pq |
the hadronic matrix element of weak @xmath0 transition ( @xmath2 : pseudoscalar meson ) is described by two form factors , whereas in general it requires four form factors to parametrize the weak matrix element for @xmath1 transition ( @xmath3 : vector meson ) .
heavy quark symmetry predicts that , all the mesonic form factors in the infinite quark mass limit @xmath4 are related to a single universal isgur - wise function @xcite .
the symmerty breaking @xmath5 corrections can be studied in a systematic framework , namely the heavy quark effective theory ( for a review , see @xcite ) .
the isgur - wise function is normalized to unity at zero recoil , but otherwise remains unknown .
phenomenologically , the hadronic form factors can be evaluated in various models among which the quark model is a popular one . however , since usual quark - model wave functions best resemble meson states in the rest frame or where the meson velocities are small , hence the form factors calculated in non - relativistic quark model or the mit bag model are trustworthy only when the recoil momentum of the daughter meson relative to the parent meson is small . as the recoil momentum increases ( corresponding to a decreasing @xmath6 )
, we have to start considering relativistic effects seriously .
in particular , at the maximum recoil point @xmath7 where the final meson could be highly relativistic , there is no reason to expect that the non - relativistic quark model is still applicable .
a consistent treatment of the relativistic effects of the quark motion and spin in a bound state is a main issue of the relativistic quark model . to our knowledge
, the light - front quark model @xcite is the only relativistic quark model in which a consistent and fully relativistic treatment of quark spins and the center - of - mass motion can be carried out .
this model has many advantages .
for example , the light - front wave function is manifestly lorentz invariant as it is expressed in terms of the momentum fraction variables ( in + " components ) in analog to the parton distributions in the infinite momentum frame .
moreover , hadron spin can also be correctly constructed using the so - called melosh rotation .
the kinematic subgroup of the light - front formalism has the maximum number of interaction - free generators including the boost operator which describes the center - of - mass motion of the bound state ( for a review of the light - front dynamics and light - front qcd , see @xcite ) .
the light - front quark model has been applied in the past to study the heavy - to - heavy and heavy - to - light weak decays form factors @xcite .
however , the weak form factors were calculated only for @xmath8 , whereas physical decays occur in the time - like region @xmath9 , with @xmath10 being the initial and final meson masses .
hence extra assumptions are needed to extrapolate the form factors to cover the entire range of momentum transfer . in @xcite
an ansatz for the @xmath6 dependence was made to extrapolate the form factors in the space - like region to the time - like region .
based on the dispersion formulation , form factors at @xmath11 were obtained in @xcite by performing an analytic continuation from the space - like @xmath6 region . finally , the weak form factors for @xmath0 transition were calculated in @xcite for the first time for the entire range of @xmath6 , so that additional extrapolation assumptions are no longer required .
this is based on the observation @xcite that in the frame where the momentum transfer is purely longitudinal , i.e. , @xmath12 , @xmath13 covers the entire range of momentum transfer .
the price one has to pay is that , besides the conventional valence - quark contribution , one must also consider the non - valence configuration ( or the so - called @xmath14-graph ) arising from quark - pair creation from the vacuum ( see fig . 1 ) .
the non - valence contribution vanishes if @xmath15 , but is supposed to be important for heavy - to - light transition near zero recoil @xcite .
unfortunately , a reliable way of estimating the @xmath14-graph contribution is still lacking . in the present paper
we calculate the @xmath1 form factors directly at time - like momentum transfers for the first time .
we then study the mesonic form factors in the infinite quark mass limit to check if the light - front model calculations respect heavy quark symmetry .
we are able to compute the isgur - wise function exactly since the non - valence contribution vanishes in the heavy - quark limit .
it turns out that not all light - front wave functions give a correct normalization for the isgur - wise function at zero recoil in @xmath1 decay . in other words , the requirement of heavy quark symmetry
can be utilized to rule out certain phenomenological wave functions .
another issue we would like to address in this work has to do with reference frame dependence of form factor . for a given @xmath6
, one can choose whether the recoiling daughter meson moves in the positive or negative @xmath16-direction relative to the parent meson , which we call the + " and @xmath17 " reference frame , respectively .
for some form factors in @xmath1 transition , namely @xmath18 , valence - quark and non - valence contributions are separately dependent on the choice of the + " or
@xmath17 " frame , but their sum should not .
this demonstrates the fact that it is mandatory to take into account the non - valence configuration in order to have physical predictions for form factors .
this issue will be discussed in more details in sections iid and ivc .
this paper is organized as follows . in section
ii , the basic theoretical formalism is given and form factors for @xmath0 and @xmath1 transitions are derived .
section iii is devoted to the discussion of the isgur - wise function .
numerical results are present and discussed in section iv , and finally a summary is given in section v.
we will describe in this section the light - front approach for the calculation of the weak mesonic form factors for pseudoscalar - to - pseudoscalar and pseudoscalar - to - vector transitions .
the hadronic matrix elements will be evaluated at time - like momentum transfers , namely the physically accessible kinematic region @xmath19 . a meson bound state consisting of a quark @xmath20 and an antiquark @xmath21 with total momentum @xmath2 and spin @xmath22
can be written as @xmath23 where @xmath24 and @xmath25 are the on - mass - shell light - front momenta , @xmath26 and @xmath27 in terms of the light - front relative momentum variables @xmath28 defined by @xmath29 the momentum - space wave - function @xmath30 can be expressed as @xmath31 where @xmath32 describes the momentum distribution of the constituents in the bound state , and @xmath33 constructs a state of definite spin ( @xmath34 ) out of light - front helicity ( @xmath35 ) eigenstates .
explicitly , @xmath36 where @xmath37 are the usual pauli spinor , and @xmath38 is the melosh transformation operator : @xmath39 with @xmath40 , a unit vector in the @xmath16-direction , and @xmath41 in practice it is more convenient to use the covariant form for @xmath33 @xcite : @xmath42 where @xmath43 with @xmath44 , \quad \vec \varepsilon_\bot ( \pm 1)=\mp(1,\pm i)/\sqrt{2 } , \nonumber\\ & & \hep^\mu(0)={1\over m_0}\left({-m_0 ^ 2+p_\bot^2\over p^+},p^+,p_\bot\right ) . \label{polcom}\end{aligned}\ ] ] note that the longitudinal polarization 4-vector @xmath45 given above is not exactly the same as that of the vector meson [ cf . eq.([polvec ] ) ] .
we normalize the meson state as @xmath46 so that @xmath47 0.3 cm in principle , the momentum distribution amplitude @xmath32 can be obtained by solving the light - front qcd bound state equation@xcite .
however , before such first - principles solutions are available , we would have to be contented with phenomenological amplitudes .
one example that has been often used in the literature for heavy mesons is the so - called bauer - stech - wirbel ( bsw ) amplitude@xcite , which for a meson of mass @xmath48 is given by @xmath49 , \label{bswamp}\ ] ] where @xmath50 is a normalization constant , @xmath51 is the longitudinal momentum fraction carried by the light antiquark , @xmath52 , and @xmath53 is a parameter of order @xmath54 .
an other example is the gaussian - type wave function , @xmath55 where @xmath56 , and @xmath57 of the internal momentum @xmath58 is defined through @xmath59 with @xmath60 .
we then have m_0=e_1 + e_2 , k_z = xm_02-m_2 ^ 2+k_^2 2 xm_0 , [ kz ] and @xmath61 is the jacobian of transformation from @xmath28 to @xmath62 .
this wave function has been also used in many other studies of hadronic transitions .
in particular , with appropriate parameters , it describes satisfactorily the pion elastic form factor up to @xmath63 @xcite .
a variant of the gaussian - type wave function is ( x , k_)=n(-m_0 ^ 22 ^ 2 ) , [ variant ] with @xmath64 being given by ( [ m0 ] ) .
this amplitude is equivalent to @xmath65 when the constituent quark masses are equal but becomes different otherwise .
nevertheless , we will not pursue this wave function further because it does not have an appropriate heavy - quark - limit behavior ( see sec .
iii ) . obviously , the isgur - wise function for heavy meson transitions depends on the heavy meson wave function @xmath66 chosen .
it turns out that in contrast to the gaussian - type wave function , the bsw wave function fails to give a correct normalization for the isgur - wise function at zero recoil in @xmath1 transition .
the decay constant of a pseudoscalar meson @xmath67 defined by @xmath68 can be evaluated using the light - front wave function given by ( 2.1 ) and ( 2.5 ) 0||q_2^+_5q_1|p&= & \{d^3p_1}\{d^3p_2 } 2(2)^3(p - p_1-p_2)_p(x , k_)r^00__1_2 ( x , k _ ) + & & 0||q_2^+_5q_1|q_1|q_2 . since @xmath69 , it is straightforward to show that f_p=4_p(x , k_)a , [ fp ] where = m_1x+m_2(1-x ) .
note that the factor @xmath70 in ( [ fp ] ) arises from the color factor implicit in the meson wave function .
likewise , for the vector - meson decay constant defined by 0|v^|v = f_vm_v^ , is found to be f_v & = & 4_v(x , k_)1m_v\ { x(1-x ) m_v^2+m_1m_2+k^2 _ + & & + b2m_v } , [ fv ] where = xm_1-(1-x)m_2 , w_v = m_0+m_1+m_2 .
when the decay constant is known , it can be used to constrain the parameters of the light - front wave function . with the light - front wave functions given above
, we will first calculate the form factors for @xmath0 transitions given by p_2|v^|p_1=f_+(q^2)(p_1+p_2)^+f_-(q^2)(p_1-p_2)^ , where @xmath71 . for later purposes
, it is also convenient to parametrize this matrix element in different forms : p_2|v^|p_1&= & , + & = & f_1(q^2)(p_1^+p_2 ^ -m_1 ^ 2-m_2 ^ 2q^2q^)+f_0 ( q^2)m_1 ^
2-m_2 ^ 2q^2q^ , [ ppform ] with @xmath72 , and f_1(q^2)=f_+(q^2 ) , f_0(q^2)=f_+(q^2)+q^2m_1 ^ 2-m_2 ^ 2f_-(q^2 ) .
[ ff ] in the heavy - quark limit @xmath73 , heavy - quark symmetry requires that @xcite h_+(q^2)=(v_1v_2 ) , h_-(q^2)=0 , [ pphqs ] where @xmath74 is the universal isgur - wise function normalized to unity at the point of equal velocities : @xmath75 .
the form factors @xmath76 and @xmath77 are related to the transition amplitude with the exchange of a vector ( @xmath78 ) and a scalar @xmath79 boson in the @xmath80-channel , respectively .
as explained in the introduction , we shall work in the frame where @xmath12 so that @xmath13 will cover the whole time - like region @xmath81 .
define @xmath82 ( it is denoted by @xmath83 in @xcite ) , then q^2=(1-r)(m_1 ^ 2-m_2 ^ 2r ) .
consequently , for a given @xmath6 , there are two solutions for @xmath84 : r_=m_2m_1(v_1v_2 ) , [ y12 ] where @xmath85 is related to @xmath6 by v_1v_2=m_1 ^ 2+m_2 ^ 2-q^22m_1m_2 .
the @xmath86 signs in ( [ y12 ] ) correspond to the daughter meson recoiling in the positive ( negative ) @xmath16-direction relative to the parent meson ( call them the + " and @xmath17 " reference frame , respectively ) . at zero recoil ( @xmath87 ) and maximum recoil ( @xmath7 ) , @xmath88
are given by & & r_+(q_max^2)=r_-(q_max^2)=m_2m_1 , + & & r_+(0)=1 , r_-(0)=(m_2m_1)^2 .
[ y12val ] the form factors @xmath89 of course should be independent of the reference frame chosen for the moving direction of the daughter meson . for a given @xmath6 ,
suppose we obtain p_2|v^+|p_1|_r = r_+=2p_1^+h(r_+ ) , p_2|v^+|p_1|_r = r_-=2p_1^+h(r_- ) . [ mel1 ]
it follows from ( [ ppform ] ) that f_+(q^2 ) & = & ( 1-r_-)h(r_+)-(1-r_+)h(r_-)r_+-r_- , + f_-(q^2 ) & = & -(1+r_-)h(r_+)-(1+r_+)h(r_-)r_+-r_- .
[ fpm ] it is easily seen that @xmath89 are independent of the choice of + " or
@xmath17 " frame , as it should be .
as noted earlier , in a frame with @xmath90 , there are actually two distinct contributions to the hadronic matrix element @xcite : valence ( partonic ) contribution calculated with relativistic light - front bound - state wave functions , and non - valence ( non - partonic ) contribution ( or the so - called @xmath14-graph ) arising from quark - antiquark pair creation from the vacuum . in the following
, we shall first provide some details for calculating the valence contribution , and then come back to the non - valence subprocess in sec .
ii.d . for @xmath91 and @xmath92 ,
the relevant quark momentum variables are & & p_1^+=(1-x)p_1^+ , p^+_=xp_1^+ , _ _ 1=(1-x )
_ 1 + _ , _ _
= x_1- _ , + & & p_2^+=(1-x)p_2^+ , p^+_=xp_2^+ , _ _ 2=(1-x ) _ 2+ _ , _ _ = x_2- _ , [ transmom ] where @xmath93 is the momentum fraction carried by the spectator antiquark @xmath94 in the initial ( final ) state .
the spectator model requires that p^+_=p_^+ , _ _ = _ _ . [ momeq ] taking a lorentz frame where @xmath95 amounts to having @xmath96 and @xmath97
. then we readily obtain p_2|v^+|p_1=__1,_2,| & & \{d^3p _ _ } ^*_2(x,k _ ) _ 1(x , k _ ) + & & r^00__2|(x,k _ ) |u(p_2,_2)^+ u(p_1,_1 ) r^00__1|(x , k _ ) , [ mel2 ] substituting the covariant form given in eq .
( [ covariant ] ) into eq .
( [ mel2 ] ) yields p_2|v^+|p_1= & & dx d^2k_2(2)^3 ^*_2(x,k _ ) _ 1(x , k _ ) -1 2 m_02 m_01 + & & .
[ mel3 ] after some manipulation , the trace term in the above expression is reduced to = -4x(a_1a_2+k^2_)p^+_1 , [ trace ] where _
1=m_1x+m_(1-x ) , a_2=m_2x+m_(1-x ) , [ a12 ] and use of ( [ transmom ] ) has been made . since _ 01_02= , [ den ] we find from ( [ mel1 ] ) , ( [ mel2 ] ) , ( [ trace ] ) and ( [ den ] ) that h(r)=^r_0dx^*_2(x,k _ ) _ 1(x , k_)a_1a_2+k^2 _ , [ hi ] with @xmath98 . the form factors @xmath89 can then be obtained from ( [ fpm ] ) .
as stated before , in the literature these form factors are customarily evaluated in the frame where @xmath99 .
this leads to @xmath100 , implying a space - like momentum transfer .
the advantage of the condition @xmath15 is that form factors only receive valence contributions ( see sec .
however , there are two drawbacks in this approach : first , form factors in the physical time - like region can not be obtained without making additional @xmath6 extrapolation assumptions .
second , no information can be obtained for the form factor @xmath101 since @xmath102 [ see ( 2.26 ) ] . at the maximum recoil @xmath7 , the form factor @xmath103
is evaluated to be @xcite f_+(0)=^1_0dx^*_2(x , k _ ) _ 1(x , k_)a_1a_2+k^2 _ , [ fp0 ] with @xmath104 and @xmath105 given by ( [ a12 ] ) except for that @xmath106 here .
therefore , the results of ( [ fpm ] ) and ( [ hi ] ) at @xmath7 are in agreement with ( [ fp0 ] ) .
form factors for @xmath1 transition are defined as & & v(p_v,)|j_|p(p_1)=2m_p+m_vi_^p_v^p_1^v(q^2)- , [ vpform ] where @xmath107 , @xmath108 , a_3(q^2)=m_p+m_v2m_va_1(q^2)-m_p - m_v2m_va_2(q^2 ) , [ a312 ] and ^(1)= ( 2___vp_v^+ , 0 , _ ) , ^(0 ) & = & 1m_v ( -m_v^2+p__v^2p_v^+ , p^+_v,__v ) [ polvec ] are , respectively , the transverse and longitudinal polarization vectors of the vector meson .
the form factors @xmath109 and @xmath110 are related to @xmath111 intermediate states , @xmath112 to @xmath113 states , and @xmath3 to @xmath78 states .
the @xmath1 matrix element can also be parametrized in different ways : v|j_| p&= & ig(q^2)_^p_v^p_1^+f(q^2)_+(p_1)[a_+(q^2)(p_1+p_v)_+a_-(q^2)(p_1- p_v ) _ ] + & = & \{i(q^2)_^v^v^+(q^2 ) _ + & & + ( v)[_+(q^2)(v+v)_+_-(q^2)(v - v ) _ ] } , [ vpform1 ] where @xmath114 and @xmath115 .
they are useful for later discussions .
the form factors @xmath116 and @xmath117 are related to @xmath118 via g(q^2 ) & = & 2m_p+m_vv(q^2 ) , f(q^2)=-(m_p+m_v)a_1(q^2 ) , + a_+(q^2 ) & = & 1m_p+m_va_2(q^2 ) , [ gv ] + a_-(q^2 ) & = & 2m_vq^2[a_3(q^2)-a_0(q^2 ) ] + & = & 2m_vq^2 . also , ( q^2 ) & = & g(q^2 ) , ( q^2)=f(q^2 ) , + _ + ( q^2)+_-(q^2)&=&m_p^2(a_++a_- ) , _ + ( q^2 ) -_-(q^2)=(a_+-a_- ) . in the heavy - quark limit @xmath119
, heavy - quark symmetry demands that @xcite _ + + _ -=0 , _
+ -_-==(vv ) , = -(1+vv)(vv ) .
[ vphqs ] the calculation of the @xmath1 form factors is more subtle than the @xmath0 case . if we choose a frame where @xmath120 as before , we will have @xmath121 for transverse polarization . as a result , form factors @xmath122 in ( [ vpform1 ] )
can not be separately determined .
therefore , we will let @xmath123 at the outset , and set them to zero only after the form factors are extracted . with the transverse polarization @xmath124 , form factors @xmath125 and @xmath126 can be individually determined . then using the longitudinal polarization @xmath127 , we are able to fix the remaining form factor @xmath128 .
we begin with @xmath125 .
since @xmath129 [ cf .
( [ polvec ] ) ] , it follows from ( [ vpform1 ] ) that -v(p_v)|a^+|p(p_1)&= & ( p_1)[a_+(p_1^++p_v^+)+a_-(p_1^+ -p_v^+ ) ] + & = & ( 1r-1)(__)[a_+(1+r)+a_- ( 1-r)]p_1^+ , [ vpme1 ] with @xmath130 and @xmath131 .
as will be shown below , the above matrix element at the quark level has the form v|a^+|p=2__(1-r)i(r)p^+_1 . substituting this into ( [ vpme1 ] ) and solving the equations for @xmath132 and @xmath133 yields a_+(q^2 ) & = & -r_+(1-r_-)i(r_+)-r_-(1-r_+)i(r_-)r_+-r_- , + a_-(q^2 ) & = & r_+(1+r_-)i(r_+)-r_-(1+r_+)i(r_-)r_+-r_- , [ apm ] in analog to eq.([fpm ] ) for @xmath89 . in order to illustrate several subtle points in the derivation of @xmath134
, we will go through the calculation in a bit more details .
first of all , it is straightforward to show that for @xmath135 and @xmath136 v|a^+|p=-2x^*_v(x,k _ ) _ p(x , k_)(a+b ) , where @xmath137 [ see eq.([a12 ] ) ] , and & & a = m_1(p_2 p^+_+p_p^+_2)+m_2(p_1 p^+ _
-p_p^+_1)+m_(p_2 p^+_1+p_1 p^+_2 ) , [ ab1 ] + & & b=(p_2-p_)w_v(m_1m_p^+_2-m_2m_p^+_1-m_1m_2 p^+_+p_1p_p^+_2-p_1p_2p^+_+p_2p_p^+_1 ) , with @xmath138 given by ( [ polvec ] ) and @xmath139 . by virtue of ( [ transmom ] )
we find that a & = & ( 1-r)(1 - 2x)a_p(__)p_1^++ , +
b & = & -2(1-r)a_pb_v+k^2_w_v(__)p_1^++ , [ ab2 ] with _ v =- m_2x+(1-x)m_. the ellipses in ( [ ab2 ] ) denote contributions from terms proportional to @xmath140 in ( [ ab1 ] ) . naively , these terms linear in @xmath141 are not expected to make contributions after integrating over @xmath141 . but
this is not the case .
consider the term _ v=_v(x,k _ ) and note that @xmath142 is different from @xmath143 due to a non - vanishing @xmath144 : k_=k_+(x-x)p _ , [ kp ] where we have used ( [ transmom ] ) and ( [ momeq ] ) .
consequently , _ v(k^2 _ ) & = & _ v(k^2_)+(d_v / dk^2_)(k^2_-k^2_)+ + & = & _ v(k^2 _ ) , [ thetav ] with _
v1_v(d_v dk^2 _ ) .
since d^2k_()()=12d^2k_k^2 _
( ) , it is evident that the linear term @xmath145 in ( [ ab2 ] ) will combine with the linear term @xmath146 in ( [ thetav ] ) to make a contribution to @xmath147 .
we wish to stress that this additional contribution from @xmath148 was first noticed and obtained by odonnell and xu @xcite and was neglected in the work of jaus @xcite . by the same token , in the expression of @xmath149 , the @xmath145 term in @xmath150
will also combine with the @xmath146 term in @xmath151 to yield a contribution proportional to @xmath152 after integration over @xmath143 .
the final result is i(r ) & = & -^r_0 dxx_v^ * ( x,k_)_p(x , k_)\{(1 - 2x)a_p + & + & [ ( 1 - 2x)a_p - a_v]_v(x,k _ ) k^2_-2 ( a_pb_v+k^2_)(1+_v(x,k_)k^2_)+12k^2_w_v } , [ ii1 ] with @xmath98 . in deriving ( [ ii1 ] )
we have first integrated out @xmath153 and @xmath142 .
we can of course alter the order of integration by first integrating over @xmath51 and @xmath143 and obtain i(r ) & = & -^1_0 dxx_v^ * ( x,k_)_p(x , k _ ) + & & \ { a_v -[(1 - 2x)a_p - a_v]_p(x , k _ ) k^2_+2(a_p b_v+k^2_)_p(x , k_)k^2_+12k^2 _ w_v } , [ ii2 ] with @xmath154 , where we have used the notation @xmath143 instead of @xmath142 as it is a dummy variable .
the result ( [ ii2 ] ) will be utilized in sec . 3 to show that light - front model calculations fulfill the heavy - quark - symmetry requirement ( [ vphqs ] ) . at @xmath7 , we have @xmath155 and @xmath156 .
it follows from ( [ apm ] ) and ( [ ii1 ] ) that a_2(0 ) & = & ( m_p+m_v)a_+(0)=-(m_p+m_v)i(r=1 ) + & = & ^1_0 dxx_v^ * ( x , k_)_p(x , k_)(m_p+m_v)\{(1 - 2x)a_p + & & + [ ( 1 - 2x)a_p - a_v]_v(x , k _ ) k^2_-2(a_p b_v+k^2_)(1+_v(x , k_)k^2_)+12k^2 _
w_v}. this is in agreement with eq.(30 ) of @xcite , but disagrees with the result obtained by jaus @xcite .
having fixed @xmath125 , we are ready to calculate @xmath128 in ( [ vpform1 ] ) . from ( [ gv ] ) it
is clear that once @xmath128 is determined , so are the form factors @xmath157 and @xmath158 . since the + " component of @xmath159 is needed to extract @xmath128 , we consider the longitudinal polarization @xmath127 of the vector meson @xmath3 and take a frame where @xmath160 .
hence , -v|a^+|p = f(q^2)rm_vp_1^++r2m_v(m_p^2-m_v^2 r^2)[a_+(1+r)+a_-(1-r)]p^+_1 .
let @xmath161 , then f(q^2)=-12(m_p^2-m_v^2r^2)[a_+(1+r)+a_-(1-r)]- m_vrj(q^2 ) .
[ fq2 ] after a straightforward manipulation , we obtain j(q^2)=-r_0^r dxx_v^ * ( x,k_)_p(x , k_)(c+d ) , [ jq2 ] with c & = & -2m_0v , + d & = & 1m_0v 1xw_v(a_pb_v+k^2 _ ) , [ cd ] and [ cf .
eq.([m0 ] ) ] m^2_0v = m_2 ^ 2+k^2_1-x+m_^2+k^2_x. [ m0pv ] to check the above results , we note that for @xmath162 [ see ( [ y12val ] ) ] v|a^+|p|_r(0)=r_+(0 ) & = & -[f(0)+(m^2_p - m^2_v)a_+(0)]p^+_1/m_v= 2a_0(0)p^+_1 , so that a_0(0 ) & = & 12j(0)|_r=1 where use has been made of ( [ gv ] ) . then it is not difficult to show from ( [ jq2]-[cd ] ) that a_0(0 ) & = & a_pa_v+(1 - 2x ) k^2_+2(m_1+m_2)xk^2_w_v which agrees with eq.(27 ) of @xcite obtained in the @xmath15 frame ( implying @xmath163 ) .
it will be shown in sec .
iii.b that our result for @xmath128 does respect the heavy - quark - symmetry requirement .
thus far we have imposed the condition @xmath12 to extract the form factors @xmath125 and @xmath128 . for the vector form factor @xmath126 or @xmath164
, it proves more convenient to first let @xmath165 and then set it to zero after the vector form factor is obtained .
the + " component of the vector matrix element for transverse polarization reads v|v^+|p&= & ig_+^p_v^p_1^ + & = & ig_+-xy[-^-q^x p^y+(p_1-p_v)^-^x p^y - p_1 ^ -^x q^y-(xy ) ] , [ vme1 ] where @xmath166 . at the quark level
, we have v|v^+|p&= & ^r_0dx2x_v^ * ( x,k_)_p(x , k _ ) . [ vme2 ]
the transverse momentum variables are p_1=(1-x)p_+k _ , p_2=(1-x)p_-q_+k _ , p_=xp_-k_. it suffices to set @xmath167 in ( [ vme2 ] ) to get contributions proportional to @xmath168 , which is related to the first term of ( [ vme1 ] ) at the hadron level .
it is easy to check that the transverse components @xmath169 and @xmath170 will not generate the same structure . repeating the similar derivation as before , we obtain g(q^2 ) & = & 2v(q^2)m_p+m_v = ^r_0 dx 2x_v^*(x,k_)_p(x , k _ ) + & & \ { a_p+(a_p - a_v ) _ v(x,k _ ) k^2 _ + 1w_v } , with @xmath57 being defined in ( [ kz ] ) . for @xmath171 , ( [ gq2 ] )
leads to v(0 ) & = & ^1_0 dx ( m_p+m_v)x_v^ * ( x,k_)_p(x , k _ ) + & & ( a_p+ k_^2w_v+x(m_1-m_2)_v(x , k_)k^2 _ ) , which agrees with @xcite .
therefore , we have calculated the form factors @xmath172 and @xmath125 in the time - like @xmath6 region within the light - front framework .
form factors @xmath164 and @xmath173 can then be determined via eq.([gv ] ) .
, only three of the @xmath1 form factors , namely @xmath174 and @xmath175 or @xmath176 and @xmath110 are determined .
however , @xmath112 can be fixed at @xmath7 using the relation @xmath177 and ( [ a312 ] ) . ]
thus far we have concentrated on the valence - quark contribution to the form factors . as stated in the introduction
, there also exist contributions which are generated from the quark - antiquark excitation or higher fock - states in the hadronic bound states .
this additional @xmath14-graph contribution vanishes in the frame where the momentum transfer is purely transverse i.e. , @xmath15 , but survives otherwise .
the general feature of the non - valence configuration can be recognized by considering the quark triangle diagram ( see fig . 1 ) . in terms of the + " component of momenta ,
the feynman triangle diagram in the light - front framework consists of two subprocesses : one corresponds to the valence - quark approximation for the meson wave functions , and the other to the contribution of quark - pair creation from the vacuum .
that is , through the mechanism of quark - antiquark pair creation , the spectator " quark in the second subprocess is fragmented into a meson plus an outgoing quark .
a detailed study of the quark triangle diagram for @xmath0 transition gives ( generalization to @xmath1 transition is straightforward ) @xcite p_2||q_2^+q_1|p_1=m_a^++m_b^+ , [ melmab ] with _
= & g_1g_2^r_0dxx(1-x)(1-x)n_a^+(m_1 ^ 2-m_01 ^ 2)(m_2 ^ 2-m_02 ^ 2 ) , + m_b^+ & = & -g_1g_2 ^ 1_rdxx(1-x)(1-x)n_b^+(m_1 ^ 2-m_01 ^ 2)(q^2-m_12 ^ 2)r 1-r , [ mab ] where @xmath98 , @xmath178 and @xmath179 are the quark - meson coupling constants at different vertices , @xmath180 and @xmath181 are the masses of the initial and final meson respectively , @xmath182 is the same as @xmath183 defined in ( [ m0pv ] ) , and m_12 ^ 2 & = & ( m_1 ^ 2+k^2_1-x - m^2_2+k^2_r - x ) ( 1-r ) , + n_a^+ & = & 4[p_1^+(m_2m_+p_2p_)+p_2^+(m_1m_+p_1p _ ) + p^+_(m_1m_2-p_1p_2 ) ] , + n_b^+ & = & 4[p_1^+(-m_1m_2+p_1p_2)+p_2^+(-m_1m_1 + p_1p_1)+p_1^+(m_1m_2-p_1p_2 ) ] , with @xmath184 . since @xmath185 receives contributions from the kinematic region @xmath186 or @xmath187 ( see fig . 1 )
, it corresponds to the valence - quark configuration . as for @xmath188 ,
only the region @xmath189 or @xmath190 is relevant , and it corresponds to the non - valence contribution .
it is straightforward to check that , apart from a sign difference , @xmath191 is precisely the trace term given in ( [ trace ] ) .
this implies that the previous calculation for @xmath0 form factors in the hamiltonian light - front approach is identical to the feynman triangle graph under the valence - quark approximation . obviously , making the following substitutions 1m_1 ^ 2-m_01 ^ 2 & & _ 1(x , k _ ) , + g_2x(1-x)1m_2 ^ 2-m_02 ^ 2 & & _
2(x,k _ ) in @xmath192 will reproduce the result @xmath193 of the initial meson enters into the expression of @xmath188 ; @xmath194 is not applicable for the non - valence graph because the light - front momentum @xmath195 of the spectator quark is larger than the momentum @xmath196 of the daughter meson ( see fig . 1 ) .
this makes the task of calculating the effect of the @xmath14-graph considerably more difficult .
nevertheless , some qualitative features of @xmath188 can still be comprehended .
first of all , as noted earlier , the contribution from non - valence configurations vanishes in a frame where @xmath15 or @xmath163 .
however , this frame is suitable only for space - like @xmath6 .
second , it is easy to show that @xmath197 in the limit of heavy quark symmetry @xmath4 , because it takes an infinite amount of energy to create a heavy quark - antiquark pair .
this has the important implication that we do not have to worry about the pair - creation subprocess when calculating the isgur - wise function . beyond the heavy - quark limit
, it is commonly argued that the non - valence contribution leads to a small correction in heavy - to - heavy transition but becomes more important for heavy - to - light decays @xcite .
for example , a @xmath198-pole contribution is usually believed to be the dominant non - valence effect in @xmath199 transition , especially when @xmath6 is near the zero - recoil point @xcite .
some estimates based on the @xmath198-pole contribution with the help of chiral perturbation theory indicate that for large values of @xmath6 , the @xmath14-graph provides the dominant contribution to @xmath199 form factors ( for a recent estimate , see @xcite ) . in this paper we will demonstrate that even for heavy - to - heavy transition , the importance of the non - valence contribution depends on the recoiling direction of the daughter meson . as shown in ( [ y12 ] ) , for a given @xmath6 there are two possible reference frames characterized by @xmath200 and @xmath201 , corresponding to whether one chooses the velocity of the final meson to be in the positive or negative @xmath16-direction relative to the initial meson
of course , the form factors are independent of the choice of the + " or @xmath17 " frame .
this means that the combination @xmath202 in ( [ melmab ] ) should be independent of the choice of @xmath203 or @xmath204 . from ( [ mab ] )
we see that the @xmath84 dependence of @xmath192 or @xmath188 appears both in the integrand and in the integration limit . as a consequence , @xmath192 and @xmath188
separately are in general
@xmath205"-frame dependent .
in other words , _ the valence - quark and non - valence contributions to mesonic form factors are in general dependent on the recoiling direction of the final meson , but their sum is not_. for the form factors @xmath89 in @xmath0 decay and @xmath125 in @xmath1 transition , we have demanded " that the valence contribution itself be frame independent [ see ( [ mel1 ] ) , ( [ fpm ] ) and ( [ apm ] ) ] . for form factors @xmath206 and @xmath3 in @xmath1 decay , explicit calculations in sec .
iv.b show that the valence contributions for @xmath132 and @xmath133 are indeed different ( see fig
thus in principle we can not make firm predictions for these form factors even for @xmath207 transition , unless the non - valence contributions are also calculated .
nevertheless , corrections due to the non - valence configuration are expected to be marginal for heavy - to - heavy form factors evaluated in the
+ " frame where @xmath132 , but become more significant in the @xmath17 " frame ( @xmath133 ) .
the argument goes as follows : we know that the non - valence contribution vanishes if @xmath15 .
now @xmath208 is never zero in the @xmath17 " frame , whereas in the
@xmath209 " frame @xmath15 when @xmath155 [ see ( [ y12val ] ) ] .
that means the valence - quark contribution in the + " frame is _ exact _ at the @xmath7 point . as will be shown in sec .
iv.b , the valence contributions at @xmath7 in the @xmath17 " frame are generally smaller than those in the
+ " frame ; the difference should be accounted for by the non - valence configuration .
these points will be elucidated in more detail in sec .
in sec . 2 we have computed the @xmath0 form factors @xmath89 and @xmath1 form factors @xmath210 for the entire physical @xmath6 region using the light - front wave functions .
it is very important to check if the light - front model predictions are in accord with the requirements of heavy - quark symmetry , namely ( [ pphqs ] ) and ( [ vphqs ] ) . in other words , as @xmath4 , we would like to see if there exists a universal isgur - wise function which governs all heavy - to - heavy mesonic from factors in the light - front quark model . to our knowledge , the isgur - wise function has not been calculated directly for @xmath81 within the framework of the light - front quark model , though it has been considered in @xcite .
the analysis of @xcite is based on the observation @xcite that the knowledge of @xmath0 or @xmath1 form factors at @xmath7 ( or at any point of @xmath6 ) suffices to determine the isgur - wise function in the whole kinematic region .
however , this relies on the assumption that the model calculations of form factors obey heavy - quark symmetry and that the universal form factor is only a function of @xmath211 .
the isgur - wise function is derived in @xcite from _ space - like elastic _ form factors of heavy mesons , .
thus the space - like elastic form factor is related to the isgur - wise function at time - like momentum transfers ( @xmath212 . ] while it is obtained in @xcite by performing an analytic continuation from the region @xmath8 to time - like momentum transfers .
in contrast , we do not impose heavy - quark symmetry from the outset , so that we can check explicitly if the weak decay form factors of heavy mesons can indeed be described by a single universal function when @xmath213 .
we will calculate this universal function directly at the time - like momentum transfer to see if it is independent of heavy quark masses and their ratio .
it is important to note that , since heavy quark - pair creation is forbidden in the @xmath213 limit , the @xmath14-graph is no longer a problem in the reference frame where @xmath214 .
therefore , within the light - front quark model , we are able to compute the isgur - wise function @xmath215 for time - like @xmath6 .
to proceed , we first investigate the heavy - quark - limit behavior of the wave function . in the infinite quark mass limit @xmath4 , the light - front wave function has the scaling behavior [ 17 ] : _
q_(x , k _ )
( m_qx , k _ ) , [ hqamp ] where the factor @xmath216 or @xmath217 ( @xmath48 being the mass of the heavy meson ) comes from the particular normalization we have assumed for the physical state in ( [ wavenor]-[momnor ] ) .
the reason why the light - front heavy - meson wave function should have such an asymptotic form is as follows .
since @xmath51 is the longitudinal momentum fraction carried by the light antiquark , the meson wave function should be sharply peaked near @xmath218 .
it is thus clear that only terms of the form @xmath219
" survive in the wave function as @xmath4 ; that is , @xmath219 is independent of @xmath220 in the @xmath4 limit . for the bsw wave function ( [ bswamp ] )
, we find that ( x , k_)_bsw=(^2)(-k^2_2 ^ 2)(-x^22 ^ 2 ) , [ bswhq ] where @xmath221 , and the normalization condition ( [ momnor ] ) becomes ^_0dx|(x , k _ ) |^2=1 .
[ isnom ] for the gaussian - type wave function ( [ gauss ] ) , we obtain ( x , k_)_gauss = 4(^2)^3/4(-^22 ^ 2 ) .
[ gausshq ] from ( [ kz ] ) it is clear that @xmath222/2 $ ] in the heavy - quark limit . therefore , ( x , k_)_gauss = 4(^2)^3/4(-k^2_2 ^ 2)(-(x2-m_^2 + k^2_2x)^22 ^ 2 ) . however , since @xmath223 , it is clear that the wave function ( [ variant ] ) , which is a variant of the gaussian type , does not have the correct asymptotic form in the heavy - quark limit .
hence it is not suitable for describing heavy - quark transitions . with the light - front wave function
@xmath224 constructed in the @xmath4 limit , the @xmath0 transition function @xmath225 ( [ hi ] ) in the limit of heavy - quark symmetry ( i.e. , @xmath226 ) becomes h(r)=^_0dx(x,k_)(x , k_)a(x)a(x)+k^2 _ , where @xmath227 , and @xmath228 .
note that the quantities @xmath229 and @xmath143 appearing in the integrand are all of order @xmath54 .
denote @xmath230 , then [ see ( [ y12 ] ) ] , z_=v_1v_2 .
[ zpm ] obviously , @xmath231 and @xmath232 .
let @xmath233 , so that ( [ mel1 ] ) can be rewritten as p_2|v^+|p_1=2h(z)v_1^+ . by a simple change of integration variable
, one can readily show that h(z)=zh(1/z ) .
[ relhz ] to check the validity of the heavy - quark - symmetry relation ( [ pphqs ] ) , we note that @xmath234 are related to @xmath235 via h_(q^2 ) & = & 1z1-z^2 , [ hpm ] in analog to ( [ fpm ] ) for @xmath89 . by virtue of ( [ relhz ] ) ,
the hqs relation @xmath236 given in ( [ pphqs ] ) is indeed satisfied , and the isgur - wise function is given by ( v_1v_2)=2h(z)1+z .
[ ppiw ] evidently , the isgur - wise function is independent of the heavy quark masses @xmath237 and their ratio , but it depends on the light spectator quark mass . the r.h.s . of ( [ ppiw ] )
is invariant under the exchange @xmath238 , implying that the isgur - wise function @xmath74 is independent of the choice of the recoiling direction of the daughter meson , as it should be . at zero recoil ( @xmath239 ) ,
the expression for @xmath240 becomes identical to the normalization condition ( [ isnom ] ) .
hence @xmath241 , and the isgur - wise function obeys the correct normalization condition @xmath75 .
we would like to stress again that , unlike the previous works @xcite where @xmath242 is actually evaluated for @xmath243 transition and for space - like values of @xmath6 , here the isgur - wise function is obtained in the infinite quark mass limit and calculated directly for @xmath81 .
within the specific model we have taken , our result is exact . in the limit of heavy - quark symmetry , form factors @xmath76 and @xmath77 are related to the isgur - wise function via ( v_1v_2)=2m_1+m_2f_1(q^2)=2m_1+m_2f_0(q^2 ) .
[ ppiwform ] hence the @xmath6 dependence of @xmath76 is different from that of @xmath77 by an additional pole factor .
there are four hqs relations given in ( [ vphqs ] ) for @xmath1 form factors .
we shall first focus on @xmath244 ( or @xmath122 ) . as @xmath245
, we can show that _ v(x,k_)(x,k _ ) , _
p(x , k _ ) ( x , k _ ) , with @xmath148 being defined in ( [ thetav ] ) .
the @xmath1 transition amplitude @xmath134 [ see ( [ ii1 ] ) and ( [ ii2 ] ) ] reduce to i(r ) & = & ( z)=-1[xa(x)+x(x - x)(x,k_)k^2_]f(x)f(x ) + & = & -1[xa(x)-x(x - x)(x , k_)k^2_]f(x)f(x ) , where @xmath246 , and all terms proportional to @xmath247 have been neglected in the heavy - quark limit .
it is evident that @xmath248 satisfies the relation i(z)=i(1/z ) .
therefore , from ( [ apm ] ) , _ + + _ -=2m_p^2r_+r_-i ( z_+)-r_+r_-i(z_-)r_+-r_-=0 , [ hqap ] and _
+ -_-=2-r_+i(z_+)+r_-i(z_-)r_+-r_-=-2i(z ) .
[ hqam ] by comparing this with ( [ vphqs ] ) yields the isgur - wise function ( vv ) & = & 2^_0 dx(x,k_)(x , k _ )
+ & & \{xa(x)+x(x - x)(x,k _ ) k^2_}. [ vpiw ] with @xmath232 .
it remains to show that @xmath249 is indeed the same as the isgur - wise function @xmath242 found in @xmath0 transition ( [ ppiw ] ) .
we will address this issue later in sec .
iv . after showing the hqs relations ( [ hqap ] ) and ( [ hqam ] ) for form factors @xmath244
, we turn to the vector form factor .
one can easily show from ( [ gq2 ] ) that , indeed , ( q^2)=g(q^2 ) -2(z)=(vv ) , in accord with ( [ vphqs ] ) . using the results ( [ hqap ] ) and ( [ hqam ] ) for form factors @xmath244 , we are ready to prove the remaining hqs relation for @xmath128 .
it follows from ( [ vpform ] ) , ( [ fq2 ] ) and ( [ jq2]-[cd ] ) that ( q^2)=f(q^2)&=&-12x^2-x^2xx + ( x,k _ ) ( x , k_)(xm_vm_1 c ) , where terms proportional to @xmath247 vanish in the limit of heavy - quark symmetry . we find from ( [ cd ] ) that c -2xx(a_pa_v+k^2 _ ) , hence ( q^2)=-12x^2-x^2xx(vv)-x x(1+xx)(vv ) , where use of ( [ ppiw ] ) has been made .
then , using @xmath232 and ( [ zpm ] ) , we are led to the desired hqs relation given in ( [ vphqs ] ) : ( q^2)=-(1+vv)(vv ) , provided that @xmath250 . is the function @xmath249 given by ( [ vpiw ] ) identical to the isgur - wise function @xmath242 ?
while @xmath75 is always valid irrespective of the details of the light - front amplitude used , the normalization of @xmath249 at zero recoil is nontrivial . in fact
, we find that @xmath251 depends on the choice of the light - front model wave function .
we find numerically ( see sec .
iv.a ) that the hqs requirement @xmath252 is fulfilled by the gaussian - type wave function ( [ gauss ] ) , but not so by the bsw - type wave function ( [ bswamp ] ) . in other words , _ the normalization of the isgur - wise function at zero recoil in @xmath1 transition puts a severe restriction on the phenomenological light - front wave functions .
_ since we are not able to solve the light - front qcd bound - state equation to obtain the momentum distribution amplitude @xmath66 , we see that heavy - quark symmetry is helpful in discriminating between different phenomenological amplitudes . as will be shown in sec .
iv.a , @xmath253 is numerically equal to @xmath242 if the gaussian - type wave function is used . the @xmath1 form factors in the heavy - quark limit are all related to the isgur - wise function via ( vv ) & = & 2m_p+m_vv(q^2)= 2m_p+m_va_0(q^2 ) + & = & 2m_p+m_va_2(q^2)= 2m_p+m_va_1(q^2 ) .
[ hqsvp ] that means @xmath254 all have the same @xmath6 dependence and they differ from @xmath109 by an additional pole factor .
to examine numerically the form factors derived in the last section , we need to specify the parameters appearing in the phenomenological light - front wave functions .
we shall use the decay constants to constrain the quark mass @xmath255 and the scale parameter @xmath53 .
the decay constants of light pseudoscalar and vector mesons are is obtained from the measured decay rate of @xmath256 , while @xmath257 is determined from @xmath258 . ]
f_=132mev , f_k=160mev , f_=216mev , f_k^*=210mev .
[ fpl ] the decay constants of heavy mesons are unknown experimentally , so we have to rely on model calculations and lattice results . to be specific , we take f_d=200mev , f_b=185mev , f_d^*=250mev , f_b^*=205mev , [ fph ] where the estimates for @xmath259 and @xmath260 are relatively more uncertain .
the parameters @xmath255 and @xmath53 in the gaussian - type and bsw - type wave functions fitted to the decay constants via ( [ fp ] ) and ( [ fv ] ) are listed in table i. note that the quark masses given in table i are fixed to the commonly used values , and the other fitted values are by no means unique .
presumably , other hadronic properties , for example the light - meson elastic form factor measured at a wide range of momentum transfer , would be helpful in fixing the light - front parameters . 0.5 cm table i. parameters @xmath255 ( in units of gev ) and @xmath53 in the gaussian - type and bsw - type wave functions fitted to the decay constants given by ( [ fpl ] ) and ( [ fph ] ) . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] 0.5 cm form factors for @xmath261 and @xmath262 transitions at @xmath7 predicted in various approaches ( lattice simulations , qcd sum rule , quark model , light - front quark model , and heavy quark effective theory together with chiral perturbation theory ) are summarized in table ii .
we have to await further experimental studies in order to test various models .
the @xmath263 and @xmath264 transitions arise from flavor - changing neutral currents induced by qcd corrections .
it has been found recently that there are two experimental data for @xmath265 which can not be accounted for simultaneously by all commonly used models @xcite .
hence , it is important to have a reliable estimate of the @xmath266 form factors at @xmath267 in order to test the validity of the factorization approach .
our calculation gives f_1^bk^*(m^2_j/)=0.66 , & & v_2^bk^*(m^2_j/)=0.42 , + a_0^bk^*(m^2_j/)=0.63 , & & a_1^bk^*(m^2_j/)=0.37 , a_2^bk^*(m^2_j/)=0.43 , from valence - quark configuration .
as for the @xmath6 dependence of heavy - to - light from factors , we see from figs . 8 - 11 that , except for @xmath268 , they all increase with @xmath6 , though @xmath109 is flatter than @xmath269 , and @xmath3 . as we have argued before , the valence - quark contribution evaluated in the
+ " frame should be reliable when @xmath6 is close to maximum recoil . for small @xmath6
, we have a dipole behavior for @xmath270 ( except for @xmath268 and @xmath271 ) and a monopole behavior for @xmath109 ; that is , @xmath269 , and@xmath272 increase with @xmath6 faster than @xmath109 .
the form factor @xmath3 for @xmath261 and @xmath264 in the
+ " frame do not have a dipole behavior at small @xmath6 mainly because of the large destructure contributions from the @xmath273 terms in ( [ gq2 ] ) . as a result ,
the form factor @xmath3 in @xmath261 and @xmath264 decays evaluated in the + " frame is smaller than that in the @xmath17 " frame .
the @xmath6 dependence of the @xmath1 form factors have also been studied in the qcd - sum - rule approach with some contradicting results .
for example , while @xmath274 is found to decrease from @xmath7 to @xmath275 in @xcite ( see also @xcite ) , such a phenomenon is not seen in @xcite ( see also sec .
5.3 of @xcite ) .
the sum - rule results of @xcite show that the form factors @xmath270 all have a dipole form while @xmath109 has a monopole form , in accord with our observation .
the same conclusion is also reached in @xcite based on the scaling behavior of heavy - to - light from factors in the @xmath4 limit .
a recent lattice study of the axial form factors @xmath276 and @xmath277 @xcite is consistent with the @xmath6 behavior we have in the light - front quark model .
the heavy - to - heavy and heavy - to - light form factors in @xmath0 and @xmath1 transitions are studied in the present paper . in the light - front relativistic quark model ,
the decay form factors are evaluated in a frame where @xmath278 and @xmath12 , so that it covers the entire physical range of momentum transfer and no extrapolation assumption from @xmath7 or from @xmath279 is required . in previous works using @xmath15 , one can only calculate form factors at @xmath6 = 0 ; moreover , the form factors @xmath101 in @xmath0 decay and @xmath280 in @xmath1 decay can not be studied . for the first time
, we have calculated the @xmath1 form factors directly at time - like momentum transfers .
the main results of this paper are : 1 ) .
we have investigated the behavior of heavy - to - heavy form factors in the heavy - quark limit and found that the requirements of heavy - quark symmetry ( [ pphqs ] ) for @xmath0 transition and ( [ vphqs ] ) for @xmath1 transition are indeed fulfilled by the light - front quark model provided that the universal function @xmath249 obtained from @xmath1 decay is identical to the isgur - wise function @xmath242 in @xmath0 decay . 2 ) .
contrary to the isgur - wise function in @xmath0 decay , the normalization of @xmath249 at zero recoil depends on the light - front wave function used .
we found that the bsw amplitude correctly gives @xmath75 , but @xmath281 .
therefore , this type of wave functions can not describe @xmath1 decays in a manner consistent with heavy - quark symmetry .
3 ) . using the gaussian - type amplitude
, the isgur - wise function @xmath282 has a correct normalization at zero recoil and is identical to @xmath242 numerically up to six digits .
it can be fitted very well with a dipole dependence with @xmath283 gev for @xmath243 transition .
however , the predicted slope parameter @xmath284 is probably too large .
this may be ascribed to the fact that the gaussian - type wave function does not have enough high - momentum components at large @xmath143 .
the valence - quark and non - valence contributions to form factors are in general dependent on the recoiling direction of the daughter meson relative to the parent meson , but their sum should not . although we do not have a reliable estimate of the pair - creation effect , we have argued that , for heavy - to - heavy transition , form factors calculated from the valence - quark configuration evaluated in the
+ " frame should be reliable in a broad kinematic region , and they become most trustworthy in the vicinity of maximum recoil .
the form factors @xmath285 ( except for @xmath268 and @xmath271 ) all exhibit a dipole behavior , which @xmath77 and @xmath286 show a monopole behavior in the close vicinity of maximum recoil for heavy - to - light transition , and in a broader kinematic region for heavy - to - heavy decays .
therefore , @xmath287 increase with @xmath6 faster than @xmath77 and @xmath109 .
99 iw n. isgur and m.b .
wise , , 113 ( 1989 ) ; * b237 * , 527 ( 1990 ) .
neu94 m. neubert , _ phys . rep . _ * 245 * , 261 ( 1994 ) .
terentev , _ sov .
_ * 24 * , 106 ( 1976 ) ; v.b .
berestetsky and m.v .
terentev , _ ibid . _
* 24 * , 547 ( 1976 ) ; * 25 * , 347 ( 1977 ) .
chung , f. coester , and w.n .
polyzou , , 545 ( 1988 ) .
w. m. zhang , _ chin .
j. phys . _
* 31 * , 717 ( 1994 ) ; preprint ip - astp-19 - 95 [ hep - ph/9510428 ]
. w. jaus , , 3394 ( 1990 ) ; _ ibid . _ * d44 * , 2851 ( 1991 ) ; , 611 ( 1992 ) .
don94a p.j .
odonnell and q.p .
xu , , 219 ( 1994 ) .
don94b p.j .
odonnell and q.p .
xu , , 113 ( 1994 ) . don95 p.j .
odonnell , q.p .
xu , and h.k.k .
tung , , 3966 ( 1995 ) .
jaus96 w. jaus , , 1349 ( 1996 ) .
mel d. melikhov , , 2460 ( 1996 ) ; hep - ph/9603340 ; hep - ph/9607216 ( 1996 ) .
cheung , c.w .
hwang , and w.m .
zhang , hep - ph/9602309 ( 1996 ) .
demchuk , i.l .
grach , i.m .
narodetskii , and s. simula , infn - iss 95/18 [ hep - ph/9601369 ] .
grach , i.m .
narodetskii , and s. simula , infn - iss 96/4 [ hep - ph/9605349 ] .
a. dubin and a. kaidalov , _ yad .
* 56 * , 164 ( 1993 ) [ _ phys . at .
_ * 56 * , 237 ( 1993 ) ]
. m. sawicki , , 433 ( 1991 ) .
cheung , w.m .
zhang , and g.l .
lin , , 2915 ( 1995 ) .
m. wirbel , b. stech , and m. bauer , , 637 ( 1985 ) ; m. bauer , b. stech , and m. wirbel , _ ibid . _
* 42 * , 671 ( 1989 ) .
simc s. simula , , 193 ( 1996 ) .
neu92 m. neubert and v. rieckert , , 97 ( 1992 ) . neu96 m. neubert , cern - th/96 - 55 [ hep - ph/9604412 ] . n. isgur and m. b. wise , , 151 ( 1990 ) .
pdg particle data group , , 1173 ( 1994 ) .
ra cleo collaboration , m.s .
et al . , _ , 1311 ( 1993 ) .
rb cleo collaboration , f. butler _
et al . , _ , 2656 ( 1995 ) .
chau and h.y .
cheng , , 514 ( 1994 ) .
r12 cleo collaboration , j.e .
et al . , _ , 3898 ( 1996 ) .
bes g. bernard , a. el khadra , and a. soni , , 2140 ( 1992 ) .
v. lubicz , g. martinelli , m.s .
mcarthy , and c.t .
sachrajda , , 415 ( 1992 ) .
et al . , _ , 675 ( 1994 ) .
ape collaboration , c.r .
et al . , _ , 513 ( 1995 ) .
ukqcd collaboration , d.r .
et al . , _ , 425 ( 1995 ) ; j.m .
et al . , _ , 327 ( 1996 ) ; j.m .
flynn and j. nieves , ug - dfm-3/96 [ hep - ph/9602201 ] .
gss s. gsken , g. siegert , and k. schilling , b ( proc . suppl . )
* 47 * , 485 ( 1996 )
slobodenyuk , _ sov .
phys . _ * 51 * , 696 ( 1990 ) .
p. ball , , 3190 ( 1993 ) .
abs a. ali , v.m .
braum , and h. simma , , 437 ( 1994 ) .
s. narison , , 166 ( 1995 ) .
yang and w - y.p .
hwang , nuthu-94 - 17 , to appear in _ z. phys . _
isgw n. isgur , d. scora , b. grinstein , and m.b .
wise , , 799 ( 1989 ) .
stech b. stech , , 447 ( 1995 ) ; hd - thep-95 - 54 [ hep - ph/9512414 ] .
faustov , v.o .
galkin , and a.yu .
mishurov , , 6302 ( 1996 ) ; , 516 ( 1995 ) .
iv m.a . ivanov and yu.m .
valit , hep - ph/9606404 .
r. casalbuoni , a. deandrea , n. di bartolomeo , r. gatto , f. feruglio , and g. nardulli , , 139 ( 1993 ) .
r. casalbuoni , a. deandrea , n. di bartolomeo , r. gatto , f. feruglio , and g. nardulli , ugva - dpt 1996/05 - 928 [ hep - ph/9605342 ] .
gour m. gourdin , a.n .
kamal , and x.y .
pham , , 3355 ( 1994 ) ; r. aleksan , a. le yaouanc , l. oliver , o. pne , and j .- c .
raynal , , 6235 ( 1995 ) .
bbd p. ball , v.m .
braun , and h.g .
dosh , , 3567 ( 1991 ) . * fig .
1 * the feynman triangle diagram and the corresponding light - front subdiagrams . diagram ( a ) corresponds to the valence - quark configuration and diagram ( b ) to the non - valence configuration .
0.25 true cm * fig .
2 * the isgur - wise function @xmath242 for @xmath0 transition calculated using gaussian - type ( solid line ) and bsw - type ( dashed line ) light - front wave functions . for comparison , a curve for @xmath288
is also shown .
0.25 true cm * fig .
4 * form factors @xmath289 and @xmath290 for @xmath243 transition arising from the valence - quark configuration . dashed
curves are fits to @xmath76 in a dipole form with @xmath291 gev and to @xmath77 in a monopole form with @xmath292 gev .
0.25 true cm * fig .
6 * form factors @xmath296 and @xmath297 for @xmath207 transition .
solid lines are the valence contribution evaluated in the
+ " frame where @xmath203 , and dashed lines in the @xmath17 " frame where @xmath204 .
the contribution to the form factor @xmath110 is independent of the choice of the + " or
@xmath17 " frame .
0.25 true cm * fig . 7 *
an illustration of the non - valence contribution to the form factor @xmath298 , whose general feature applies to @xmath299 and @xmath300 as well .
valence - quark contributions to @xmath298 evaluated in the
+ " frame ( solid line ) and in the @xmath17 " frame ( dashed line ) are the same as in fig.6 .
the corresponding non - valence contribution are extracted in respective frames by assuming that the full @xmath298 has a dipole behavior with @xmath301 = 6.73 gev ( dash - dotted line ) .
0.25 true cm | within the light - front framework , form factors for @xmath0 and @xmath1 transitions ( @xmath2 : pseudoscalar meson , @xmath3 : vector meson ) due to the valence - quark configuration are calculated directly in the entire physical range of momentum transfer .
the behavior of the form factors in the infinite quark mass limit are examined to see if the requirements of heavy - quark symmetry are fulfilled .
we find that the bauer - stech - wirbel type of light - front wave function fails to give a correct normalization for the isgur - wise function at zero recoil in @xmath1 transition .
some of the @xmath1 form factors are found to depend on the recoiling direction of the daughter mesons relative to their parents .
thus , the inclusion of the non - valence contribution arising from quark - pair creation is mandatory in order to ensure that the physical form factors are independent of the recoiling direction .
the main feature of the non - valence contribution is discussed .
0.25 true cm pacs numbers : 13.20 , 14.40.j 6.5 in 1.0 cm .29 in |
in recent years it has become clear that the gas - phase d / h ratio varies significantly from one sightline to another within a few hundred pc of the sun ( see , e.g. , moos et al .
2002 ; steigman 2003 ; wood et al . 2004 ; hbrard 2005 ) .
this has often been taken to result from regional variations in the extent to which nucleosynthesis in stars , followed by return of matter to the interstellar medium via stellar winds and ejecta , has lowered the interstellar d / h ratio from the primordial value established by nucleosynthesis in the big bang .
i argue here that the observed local variations in the gas - phase d / h ratio are not due to variable astration .
the low values of ( d / h)@xmath3 seen on some sightlines are instead attributed to incorporation of d into dust grains .
i discuss processes acting to add d to dust grains , and show that these can strongly deuterate pah material in the interstellar medium .
this hypothesis predicts a positive correlation of ( d / h)@xmath3 with gas - phase abundances of strongly - depleted metals , and deuteration of carbonaceous interstellar grains captured by the _ stardust _ mission .
using high resolution spectroscopy with imaps , jenkins et al .
( 1999 ) measured d / h=@xmath4ppm toward @xmath5 ori ( @xmath6pc ) , and sonneborn et al .
( 2000 ) found d / h=@xmath7ppm toward @xmath8 vel ( @xmath9pc ) .
these two high precision measurements firmly established the reality of substantial variations in the gas - phase d / h ratio within a few hundred pc of the sun . including observations with copernicus , imaps , hst , and fuse ( see the summary by wood et al . 2004 ) we now have measurements of d / h on @xmath040 interstellar sightlines , with d / h values ranging from @xmath10ppm toward @xmath11 car ( allen , jenkins , & snow 1992 ) to @xmath7ppm toward @xmath8 vel a factor of @xmath04 variation in d / h .
the gas returned to the ism from stars is expected to be nearly devoid of d , as the d is converted to @xmath12he during the pre - main - sequence evolution of @xmath13 stars ( mazzitelli & moretti 1980 ) . as a result , the d / h ratio in the ism is an indicator of the fraction of the baryons now in the ism that have passed through a star .
if the primordial value of d / h is @xmath14ppm ( spergel et al . 2003 ) then a d / h ratio of @xmath10ppm would require that @xmath15 of the baryons in the interstellar gas toward @xmath11 car have been cycled through a star at least once , whereas on the sightline toward @xmath8 vel , this fraction would be only @xmath16 .
the hypothesis that the observed variations in d / h on different sightlines are due to variations in astration therefore requires that ( 1 ) regions separated by only a few hundred pc have had extremely different star formation histories , and ( 2 ) turbulent diffusion has not homogenized the elemental abundances in the gas .
it is difficult to see how these two conditions can be true : we do not see other evidence of such variations in star formation activity , and turbulent mixing is expected to be effective at mixing gas over length scales of hundreds of pc on @xmath17yr time scales .
the gas returned to the ism from stars will , on average , be enriched in the products of stellar nucleosynthesis , such as oxygen .
models for evolution of interstellar abundances due to stellar nucleosynthesis and infall predict the joint variation of d / h and o / h .
outside of dark clouds , h@xmath18o is absent , and the o in grains is thought to be mostly in the form of amorphous olivine mg@xmath19fe@xmath20sio@xmath21 ( see draine 2003 ) . on typical sightlines 7099% of the si is in grains ( jenkins 2004 ) ; thus o@xmath22/o@xmath23(si / o)@xmath24(si / o)@xmath25 : the silicate grains contain @xmath26 of the oxygen ( for ( si / h)@xmath27ppm from asplund 2000 , and ( o / h)@xmath28ppm from asplund et al
. 2004 ) .
the gas phase oxygen abundance tracks the total oxygen abundance : ( o / h)@xmath29(o / h)@xmath30 .
figure [ fig : dvso ] shows tracks of d / h vs. o / h for two chemical evolution models ( chiappini , renda , & matteucci 2002 ) , where the degree of astration is a decreasing function of galactocentric radius .
these models , which include infall , have o / h varying from 300 to 600@xmath1ppm as d / h declines by only @xmath31 .
it is difficult to envision astration scenarios where d / h varies by a factor of 4 without associated very large variations in o / h .
however , unlike the d / h ratio , the o / h ratio does _ not _ show large variations .
figure [ fig : dvso ] shows measured values of ( d / h)@xmath3 and ( o / h)@xmath3 for 14 sightlines where @xmath32 , @xmath33 , and @xmath34 are all reported with 1-@xmath35 uncertainties not exceeding 0.10 dex . for this sample , the median
( o / h)@xmath3 is @xmath36ppm and the median ( d / h)@xmath3 is @xmath37ppm .
there is no indication of a significant anticorrelation between ( o / h)@xmath3 and ( d / h)@xmath3 : * two of the three sightlines with the lowest values of ( d / h)@xmath3 ( @xmath5 ori and hd 191877 ) have ( o / h)@xmath3 _ below _ the median . * the sightline with the highest ( d / h)@xmath3 in this sample ( lse 234 ) has the second - highest ( o / h)@xmath3 .
* the full ( factor of @xmath03 ) range of variation of ( d / h)@xmath3 is seen among the 6 sightlines with ( o / h)@xmath3 between 300 and @xmath38ppm ( @xmath5 ori , hd 191877 , @xmath39 ori , gd 246 , hz 43a , and feige 110 ) . it therefore seems very unlikely that the observed variations in ( d / h)@xmath3 are due to variations in astration of the interstellar gas .
if the ( d / h)@xmath3 variations are real , some other process must be responsible .
when we observe variations in the gas - phase abundance of , say , si or fe from one local sightline to another , we do not attribute it to variable astration , but instead to variations in the fraction of the si and fe `` depleted '' from the gas by incorporation in solid dust particles .
jura ( 1982 ) suggested that even d might perhaps be depleted from the gas phase and sequestered in dust grains . since
1982 our picture of interstellar grains has evolved to include a substantial population of very small polycyclic aromatic hydrocarbon ( pah ) grains ( see the review by draine 2003 ) .
draine ( 2004 ) argued that carbonaceous interstellar grains could incorporate enough d atoms to substantially reduce the d abundance in the gas phase . under sufficiently violent conditions , the `` depleted '' d atoms could later be released from grains and returned to the gas . in this interpretation
, astration has reduced the d / h abundance from the primordial value to ( d / h)@xmath40ppm ; incorporation of some of the d into dust will then result in ( d / h)@xmath41 .
the _ variations _ in the gas - phase d / h ratio are attributed to variations from one sightline to another in the fraction of the d sequestered in dust grains .
measurement of the d / h ratio using ultraviolet spectroscopy is necessarily limited to sightlines through `` diffuse '' clouds with @xmath42mag .
grains in clouds with @xmath43mag do not have h@xmath18o ice mantles ( whittet et al . 1988 ) ; the hydrogen present in diffuse cloud grains must be primarily in hydrocarbons .
in fact , a broad absorption feature at 3.4@xmath44 m is observed and is identified as the c - h stretching mode in aliphatic hydrocarbons , although the amount of material necessary to account for the observed feature is uncertain ( see , e.g. , pendleton & allamandola 2002 ) .
in addition to the 3.4@xmath44 m absorption feature , we also observe emission features at 3.3 , 6.2 , 7.7 , 8.6 , and 11.3@xmath44 m that are attributed to vibrationally - excited polycyclic aromatic hydrocarbon ( pah ) molecules .
models that quantitatively reproduce the observed emission require substantial amounts of c in such particles ; the model of li & draine ( 2001 ) has c / h@xmath45ppm in pah grains containing @xmath46 c atoms refers to all of the h nucleons , including those in the gas . ]
the larger pah grains are presumed to be clusters of smaller pah molecules .
these might have c : h ratios in the range of 2:1 4:1 ( e.g. , coronene c@xmath47h@xmath48 to dicircumcoronene c@xmath49h@xmath47 ) .
if so , the very small pah grains alone may contain h / h@xmath5020@xmath1ppm .
how much c is in all of the grain material ? if it is assumed that the total interstellar c abundance is equal to current estimates of the solar abundance ( c / h=@xmath51ppm ; allende prieto , lambert , & asplund 2002 ) , then the difference between this value and the gas - phase abundances of ci and cii determined from absorption line measurements allow one to estimate that dust contains c / h@xmath52ppm ( jenkins 2004 ) .
however , ( 1 ) solar abundances are uncertain , and ( 2 ) interstellar abundances may differ from solar abundances . when one seeks to construct a grain model to reproduce the observed extinction per h nucleon as a function of wavelength
, one finds that in addition to silicate material , it is necessary to include a substantial amount of carbonaceous material .
the dust model of weingartner & draine ( 2001a ) assumes c / h@xmath53ppm in dust ( including the small pah particles ) .
the dust models of zubko , dwek & arendt ( 2004 ) use c / h@xmath54@xmath55ppm .
pendleton & allamandola ( 2002 ) estimate that about 85% of the c in dust is aromatic ( @xmath56 bonding , e.g. , pahs and graphite ) , and 15% is aliphatic ( `` chain hydrocarbons '' ) .
they estimate a c : h ratio @xmath02.9:1 for the aromatic material , and @xmath00.46:1 for the aliphatic material . for purposes of discussion
, i will assume c / h@xmath57ppm and ( h+d)/h@xmath58ppm in aromatic hydrocarbon material , with an additional c / h@xmath59ppm and h / h@xmath60ppm in aliphatic material . because the smallest grains , which contribute most of the surface area , appear to be pahs ,
i will focus on deuteration of the aromatic hydrocarbons alone , and neglect possible deuteration of the aliphatic material .
observed values of ( d / h)@xmath3 range from @xmath022@xmath1ppm to @xmath61ppm.ppm toward @xmath8 vel ( sonneborn et al .
2000 ) , @xmath62ppm toward feige 110 ( friedman et al . 2002 ) , and @xmath63ppm toward lse 234 ( lecavelier et al .
some low values : @xmath10ppm toward @xmath11 car ( allen et al .
1992 ) , @xmath4ppm toward @xmath5 ori ( jenkins et al .
1999 ) , @xmath64ppm toward lse 1274 ( wood et al .
2004 ) . ] for dust to account for the variations in ( d / h)@xmath3 , the dust grains must be able to hold at least d / h@xmath65ppm . if the total amount of hydrogen in the aromatic material is h / h@xmath66ppm , then it must be possible to enrich the deuterium concentration to the point where ( d / h)@xmath67 .
given that the overall d / h ratio is of order @xmath68 , processes must act to increase the d / h ratio in the dust by 4 orders of magnitude !
examples of extreme deuteration are already known .
for example , fragments of the hydrocarbon material in one interplanetary dust particle ( idp ) analyzed in the laboratory have d / h @xmath69 ( messenger 2000 ) . in some molecular cloud cores , extreme deuteration of small gas - phase molecules
is seen , e.g. , d@xmath18co
/ h@xmath18co = 0.01 0.1 ( ceccarelli et al . 2001 ; bacmann et al .
2003 ) , nh@xmath18d / nh@xmath70 = 0.07 ( shah & wooten 2001 ) , nd@xmath70/[email protected] ( lis et al . 2002 ; van der tak et al .
2002 ) , and ch@xmath18doh / ch@xmath70oh@xmath72 ( parisi et al .
2002 ) .
sandford , bernstein , & dworkin ( 2001 ) discussed 4 processes that could lead to deuterium enrichment in carbonaceous material : low - temperature gas phase ion - molecule reactions ; low - temperature gas - grain reactions ; uv photon absorption followed by ejection of h ; and deuteration of carbon molecules embedded in d - rich ices .
i focus here on processes that can deuterate grains in diffuse clouds , and show that d could be removed from the gas on timescales as short as a few myr . while h and d have the same electronic structure , and therefore the same `` chemistry '' , the c - d bond is stronger than the c - h bond because of the reduced zero - point energy .
it is easy to estimate the magnitude of this energy difference . to be specific ,
let us consider pah molecules , since we have direct evidence that at least some of the solid carbon is in this form .
pahs show three emission features associated with vibrations of the c - h bond : the c - h stretch at @xmath73 m , in - plane c - h bending at @xmath74 m , and out - of - plane c - h bending at @xmath75 m .
the stretching and bending modes can be approximated as harmonic oscillators .
if the h is replaced by d , the reduced mass of the c - h oscillator will increase by a factor 13/7 , and the frequency will be reduced by a factor @xmath76 .
each of the vibrational modes has zero - point energy @xmath77 associated with it , so ( summing over the three modes ) the zero - point energy will be lowered by @xmath78 therefore it is energetically favorable for the h in pahs to be replaced by d. if atomic h and d in the gas were in thermodynamic equilibrium with deuterated pah material at temperature @xmath79 , we would expect to have @xmath80 where the inequality is because we may not reach a steady - state , and there may be other processes acting to return d to the gas .
now suppose that ( d / h)@xmath30=22@xmath1ppm .
if the aromatic grains contain d / h@xmath81ppm and h / h@xmath82ppm , eq .
( [ eq : condition on tdust ] ) becomes @xmath83 or @xmath84 = 91$]k .
the ism is , of course , far from lte , so thermodyamic arguments are suspect . nevertheless ,
since interstellar grain material is generally at temperatures @xmath85k , we can at least contemplate the possibility that pah grains exposed to atomic h and d might attain d : h ratios as high as @xmath00.3 .
it can easily be shown that thermal desorption also can not be important .
at such temperature @xmath86k , the rate for thermal desorption of chemically - bound h in pahs is entirely negligible .
therefore it does not appear that the required degree of deuteration can be achieved by the thermal desorption process discussed by allamandola , tielens , & barker ( 1989 ) . because grains can catalyze h@xmath18 formation , it is important to note that the energy difference @xmath87 is 0.048 ev smaller than @xmath88 , so that it is energetically favorable to leave the d attached to c and instead abstract an h to form h@xmath18 rather than hd ( 2.4@xmath44 m is the wavelength of h@xmath181 - 0 q branch lines ) .
thus even if the h@xmath18 formation process involves hydrogen abstraction , h@xmath18 will be strongly favored over formation of hd , leaving d atoms chemically bound in the hydrocarbon grain .
this differs from the prediction by lipshtat , biham , & herbst ( 2004 ) of enhanced hd and d@xmath18 formation if h and d are only physisorbed on the grain surface .
now the key question is : can we imagine a kinetic pathway whereby extreme deuteration of grains might come about ? while it is known that pahs can form in the high - temperature chemistry of flames , the processes responsible for the formation of interstellar pahs and other carbonaceous material remain unclear .
we will assume that at the time of their formation , the carbon solids are minimally deuterated , incorporating h and d approximately in proportion to gas - phase abundances .
given that carbonaceous grains incorporate only @xmath0145@xmath1ppm of all of the hydrogen , a negligible fraction of the d would initially be trapped in pahs . for grains to become heavily deuterated , it is necessary that 1 .
d or d@xmath89 arriving at a carbonaceous grain must have a significant probability of being incorporated into the grain ; 2 .
d that has been incorporated must not be removed by exchange reactions with arriving h or h@xmath89 if this were to happen with any appreciable probability , the steady - state abundance of d in the grain would be very low .
we consider conditions in cold hi clouds ( the `` cold neutral medium '' , or cnm ) , with density @xmath90 , electron density @xmath91 , with perhaps 1/3 of the electrons contributed by ionized metals , and 2/3 by ionized hydrogen .
thus the hydrogen ionization fraction @xmath92 ; because of rapid charge exchange , deuterium will have the same ionization fraction .
we take the local average starlight background .
impinging low - energy d atoms are probably unable to react chemically with an already saturated neutral hydrocarbon
the arriving d will most likely be physisorbed or chemisorbed on the surface of the neutral pah , from which it can subsequently be removed by reacting with an arriving h to form hd , or by thermal desorption following heating of the pah by photon absorption .
as discussed by bauschlicher ( 1998 ) for the naphthalene cation c@xmath93h@xmath94 , a d atom can react with a pah cation , producing a hydro - pah cation , with two hydrogens sharing a single carbon ( see figure [ fig : react ] ) .
how rapidly will this occur ?
it is not known if thermal d atoms will react with pahs much larger than naphthalene if they are only singly or doubly charged , so let us consider only pah cations containing @xmath95 c atoms .
collision cross sections are estimated following draine & sutin ( 1987 ) , the pah size distribution is from li & draine ( 2001 ) and weingartner & draine ( 2001a ) , and the pah charge distribution is estimated following weingartner & draine ( 2001b ) . for a neutral d in the cnm , i find the rate to collide and react with pah cations containing @xmath96 atoms is @xmath97 ; if the probability of reaction is of order unity , and there is no process returning d to the gas phase , then the gas - phase d abundance in the cnm would decline by a factor @xmath98 in only @xmath99myr .
another pathway begins with d@xmath89 colliding with a neutral ( or negative ) pah ; the d@xmath89 will arrive at the pah with sufficient kinetic energy that atomic rearrangements can take place among the peripheral hydrogens near the point of impact .
the excess energy will be thermalized in the vibrational modes ; the collision complex is expected to maximize its binding energy by an exchange reaction that would replace a c - h bond with a c - d bond , thus incorporating the d and creating a hydro - pah cation just as for the case of a neutral d colliding with a pah cation .
it seems likely that every arriving d@xmath89 ion will be incorporated into the pah via this pathway , thus deuterating the pah and removing d from the gas .
the rate for species @xmath100 to collide with a grain can be written @xmath101 , where @xmath102 depends on the pah charge distribution .
ions have relatively high rates for collisions with neutral pahs because of the induced polarization interaction , and even higher rates for negative pah ions because of coulomb focussing .
weingartner & draine ( 2001c ) estimated @xmath103 in the cnm ; @xmath104 will be smaller by @xmath105 , @xmath106 .
thus the rate for d nuclei to react with a grain by this pathway is @xmath107 .
this is a factor of 16 slower than the rate for d nuclei to react with pah cations , and therefore is of only secondary importance .
together , the gas - phase processes appear able to remove d from the gas , and deuterate pahs , on a time scale of @xmath108myr .
evolution of the grain population is not well understood , but it appears that the larger grains move relative to the gas with speeds of @xmath109 resulting from mhd turbulence ( yan et al .
this can sweep up the pahs on @xmath110yr timescales and incorporate them into larger grains ; occasional shattering of larger carbonaceous grains when they collide replenishes the pah population . in this scenario ,
all of the aromatic grain material eventually becomes enriched with d , not just the fraction that is instantaneously present in small free - flying pahs .
the proposed extreme deuteration of pah material imposes strong requirements on branching ratios for reactions of pah material with impinging h and d atoms and ions , and is therefore subject to laboratory test .
in particular , an experiment with low energy d@xmath89 impinging on partially - deuterated pah material should have a high probability of incorporating the d into the solid , while reactions of low energy h@xmath89 with the same sample should have a very small probability of abstracting d from the sample .
grains do nt last forever sputtering and grain - grain collisions following the passage of @xmath111 shock waves can return atoms including d from the solid to the gas phase .
when this occurs , ( d / h)@xmath3 should increase , along with the gas - phase abundances of other elements , such as si , ti , and fe , that are normally strongly depleted . the hypothesis that dust is responsible for the observed variations in (
d / h)@xmath3 therefore predicts that the gas - phase abundances of si , ti , fe and other refractory metals should show a positive correlation with the gas - phase abundance of d. this prediction appears to be confirmed by a recent study finding a positive correlation between ( d / h)@xmath3 and ( fe / h)@xmath3 ( linsky et al .
the emission features at 3.3 , 8.6 , and 11.3@xmath44 m provide direct spectroscopic evidence for the presence of c - h bonds in vibrationally - excited interstellar pah material .
can c - d bonds be observed through corresponding emission features ?
the c - d in - plane bending mode at @xmath112 would be confused with the c - h out - of - plane bending modes at @xmath113 , and the out - of - plane bending mode at @xmath114 falls in a region where other pah features ( c - c skeleton modes ) are present .
the best hope appears to be the c - d stretching mode at @xmath115 .
in fact , peeters et al . ( 2004 ) report detection of weak emission features at 4.4@xmath44 m and 4.68@xmath44 m from the orion bar and m17 photodissociation regions .
identification of the features as due to c - d is uncertain , but if the features are indeed due to c - d modes , peeters et al.estimate ( d / h)@xmath116 in orion and @xmath117 in m17 .
these d / h values are just in the range that we have estimated above to be necessary to lower ( d / h)@xmath3 to the observed values
! additional spectroscopy of the emission from pdrs in the 45@xmath44 m region to confirm these detections and identifications would be of great value .
unfortunately , the irs instrument on spitzer only works longward of 5.2@xmath44 m .
detection of absorption features produced by deuterated pah material is unpromising , because absorption by the c - h stretching mode in aromatic hydrocarbons is quite weak even the 3.3@xmath44 m aromatic c - h stretch feature has not been seen in absorption .
detection of the 4.5@xmath118 c - d stretch in absorption does not appear to be possible .
the c - d bending modes will also be too weak to detect , and in any case would be confused with other pah modes , as discussed above .
the local interstellar medium appears to have ( d / h)@xmath119ppm ( see fig .
6 of wood et al .
2004 ) . if the total d / h @xmath120ppm
, then the local dust grains would be deuterated . for our above provisional value of ( h+d)@xmath121ppm ( including aliphatic hydrocarbons ) ,
we would estimate ( d / h)@xmath122 . the _ stardust _ mission ( brownlee et al .
1994 ) should have captured interstellar dust grains in its aerogel dust collectors , to be returned to earth in january 2006 .
if it has collected grain material that is representative of the interstellar grain population , this would be expected to have d / h @xmath123 if dust is responsible for the observed variations in ( d / h)@xmath3 ; the aromatic component of such grains , if isolated , might have an even higher d / h ratio .
laboratory measurements of d / h in the returned interstellar grain samples will test this hypothesis .
as noted above ( [ sec : observations ] ) , idps are sometimes seen to be strongly deuterated .
the largest d / h ratio seen thus far is @xmath124 ( messenger 2000 ) . if interstellar grains have d / h ratios as large as @xmath125 , as proposed here , why are these not seen in idps that are thought to originate in comets ?
first , while the median ( d / h)@xmath126ppm in figure [ fig : dvso ] , two of the 14 sightlines have ( d / h)@xmath127ppm . presumably this material was shocked relatively recently .
perhaps the solar system formed out of such matter 4.5@xmath1gyr ago .
second , while comets are thought to be made of relatively primitive material , it is possible that accretion into the solar nebula might involve a phase with sufficiently high temperatures and densities for significant exchange of h and d with the gas to take place , thereby lowering the d / h ratio in the grains .
the observed d / h ratios in interplanetary dust particles are compatible with the hypothesis that interstellar grains can reach ( d / h ) values as large as @xmath128 .
the lack of correlation between ( d / h)@xmath3 and ( o / h)@xmath3 in figure [ fig : dvso ] argues strongly against variations in astration as the explanation for the observed variation in ( d / h)@xmath3 .
2 . it appears possible for dust grains to remove d from the gas phase in amounts sufficient to account for the observed variations in ( d / h)@xmath3 from one sightline to another .
collisions of d with pah cations , and collisions of d@xmath89 with pahs , are expected to result in incorporation of d into the pah .
the rate of such collisions is sufficient to deplete d from the gas on time scales of @xmath02 myr in cool h i clouds .
4 . grain destruction should return d to the gas along with other elements found in refractory grains .
there should therefore be an observable correlation between ( d / h)@xmath3 and the gas - phase abundances of mg , si , ca , ti , fe , and other elements that exhibit strong depletions .
this prediction appears to be confirmed by a recent study finding a positive correlation between ( d / h)@xmath3 and ( fe / h)@xmath3 ( linsky et al .
carbonaceous interstellar material collected by _
stardust _ , if representative of grains from the local interstellar medium , should be strongly deuterated .
laboratory studies can test whether the proposed deuteration reactions have the required branching ratios .
i am grateful to the scientific organizing committee for invitating me to participate .
i thank j. black , g. hbrard and j.l .
linsky for helpful discussions , and r.h .
lupton for making available the sm plotting package .
this work was supported in part by nsf grant ast-9988126 and by grants from the w.m .
keck foundation and the monell foundation .
brownlee , d.e .
, tsou , p. , clark , b.c . , swanson , p.n . , & vellinga , j. 1994 , comet coma sample return plus interstellar dust , science and technical approach http://stardust.jpl.nasa.gov/science/details.html hbrard , g. , chayer , p. , dupuis , j. , moos , h.w .
, sonnentrucker , p. , tripp , t.m . , & williger , g.m .
2005 , in astrophysics in the far ultraviolet , ed .
g. sonneborn , h.w .
moos , & b .-
andersson ( this volume ) | the d / h ratio in interstellar gas varies on scales of a few hundred pc in the local milky way , with d / h values ranging from @xmath07@xmath1ppm to @xmath022@xmath1ppm .
the reduction in d / h relative to the primordial value of d / h@xmath2ppm is usually attributed to `` astration '' conversion of d into other elements by nuclear fusion in stars .
however , it is shown here that astration has difficulty accounting for the observations because the expected associated variations in o / h are not seen . the lower d / h values are instead likely due to `` depletion '' of the d onto dust grains .
polycyclic aromatic hydrocarbons ( pahs ) are a possible repository for the missing d , and it appears possible for gas - grain reactions to achieve extreme deuteration of carbonaceous grain material .
grain destruction will release d from the grains ; the gas phase abundance of d should therefore be positively correlated with the gas phase abundances of other elements that exhibit strong depletions , such as mg , si , ti , and fe , which also will be returned to the gas by grain destruction . |
among many numerical techniques for condensed matter physics , the monte carlo method is a popular choice when a long correlation length or a small excitation gap is anticipated .
apart from the negative sign difficulty , most of the shortcomings of the quantum monte carlo ( qmc ) method for spin systems have been removed or reduced . in particular , the qmc for finite temperature based on the path - integral representation has been improved considerably during the past decade .
the improvement was achieved mainly by the development of the loop - cluster algorithms@xcite and related methods .
for instance , the critical slowing down was tamed by the loop - cluster algorithms for a broad class of quantum spin systems@xcite .
it was shown@xcite that the typical size of the clusters coincides with the correlation length .
because of this property , an effective update of configurations is possible .
the other slowing down , due to small intervals for discretization of the imaginary time , was completely removed also by the loop - cluster qmc@xcite , which became even more evident by the extension to continuous imaginary time@xcite .
an efficient measurement of some of important off - diagonal quantities was made possible through the improved estimator@xcite .
one of the difficulties that have been left unsolved until recently was the freezing of configurations due to an external field competing with the exchange couplings . in the conventional framework of the loop - cluster algorithms ,
the field term does not affect the graph assignment probabilities .
it is taken into account only in the flipping probabilities of clusters . as a result
, the cluster size does not correspond to the physical correlation length any more .
it was demonstrated@xcite that this difficulty can be removed , in the case of the @xmath2 antiferromagnetic heisenberg model , by introducing two singular points at which the local conservation rule of particle number ( or magnetization ) is violated .
these singular points are called `` worms . ''
this extension of the configuration space makes it possible to take the external field into account in the hopping probability of worms .
another difficulty is large memory requirement due to the split - spin representation@xcite . when one uses the loop algorithm for a spin problem with large @xmath1 , it is customary to replace each spin operator by a sum of @xmath3 pauli matrices .
therefore , for larger @xmath1 , the algorithm consumes more memory . the stochastic series expansion ( sse)@xcite does not have this difficulty , since it works directly on the original spin - configuration space . the sse is based on the high - temperature series expansion of the partition function , rather than the path - integral formulation . however , it was pointed out@xcite that these two apparently different formulations are essentially equivalent in the limit of the infinite order expansion .
the apparent difference was due to the different updating method , rather than the formulations themselves .
quite recently@xcite , syljusen and sandvik introduced the notion of `` directed loops '' and proposed a framework that accommodates all of the above - mentioned ideas , i.e. , the loop updating , the worm updating , and the two formulations . their framework can be compared with kandel and domany s framework @xcite for the loop - cluster algorithms .
in fact , the mathematical formulation of the syljusen - sandvik ( ss ) scheme has a similar structure to kandel and domany s ( see appendix [ sec : framework ] ) , and the resulting algorithm coincides with the loop - cluster algorithm in some cases . in this sense
, the ss scheme can be viewed as a generalization of the kandel - domany framework .
an algorithm based on the ss scheme is characterized by the scattering probabilities of worms .
although the detailed balance condition imposes a set of equations to be satisfied by these probabilities , there are still a lot of degrees of freedom .
some of the solutions to these detailed balance equations lead to the single - cluster version of the conventional loop - cluster algorithm at zero magnetic field , which are known to be efficient .
however , all the solutions are not necessarily efficient or practical .
there are obviously many bad solutions in which the `` back - tracking '' probability @xcite or `` turning - back '' probability are dominating .
in addition , the straightforward solutions of the heat - bath type do not work either , as we see below in sec .
[ sec : performance ] .
although an efficient solution was discussed@xcite for @xmath2 @xmath0 models , the prescription was not given for general @xmath1 .
similar to the kandel - domany framework , the ss scheme does not give a concrete prescription for obtaining a good solution that leads to an efficient algorithm for specific models . a rule of thumb for obtaining a good solution is to minimize the turning - back probability .
however , even if the turning - back probability is fixed , we still have many degrees of freedom to play with , and the efficiency of the algorithm strongly depends on the choice of the worm - scattering probabilities , as we demonstrate in sec . [
sec : performance ] .
while this freedom can be quite useful for constructing new types of efficient algorithms , it makes finding a reasonable solution a nontrivial task . in this paper , we propose a natural extension of existing algorithms that determines a unique set of scattering probabilities of worms .
the resulting algorithm is within the ss scheme and expected to be efficient for a wide class of quantum spin systems .
the algorithm can be obtained by the coarse graining mapping applied to an algorithm in the split - spin representation . in sec .
[ sec : splitspin ] , we first discuss algorithms in the split - spin representation . in sec .
[ sec : coarsegraining ] , we show how we can obtain an algorithm in the original spin representation by coarse - graining the split - spin algorithm . based on this general prescription for the coarse - grained algorithms , we present in sec . [
sec : xxz ] a complete table of the worm scattering probability for the @xmath0 models with arbitrary @xmath1 . finally , in sec .
[ sec : performance ] , we compare the present algorithm with other algorithms such as the directed loop algorithm with the straightforward heat - bath solution .
it has been pointed out in previous papers@xcite that , in the limit of the infinite order expansion , a monte carlo algorithm based on the series expansion can be reformulated in terms of the language of the path - integral with continuous imaginary time , and vice versa . in what follows , therefore , we describe algorithms in this limit for the sake of simplicity , and use the path - integral language . the translation into the series - expansion language and its modification for a finite order expansion should be straightforward . a simulation based on the path - integral representation ( or the sse in the infinite order expansion limit )
can be visualized in a @xmath4 - dimensional space - time where @xmath5 is the real - space dimension . at each point in this space - time an integral variable is defined , and it takes on one of the @xmath6 values @xmath7 and @xmath1 . in the case of @xmath2
the variables are one - bit ( or ising ) variables .
accordingly , we consider world lines , which are trajectories of up - spins in this space - time . in the present paper ,
we use the term `` world - line configurations '' to refer to the spin configurations in the space - time for general @xmath1 , although they are not represented by simple lines for @xmath8 .
a monte carlo algorithm is nothing but a procedure by which the world - line configuration is updated so that the limiting probability distribution may coincide with the weight of the configuration , i.e. , the exponential of the action . in the ss scheme
, we deal with objects defined in the @xmath4- dimensional space - time ( fig.[fig : objects ] ) .
a vertical line of length @xmath9 represents a spin .
kink _ is a point at which the local spin configuration changes .
a particle ( or an up - spin ) jumps from one vertical line to another only at kinks . in models without particle number ( or magnetization )
conservation , a point at which a particle disappears or appears is also a kink .
every kink is located on a _
vertex_. vertices in the ss scheme play a role comparable to that of local graph elements in the conventional loop - cluster algorithms .
in particular , for models in which the magnetization conserves , vertices are represented by short horizontal lines , each connecting two or more neighboring vertical lines .
if a vertex connects two lines , we call such a vertex _ four legged _ since it joins four _ segments _ , where a segment is a part of a vertical line which is delimited by two vertices . a _
worm _ is a kink of a special kind located on a segment @xcite .
a worm can move continuously as the simulation proceeds , while locations of ordinary kinks and vertices are fixed until they are deleted . in the applications discussed in the present paper
, there are only zero or two worms at the same time in the whole system . for quantum spin systems , one cycle of update in the ss scheme
consists of the following operations on these objects : 1 ) assigning vertices to a given world - line configuration , 2 ) creating a pair of worms , 3 ) letting one of them move along segments and be scattered by vertices until it comes back to the other worm to be annihilated , and 4 ) deleting all the vertices with no kinks on them . in the rest of the present paper ,
we see these operations in more detail .
when a world - line configuration is given , we first assign vertices .
vertices are assigned to every part of the system probabilistically with a density that depends on the local world - line configuration .
in addition , all the kinks are regarded as vertices . after placing all vertices
, we choose a point on a segment at random and create a pair of worms there .
then , one of the worms starts moving .
as a worm passes a point in the space - time , it changes the local spin value there .
when the moving worm encounters a vertex , it may be scattered . the outgoing direction after the scattering is determined stochastically with certain predetermined scattering probabilities . when a moving worm meets the other worm , they annihilate .
therefore , what we have to specify in order to define an algorithm are the density of vertices and the scattering probability of worms .
the ss scheme imposes conditions on these .
the conditions are summarized in appendix [ sec : framework ] .
when spins in a given model are larger than s=1/2 , it is customary to replace the original spin operators by the sum of @xmath3 pauli spins@xcite , i.e. , @xmath10 where @xmath11 is an @xmath2 spin operator .
the partition function is expressed in terms of these @xmath12 degrees of freedom .
since the original phase space corresponds to the subspace in which @xmath13 holds , we have to project out all the states orthogonal to this subspace to obtain the correct partition function .
this can be done by inserting the projection operator @xmath14 : @xmath15 } ) = { \rm tr}_{\{\sigma_{i\mu}\ } } ( p e^{-\beta h[\{\tilde s_i\}]}).\ ] ] hereafter , the representation based on @xmath16 degrees of freedom is referred to as the `` original spin representation '' whereas that based on @xmath17 the `` split - spin representation . '' for many models , it is rather straightforward to obtain an algorithm in the split - spin representation . for example
, we can obtain an algorithm for the @xmath0 models with arbitrary magnitude of spins @xmath1 from that for the corresponding model with @xmath2 , simply by regarding the former as a superposition of many of the latter .
if we do so , we consider @xmath3 vertical lines for each original spin .
accordingly , a point in the space - time is specified by three numbers @xmath18 rather than two @xmath19 .
the coupling between two original spins @xmath16 and @xmath20 is transformed into @xmath21 couplings , each couples @xmath22 and @xmath23 : @xmath24 where @xmath25 , @xmath26 , and @xmath27 is the external field per bond ( e.g. , @xmath28 for the hypercubic lattice where @xmath29 is the external field per site ) . for the vertex assignment
, we apply the procedure for the directed loop algorithm for @xmath2@xcite to every one of @xmath21 combinations of @xmath12 spins .
to be more precise , the density for the vertex between two split spins is the same as that in the directed loop algorithm for @xmath2 with @xmath27 replaced by @xmath30 .
similarly , the worm - scattering probabilities for @xmath2 can be used for split spins with the same modification of @xmath27 .
for the projection operator , we do essentially the same as we usually do in the conventional loop algorithm for @xmath8@xcite . in the present framework , we represent it by special vertices , each located at @xmath31 connecting all the @xmath32 vertical lines on a site @xmath33 . to be specific ,
when a worm moves upwards along the vertical line @xmath34 and hits the point @xmath35 from below , it jumps to @xmath36 and go on upwards . the line to which the worm jumps , @xmath37 , is chosen with equal probability among those on which the local spin state is the same as the spin state right above the incoming worm .
namely , it is chosen among such @xmath38 s that @xmath39 may hold .
one of the drawbacks of the split - spin representation mentioned above is that it may require much more memory than the original spin representation . for example , in the loop algorithm for the @xmath40 models @xcite , we insert graphs that involve all pauli spins on two neighboring sites at the same imaginary time ( fig.[fig : sun ] ) . in the split - spin representation ,
insertion of a graph of this type creates @xmath41 new segments . since the memory requirement is roughly proportional to the number of segments , a loop algorithm for the @xmath40 model requires memory resources proportional to @xmath42 .
if we can construct an algorithm in the original spin representation , insertion of a graph would create only two new segments .
this leads to a memory requirement smaller by factor @xmath43 than that of the split - spin representation .
another drawback is the lack of portability of the code . in the split - spin representation , there are many kinds of vertices , in principle , depending on the number of legs .
therefore , we have to change the core part of the code to accommodate new kinds of graphs for each model unless we implement all possible sorts of graphs initially , which is impractical . on the other hand , in the coarse - grained representation ,
all the vertices are four - legged ( for models with two - body interactions ) and there are only four different types of scattering of worms .
therefore , the core part of codes for all models are the same except for the densities of vertices and the scattering probabilities of worms . for example , if we have a code based on the ss scheme for the @xmath40 model we can immediately obtain a code for the @xmath44 model simply by changing the arrays of the probability tables . in order to take full advantage of the ss scheme , therefore
, we have to construct probability tables for algorithms based on the original spin representation rather than the split - spin representation . for this purpose
, we consider a `` coarse - graining '' map and its stochastic inverse . the map is basically disregarding the detailed information of split spins .
the inverse of the map is to choose stochastically one of the split - spin configurations which are transformed by coarse graining into a given original - spin configuration . with these maps and an algorithm in the split - spin representation
, we can construct an algorithm in which we manipulate only original spin degrees of freedom .
to illustrate the idea in more detail , we again take the @xmath0 model with an arbitrary magnitude of spins @xmath1 .
we can define a coarse - graining map from a split - spin world - line configuration @xmath45 into an original spin world - line configuration @xmath46 as @xmath47 where @xmath48 is the value of @xmath49 at the imaginary time @xmath50 , whereas @xmath51 is the sum of them , i.e. , @xmath52 .
similarly , we can define a map for vertices .
since the only interaction is of second order in the spin operators , a vertex of the @xmath0 model has only four legs ( i.e. , connects only two lines ) .
if a vertex connects two lines @xmath34 and @xmath53 , we associate with it a vertex that connects two coarse - grained lines @xmath33 and @xmath54 .
of course , in the latter representation , the information concerning the indices @xmath55 and @xmath38 is missing . obviously , these mappings are many - to - one mappings
. however , we can define the inverse of this coarse - graining map . in the following ,
we adopt the `` particle '' picture in which an up - spin is regarded as a particle whereas a down - spin a hole . correspondingly , we use particle numbers @xmath56 and @xmath57 , instead of @xmath58 and @xmath48 , to specify local states of spins .
these are related to @xmath58 and @xmath48 by @xmath59 and @xmath60 .
the inverse mapping of a local state is rather simple .
suppose that a model is an @xmath61 model and a local spin state at the point of interest is @xmath62 in the coarse - grained representation .
there are two split - spin states that are mapped to this state , i.e. , @xmath63 and @xmath64 .
both configurations are chosen with the same probability ( i.e. , 1/2 ) since there is no reason to put any bias .
for general @xmath1 , all configurations that satisfy @xmath65 are chosen with the same probability , where @xmath66 is the local state on the coarse - grained line .
the inverse mapping of a vertex can be defined in a similar way . when two space - time points @xmath19 and @xmath67 are connected by a vertex ( with no kink on it ) in the coarse - grained representation , we can map it to a vertex connecting @xmath18 and @xmath68 with some probability .
when @xmath61 , there are four different ways of choosing @xmath55 and @xmath38 that are to be connected . however , the probability for taking one of them is not 1/4 in this case .
this is because the density of vertices depends on the spin states at their legs .
if , for example , the density of vertex between two particles is higher than that between a particle and a hole , a given coarse - grained vertex should be mapped to the former with larger probability than the latter . in other words ,
the probability for associating a coarse - grained vertex with a particular split - spin vertex should be proportional to the density of the latter .
therefore , the probability for associating a coarse - grained vertex between @xmath33 and @xmath54 with a split - spin vertex between @xmath34 and @xmath53 is given by @xmath69 here , @xmath70 is the density of vertices in the split - spin representation where the local spin values are @xmath71 and @xmath72 at the legs of vertices .
the coarse - graining map and its inverse can be used for obtaining the vertex density and the worm - scattering probability in the coarse - grained representation from those in the split - spin representation .
suppose an imaginary - time interval in which the state of two neighboring sites @xmath33 and @xmath54 are specified by @xmath73 and @xmath74 , respectively .
the site @xmath33 consists of @xmath73 particles and @xmath75 ( @xmath76 ) holes whereas the site @xmath54 consists of @xmath74 particles and @xmath77 holes .
then , there are @xmath78 , and @xmath79 possible combinations of @xmath80 , @xmath81 , @xmath82 , and @xmath83 pairs of @xmath12 spins , respectively .
since we assign a vertex with density @xmath84 for each @xmath80 pair , the total density of vertices connecting @xmath80 pairs is @xmath85 .
the densities for other combinations can be obtained in a similar fashion .
thus , the total density of vertices is @xmath86 for two neighboring segments with spin values @xmath73 and @xmath74 .
next , we consider the scattering probability of worms at a vertex with no kink on it .
suppose a spin - lowering worm hits the lower - left leg of the vertex from below in the coarse - grained picture . in order for this worm to be scattered
, the worm and one of the legs of the vertex must be mapped onto the same line by the inverse map ( fig.[fig : scattering ] ) .
there are two such cases : the case where the spin value is 1 [ fig.[fig : scattering](b ) ] on the legs on the other line and the case where it is 0 [ fig.[fig : scattering](c ) ] . in the first case
there are @xmath74 different choices of the line , whereas we have @xmath77 choices in the second case .
each individual choice in the first case has the weight @xmath84 whereas that in the second case has the weight @xmath87 .
therefore , the probability for choosing the first case is @xmath88 , whereas that for the second case is @xmath89 .
if we choose the first case , the probability with which the worm is scattered in the direction @xmath90 is @xmath91 the worm scattering probability for the second case is given similarly .
all in all , the probability of a spin - lowering worm being scattered into the direction specified by the directed graph @xmath90 ( @xmath92 ) becomes @xmath93 the probability for going through ( @xmath94 ) is simply equal to @xmath95 ( probabilities of the three proper scatterings ) .
the symbol @xmath96 denotes the state where the spin states on the upper - left , upper - right , lower - left , and lower - right legs are @xmath97 , and @xmath74 , respectively , and there is an incoming spin - raising ( @xmath98 ) or spin - lowering ( @xmath99 ) worm on the lower - left leg .
@xmath100 is the probability of the worm being scattered into the direction @xmath90 when the initial state of the vertex is @xmath101 in the split - spin representation .
this probability coincides with that in the @xmath2 case with the replacement @xmath102 .
the scattering probability of a spin - raising worm can be obtained in the same fashion .
the scattering probability at a vertex with a kink is simpler than that for a vertex with no kink on it , because in this case there is at most one type of vertex that may lead to a proper scattering ( diagonal , horizontal , or turning back ) .
for example , suppose a particle jumps from left to right at the kink at the imaginary time @xmath50 , and the spin - lowering worm is approaching the vertex on its lower - left leg .
the local state @xmath101 is given by @xmath103 then , the vertex s lower - left leg must be footed on positive segments [ @xmath104 because otherwise no particle can hop to the neighboring site there .
similarly , the lower - right , upper - left , and upper - right legs must be footed on negative , negative , and positive segments , respectively .
there are @xmath105 such choices of segments , and all the choices are equally probable . among them
, there are @xmath77 choices where the lower - left leg is footed on the segment where the worm is located . therefore , the probability of the worm being located on one of the legs of the vertex is @xmath106 .
then , the scattering probability for @xmath90 s corresponding to proper scatterings [ i.e. , @xmath107 is @xmath108 for spin - lowering worms .
probabilities for spin - raising worms can be obtained similarly .
thus , we have described the way we derive the density of vertices and the scattering probability of worms from an algorithm in the split - spin representation . although our description above may seem to give an actual procedure for coarse - graining mapping and its inverse
, we do not perform these mappings in real simulation .
they are only for deriving the density ( [ generaldensity ] ) and the probabilities ( [ generalscatteringprobability ] ) and ( [ generalscatteringprobabilityii ] ) . in the actual simulation
, we manipulate only coarse - grained variables . in order to complete the description of the algorithm
, we have to specify the procedure for the pair creation and annihilation of worms .
again , this can be done by the coarse - graining map and its inverse . in the split - spin representation , the pair creation of worms
is done simply by choosing a point @xmath18 in the system with a uniform probability distribution .
if there is a hole at the chosen point [ i.e. , @xmath109 , we create a pair of spin - raising worms there .
if there is a particle instead , we create spin - lowering worms .
when coarse grained , this procedure is mapped to choosing a point from the whole space - time with uniform probability distribution and creating a pair of spin - raising worms with probability @xmath110 or spin - lowering ones with probability @xmath111 , where @xmath73 is the spin state at the chosen point .
the moving worm travels according to the scattering process described above until it comes back to the original position @xmath18 where the other worm waits . when coarse grained , this `` coming - home '' event is mapped to an event in which a worm comes back to @xmath19 .
however , several other split - spin events are mapped to this same coarse - grained event .
namely , there are cases where the moving worm comes to the point corresponding to a different @xmath12 spin , i.e. , @xmath112 with @xmath113 .
worms in this case should not annihilate .
it has to be mapped , therefore , to a `` going - through '' event .
suppose that the worms are spin - lowering ones and that the local value of the coarse - grained spin is @xmath73 ( before the passage of the worm ) .
then , there are @xmath73 cases in total which are mapped to the same coarse - grained state . only one of them leads to the collision of two worms .
therefore , the probability of pair annihilation is @xmath114 and that for going through is @xmath115 . for the same reason
, the probability of annihilation should be @xmath116 if the worms are spin - raising ones .
the whole procedure of one monte carlo sweep ( mcs ) with the algorithm described in this section can be summarized as follows .
step 1 : : : place vertices at random with the density , @xmath117 , that depends on the local spin state , @xmath101 . set @xmath118 .
step 2 : : : increase @xmath119 by 1 .
choose a point in the whole space - time at random , and create two worms there , one is to move and the other is to stay .
for the moving worm , choose the initial direction of motion , upward or downward , with the probability 1/2 .
then choose its initial type , spin - lowering or spin - raising , with the probability @xmath111 or @xmath120 , respectively .
step 3 : : : let the moving worm go until it hits a vertex or comes back to the original position where the other worm stays .
if it hits a vertex before it comes back to the original position , go to step 4 .
otherwise , go to step 5 .
step 4 : : : choose the scattering direction @xmath90 with the probability @xmath121 , where @xmath101 is the local spin state at the vertex before the worm s arrival .
change the type of the worm as specified by @xmath90 . then , go back to step 3 .
step 5 : : : if the moving worm is a spin - lowering one , let it go through the original point with the probability @xmath115 and go to step 3
. otherwise let it go through with the probability @xmath122 and go to step 3 .
if the moving worm does not go through , let it annihilate with its partner and go to step 6 .
step 6 : : : if @xmath119 is smaller than @xmath123 , go back to step 2 . otherwise , erase all the vertices with no kink and go back to step 1 .
one monte carlo sweep is defined as a process between two successive resets of vertices ( i.e. , two successive passages of step 1 ) .
the number @xmath124 , the number of pair creations of worms during 1 mcs , can be an arbitrary positive integer .
we choose it so that every vertex may be visited by a worm once in average during 1 mcs .
since the @xmath0 models are of particular importance , we summarize the probability of vertices and the scattering probability of worms for the models in table [ tab : scatteringprobability ] . besides the coupling constants , the scattering probability depends upon the initial configuration of the scatterer ( i.e. , vertex ) , the type of the worm ( `` spin - raising '' or `` spin - lowering '' ) , and the incoming and outgoing direction . because of the mirror image symmetries with respect to the horizontal and vertical axes , scattering probabilities for
any two cases which can be transformed to each other by mirror image transformations should be the same .
therefore , without loss of generality , we assume that the incoming worm is located on the lower - left leg of the vertex . then , the initial states can be categorized into six classes , each specified by the spin states on all the legs and the kind of the incoming worm ( table [ tab : scatteringprobability ] ) .
there are only a few possible final states for each initial state .
those final states can be specified by the outgoing direction ( @xmath90 ) of the scattered worm ( fig.[fig : finalstates ] ) .
there are four such directions : turning - back , diagonal , horizontal , and straight , as indicated in the top row in fig.[fig : finalstates ] .
the probabilities for scattering in these directions are denoted by @xmath125 and @xmath126 respectively , where @xmath101 is the local state in the coarse - grained spin representation . in table
[ tab : scatteringprobability ] , we present the first three only . the probability for going straight
@xmath127 can be readily obtained by subtracting the other three from unity .
the scattering probability also depends upon the coupling constants , @xmath128 and @xmath129 . from the algorithmic point of view
, the whole parameter space is divided into six regions ( fig.[fig : phasediagram ] ) . within each one of the six regions ,
the scattering probability is a simple analytic function of the coupling constant , and it is continuous in the whole parameter space .
however , its first derivatives are discontinuous at the boundary between two adjacent regions . in the case of @xmath2 ,
the division is the same as that in the previous paper@xcite .
c|lcccccc @xmath101 & & region i & region ii & region iii & region iv & region v & region vi + @xmath130 & @xmath131 & a & b & b & b & a & c + & @xmath132 & 0 & @xmath133 & 0 & 0 & 0 & @xmath134 + @xmath135 & @xmath136 & @xmath137 & 0 & 0 & 0 & @xmath137 & @xmath138 + & @xmath139 & @xmath140 & @xmath141 & @xmath142 & 0 & 0 & 0 + & @xmath132 & 0 & @xmath143 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + @xmath147 & @xmath136 & @xmath148 & 0 & @xmath149 & @xmath141 & @xmath150 & @xmath151 + & @xmath139 & @xmath152 & @xmath153 & @xmath153 & @xmath153 & @xmath152 & 0 + & @xmath132 & 0 & 0 & 0 & 0 & 0 & 0 + @xmath154 & @xmath136 & @xmath155 & 0 & @xmath155 & @xmath156 & @xmath156 & @xmath156 + & @xmath139 & @xmath157 & @xmath156 & @xmath157 & 0 & 0 & 0 + & @xmath132 & 0 & 0 & 0 & 0 & 0 & 0 + @xmath158 & @xmath136 & @xmath159 & 0 & 0 & 0 & @xmath159 & @xmath160 + & @xmath139 & @xmath161 & @xmath160 & @xmath160 & @xmath160 & @xmath161 & 0 + @xmath162 & & + @xmath163 & & +
it is practically impossible to evaluate the efficiency of the algorithm for all possible combinations of coupling constants , the external field , the magnitude of spins , and the number of dimensions .
therefore , here we only show an example and make a few remarks concerning the efficiency of the algorithm described above . of particular interest
is the algorithm in region iii , because the primary motivation for developing the algorithms based on the ss scheme is to solves the freezing problem of the conventional loop algorithms in this region . for @xmath2 ,
good performance was demonstrated@xcite in the isotropic case @xmath164 for various values of @xmath29 .
most importantly , no severe freezing was observed at low temperature . in what follows ,
we show that the present algorithm solve this problem for an arbitrary @xmath1 .
several other directed loop algorithms ( algorithms 1 4 ) in the original spin representation are also examined for comparison with the present algorithm .
algorithms 1 3 are obtained by tuning solutions of the weight equation([weightequation ] ) and the detailed balance([detailedbalance ] ) so that the turning - back probabilities may be minimized .
all of these three algorithms have exactly the same turning - back probabilities .
algorithm 1 is characterized by the vanishing probability for the diagonal scattering when the field is zero [ i.e. , @xmath165 for all @xmath101 ] , whereas it is finite even at @xmath167 in algorithm 2 .
algorithm 3 is a mixture of algorithms 1 and 2 .
the details of these algorithms are given in appendix [ sec : examples ] .
algorithm 4 , on the other hand , is the heat - bath - type algorithm that can be obtained in the most straightforward way , although this is also a solution of eqs .
( [ weightequation ] ) and ( [ detailedbalance ] ) . in algorithm 4 ,
one of the four possible directions @xmath90 is chosen with the probability proportional to the weight of the final state . to be specific ,
@xmath168 where @xmath101 is the initial state of the vertex and @xmath169 is the final state of the vertex when the worm is scattered into the direction @xmath90 . in order to check the validity of these algorithms , we first performed simulations for a small one - dimensional system ( @xmath170 ) and compared the results with the exact solution for various set of parameters , @xmath129 , @xmath29 , and @xmath9 .
it turned out that all the algorithms yielded correct results with 1% or less of the statistical error .
the present algorithm and algorithm 1 yielded roughly the same magnitude of error whereas the other three yielded larger errors than the first two .
for a longer chain ( @xmath171 ) , 50 sets of simulations were performed using each algorithm where each set consists of 20000 creations and annihilations of worm pairs .
we can see the performances of five algorithms in fig.[fig : statisticalerror ] .
plotted in fig.[fig : statisticalerror ] is @xmath172 , where @xmath173 is the estimated statistical error of the squared staggered magnetization , @xmath174 is the system size and @xmath175 is the average number of the vertices visited by the worm during its lifetime .
since the scattering process is the most time - consuming part of the code , the total cpu time is roughly proportional to the total number of scattering events of worms , including the `` straight '' scatterings .
therefore , the cpu time is proportional to @xmath175 .
this is why the statistical error should be multiplied by @xmath176 in order to make the comparison fair . in fig.[fig : statisticalerror ] , we can clearly see that the present algorithm performs as well as the best algorithm among the the other four ( i.e. , algorithm 1 ) .
obviously , there is no exponential slowing down for the present algorithm and algorithm 1 , as was the case with syljusen and sandvik s algorithm for @xmath2 .
for a larger system ( @xmath171 ) with zero magnetic field , the results of the present algorithm and algorithm 1 agree with each other , while those of the other algorithms do not ( fig.[fig : systematicerror ] ) .
we consider that the correct result for @xmath171 is the one that is obtained by the present algorithm and algorithm 1 , and that the other algorithms fail to achieve equilibrium within the performed simulation .
we performed similar simulations for various values of @xmath29 ranging from @xmath177 to @xmath178 .
it turned out that the good algorithms ( the present and algorithm 1 ) always perform better than the bad ones ( algorithm 2 4 ) although the difference between them becomes smaller as the field is increased .
we can explain these facts in terms of the compatibility of the clusters with the order parameter .
one of the reasons why the conventional loop algorithm works well even in the vicinity of the critical point is the accordance between the typical cluster size and the correlation length .
this accordance is guaranteed by the two features of the algorithm : ( i ) independent flipping of clusters , and ( ii ) perfect ordering within each cluster .
although the staggered magnetization is not strictly the order parameter in one dimension , this criterion of good performance of loop - cluster algorithms still applies because finite but relatively long - range correlation exists even in one dimension .
it is easy to see that the present algorithm and algorithm 1 satisfy both conditions ( i ) and ( ii ) in the zero - field limit whereas the other algorithms do not satisfy condition ( ii ) regardless of the field .
this is the reason why the former two algorithms perform much better than the latter three in the weak field region .
therefore , we expect that the difference in the efficiency is even more pronounced near real phase transitions such as these in three - dimensional systems .
the results of the five algorithms , all based on the ss scheme , illustrate that it is not trivial in general to obtain the most efficient algorithm among many possible ones and also that the straightforward heat - bath algorithm is rather poor in some important cases .
it should be noted that the coarse - grained algorithm discussed in the present paper satisfies criterion ( ii ) mentioned above for an arbitrary @xmath0 model when the external field is vanishing .
therefore , we consider that the present algorithm performs well for a rather wide class of models in the weak magnetic field region . even for strong magnetic field , we consider that the present algorithm is at least as good as most of the other algorithms based on the ss scheme as we see above . for the performance of the present algorithm in the regions other than iii , we can not conclude much at the moment .
in the case of @xmath2 , the algorithm in region i is equivalent to the wolff version ( i.e. , the single - loop update version ) of the loop algorithm when the external field is vanishing .
a good performance of the algorithm in this region has been demonstrated in many applications@xcite , as well as for @xmath8 @xcite .
since the turning - back probability is vanishing in the whole region i , we expect good performance of the algorithm not only on the line @xmath167 but also in the whole region i. the algorithm in regions ii and vi , on the other hand , may not be very efficient because of the presence of the relatively large turning - back probability .
it is easy to see that , in the classical limit ( @xmath179 ) at zero external field , the turning - back probability dominates in regions ii and vi , leading to poor performance . for region
vi , this may not be very problematic because in this region ( at least in the classical ising limit ) the conventional loop - cluster algorithm works efficiently . for region
ii , it is not known whether this is a real problem or not .
the algorithm in region iv is reduced to a single - spin - flip metropolis algorithm in the limit of @xmath180 and @xmath181 .
the performance of the metropolis algorithm should be good in this limit , although the region is physically not very interesting .
we have proposed an algorithms based on the syljusen and sandvik scheme by introducing the split - spin representation and the coarse - graining procedure .
the algorithm is a natural extension of the directed loop algorithm for @xmath2 , in that the present algorithm coincides with it for @xmath2 .
in addition , the present algorithm is a natural extension of the conventional loop algorithm , because if the external field is vanishing , the present algorithm can be obtained through coarse graining the conventional split - spin loop algorithm .
compared with the algorithms in the split - spin representation , the coarse - grained algorithm requires much smaller memory , in general .
in particular , when the algorithm consists of vertices with more than four legs , as is generally the case with the loop algorithms for models with high order interaction terms , the memory can be reduced considerably .
the coarse - grained algorithm is also advantageous since the codes based on it can be very easily modified for other models ( we only need to change the arrays of the probability tables ) . in the case of the @xmath61 heisenberg antiferromagnet in one dimension ,
the algorithm s performance is almost the same as the best algorithm obtained by directly working on the original spin representation .
many other algorithms can also be obtained in the same way .
however , most of them , including the heat - bath algorithm , are much worse than the present one .
existence of algorithm 1 , a good direct solution to eqs .
( [ weightequation ] ) and ( [ detailedbalance ] ) , suggests the existence of similar solutions for an arbitrary @xmath1 .
we have not succeeded in finding a complete set of such solutions , although we believe that such solutions exist .
it would be an interesting future problem to find such solutions for various models .
the coarse - graining procedure presented in this paper applies not only to the @xmath0 spin systems but also to any model for which a directed loop algorithm can be constructed in the split - spin representation .
for example , on - site easy axis or easy plane anisotropy terms may be treated as the couplings between @xmath12 spins on the same site .
another example is the su(n ) model where the split - spin algorithm is known .
the authors are grateful to o. syljusen , a. w. sandvik , and m. troyer for enlightening comments .
the present work is supported by grants - in - aid for scientific research program ( no .
12740232 and no .
14540361 ) from monkasho , japan .
in general , a directed loop algorithm is characterized by the density of vertices @xmath117 and the scattering probability of worms @xmath121 .
the density @xmath182 is simply given by @xmath183 where the weight @xmath184 is defined as @xmath185 where @xmath186 when @xmath187 is true and @xmath188 otherwise .
the symbol @xmath189 stands for the integer by which the worm changes the spin value , e.g. , @xmath190 for a @xmath191 worm that lowers the spin value by 2 .
the variable @xmath192 is the only free parameter related to the vertex density . the scattering probability ,
on the other hand , has a lot of freedom .
the algorithm can be explained very clearly by introducing an extended weight @xmath193 that is related to the weight @xmath184 as @xmath194 here , we consider a scattering event in which the initial state of the vertex is @xmath101 and the out going direction of the worm is @xmath90 .
we denote the final state of this event as @xmath169 .
it should be noted that the ` state ' @xmath101 is directed in contrast to the state in the ordinary monte carlo simulation .
we consider the balance between an arbitrary sequence of scatterings and its reverse .
each sequence starts from pair creation of the worms and ends at pair annihilation , the detailed balance condition should be considered between @xmath101 and the reverse of @xmath169 .
therefore , the detailed balance condition is expressed as @xmath195 here , @xmath196 is the reverse of @xmath169 obtained by inverting the direction and changing the type of the worm in @xmath169 , whereas @xmath197 is the inverse of @xmath90 , obtained by inverting the direction of the arrow in @xmath90 .
it is worth mentioning that eqs .
( [ weightequation ] ) and ( [ detailedbalance ] ) are quite similar to the equations that appear in the general formulations of the loop - cluster algorithm@xcite .
the only difference is that here we use directed graphs and directed states whereas only nondirected graphs and states appear in the conventional loop - cluster algorithm .
it is easy to see that in the case of zero magnetic field the extended weight in the present scheme can be made independent of the directions of states and graphs , and all the equations discussed in this appendix coincide with those for conventional loop - cluster algorithms .
once we obtain any set of constant @xmath192 and positive @xmath193 s that satisfy eqs .
( [ weight ] ) , ( [ weightequation ] ) , and ( [ detailedbalance ] ) , we obtain a scattering probability @xmath121 of worms that satisfies the detailed balance condition as @xmath198
in tables [ tab : pmoneweightforafh ] and [ tab : pmtwoweightforafh ] , we show the weight and the extended weight that satisfy eqs .
( [ weight ] ) , ( [ weightequation ] ) , and ( [ detailedbalance ] ) for the @xmath61 antiferromagnetic heisenberg model . the vertex density and the worm - scattering probability can be obtained through eqs .
( [ density ] ) and ( [ scatteringprobability ] ) .
table [ tab : pmoneweightforafh ] shows the scattering probability for worms that change values of spin by 1 .
the weights contain two free parameters @xmath199 and @xmath200 .
note , however , that the turning - back probability does not depend on the choice of the free parameters .
algorithm 1 3 correspond to the following choices , respectively : @xmath201 in addition to the worms changing spin values by 1 , we can introduce worms that change spin values by an arbitrary amount . for @xmath61
, we can introduce @xmath202 worms . in the examples presented in sec .
[ sec : performance ] , we used both @xmath203 worms and @xmath202 ones .
when a pair of worms are created , the type of worm is chosen with equal probability from all possible ones .
for instance , when the initial spin state is @xmath204 at the point chosen for the pair creation , a @xmath205 or @xmath206 worm is possible .
each one of them is chosen with probability @xmath207 . the extended weight for @xmath202 worms
are listed in table [ tab : pmtwoweightforafh ] .
in fact , only the @xmath203 worms are necessary for making the algorithm ergodic and for satisfying the detailed balance . although it is likely that the @xmath202 worms are useful for improving the efficiency of the algorithm for more complicated models , we have not encountered such a case yet .
this wording is different from the original one by prokofv _
et al . _ , where the whole path connecting the two discontinuity points is called a worm . in the present paper
we call the discontinuities themselves worms .
therefore , the worms in the present paper should be called the worm heads " and the worm tails " in the original definition . however , since we did not feel necessity of referring to the `` body '' of the worm , we call both worm heads and worm tails worms in short . | recently , syljusen and sandvik proposed a new framework for constructing algorithms of quantum monte carlo simulation .
while it includes new classes of powerful algorithms , it is not straightforward to find an efficient algorithm for a given model . based on their framework , we propose an algorithm that is a natural extension of the conventional loop algorithm with the split - spin representation .
a complete table of the vertex density and the worm - scattering probability is presented for the general @xmath0 model of an arbitrary @xmath1 with a uniform magnetic field . |
since the inception of quantum computing @xcite , a large effort has been devoted to making quantum computers function in a noisy environment and securing them against imperfections in the physical setup itself .
the theoretical literature offers several ways to cope with such errors .
the leading idea is to encode quantum states into a larger system @xcite such that noise hits can be recognized and subsequently corrected . a quantum computation
can then be performed in a fault - tolerant way directly on these encoding systems @xcite , and nested levels of error - correction can make the computation error - free .
the last statement assumes that the initial error rate is not too high and that the error hits are not too correlated , e.g. occur locally . beyond these assumptions ,
quantum error - correction schemes require significant resources in terms of circuit size and experimental control @xcite .
other approaches use decoherence - free subspaces , employing for the quantum computation those parts of a system s hilbert space which are relatively unaffected by the noise @xcite .
the latter approach works only in more limited circumstances and requires detailed knowledge of the noise . here ,
we follow a different idea which tries to avoid the disadvantages just mentioned . rather than devoting large efforts to eliminating the errors at all costs ,
we accept them in the computation and try to design algorithms that eventually still find the desired result .
the spatial size of the quantum circuit should however not be enlarged much beyond the noiseless version of the algorithm ; one may e.g. allow only a number of extra qubits that remains bounded by a constant as the problem size becomes large @xcite ( whereas it seems reasonable to allow for an exponentially large noiseless classical memory ) .
furthermore , the algorithms should find the desired result under any level of background noise this level may actually be unknown to the algorithm , hence the term _ fault - ignorant computation_. the algorithm is , however , allowed to take more time the larger the actual noise gets . in this sense , we are trading spatial resources ( circuit size ) for temporal resources ( runtime ) .
still , when the actual noise level is low ( but constant in the problem size ) , we want our algorithms to produce the desired result faster than any classical algorithm even in a noiseless environment this indeed will be the case for the explicit algorithms we present .
we describe the fault - ignorant approach and these desiderata in more detail in section [ moreformaldescription ] . while under noise the resulting fault - ignorant algorithms may not give the full quantum speedup for large problem sizes
, they may still be useful for initial and medium - term realizations of quantum computers , in particular in the non - scalable low - qubit number regime which does not allow for full - blown quantum error correction schemes . in this paper
, we analyze the above _ fault - ignorant idea _ for the example of quantum search on a noisy quantum computer .
the unstructured search problem , i.e. the search for a marked item in an unordered list with oracle access , can be solved on a noiseless quantum computer by grover s algorithm quadratically faster than by any classical algorithm @xcite .
this quantum speedup is optimal @xcite .
good algorithms or optimality issues for the noisy case are however much less clear .
many studies in this direction investigate how ( different models of ) background noise or a faulty oracle affect specifically grover s algorithm
( e.g. @xcite ) , and it is often found that the quadratic speedup persists as long as the noise stays below a certain threshold depending on the database size
. only the paper @xcite by regev and schiff gives , for a specific model of a faulty oracle , a general lower bound on the number of oracle invocations necessary to find the marked item by _ any _ quantum algorithm ; ref .
@xcite proves the analogous result for any continuous - time implementation of oracle search .
a concrete algorithm that functions under a faulty oracle is briefly suggested in @xcite , advocating the avoidance of active error correction as well , as we do . in the quantum search part of this paper ,
we address the strengths and weaknesses of general oracular search algorithms under noise in more detail .
notably , we state all our upper and lower runtime bounds with explicit prefactors as e.g. in theorems [ thmsymmalgomemoryless][thmlowerboundwithmemory ] , which in particular allows a comparison with the performance of classical search algorithms already for `` small '' ( i.e. non - asymptotic ) database sizes @xmath0 . on the one hand
we give fault - ignorant algorithms , on the other we investigate their optimality .
more detailed comparisons to the works @xcite are made in section [ moregeneralmemoryalgos ] .
the hilbert space in our setup consists of a search space , possibly supplemented by an ancillary quantum system and both suffering from noise ; our constructive algorithms will actually not use ancillary systems , but we allow for them in the proofs of our lower runtime bounds , which strengthens these . additionally , the algorithms may have access to a noiseless classical memory , which is a technologically realistic assumption .
the noise itself is modelled as discrete hits of some noise channel that is to be applied between any two oracle invocations .
the problem of fault - ignorant quantum search is then as follows : devise a quantum algorithm that , except for some specified maximal failure probability @xmath1 , returns the marked item under any decoherence rate @xmath2 , using as few oracle calls as possible .
the fastest ( noiseless ) classical algorithm needs @xmath3 table lookups for this task , examining the database entries one by one .
we exhibit quantum algorithms which , under low but constant depolarizing level @xmath2 , require fewer oracle calls than this classical algorithm , see e.g. theorem [ thmunknownpwithoutexcluding ] .
the paper is organized as follows .
we give a more detailed description of the fault - ignorant idea in sec .
[ moreformaldescription ] , while referring to appendix [ sec : faultignorance ] for a general and precise mathematical definition of _ fault - ignorant algorithms_. those readers interested mainly in the problem of quantum search on a noisy quantum computer are directed to the remaining sections . in sec .
[ sec : memoryless ] we introduce quantum search in the presence of decoherence , and develop a fault - ignorant algorithm ( algorithm [ algounknownpnotexcluding ] ) that consists of `` quantum building blocks '' and uses the noiseless classical memory merely to store the marked item in case of a previous successful round .
a matching lower bound on the runtime of such algorithms is given by theorem [ thmsymmalgomemoryless ] .
we expand this analysis in sec .
[ sec : withmemory ] and allow for a noiseless classical memory that can store all previously falsified items , which enhances search efficiency ( algorithm [ algounknownpexcluding ] ) . in appendix
[ sec : noisemodels ] we discuss in more detail the noise models for which our results from the main text apply .
here we describe the main desiderata on fault - ignorant algorithms , with particular emphasis on oracular algorithms that are the main topic of the following sections . for a precise mathematical formalization of fault - ignorance , which also includes algorithms computing probabilistic functions as in sampling problems and computational algorithms such as factoring ,
we refer the reader to appendix [ sec : faultignorance ] ( in particular , definitions [ definitionnoisytask ] and [ definefaultignorantalgo ] there ) .
the tasks we consider consist in the computation of a classical output @xmath4 as a function of an oracle index @xmath5 and an input @xmath6 , which we assume is given via a specified encoding @xmath7 in the quantum register at the start of the algorithm .
the specification of the task contains already the available size of the quantum register , i.e. the hilbert space @xmath8 on which the @xmath7 act and which is assumed to be fixed throughout the computation ; in early implementations of quantum computers this size may be severely limited and is thus assumed to be part of the problem specification . besides the quantum register , we allow for a classical register that may serve several purposes : _ ( i ) _ to be used during the computation , _
( ii ) _ to store the output , _ ( iii ) _ to hold a binary flag indicating whether the output has already been written into the register , such that it can be read out by an outside agent without disturbing the ( quantum ) computation that may still be ongoing ( such an indication is necessary since the noise level and the algorithm runtime may not be known in advance , see below ) . for the task of oracle search among @xmath0 items
, we thus have @xmath9 and would like to produce the output @xmath10 , whereas the quantum register can be initialized in any fixed state @xmath11 as there is no other input to the task , i.e. @xmath12 .
we can allow for any quantum register @xmath13 ; @xmath14 would for example enable to perform grover s algorithm using the oracle eq .
( [ oracleaction ] ) ( see section [ setupnoiselessandmodels ] ) , but we may also allow for additional quantum registers such that e.g. @xmath15 .
we have not yet specified in what way fault - ignorant algorithms should behave with respect to noise . for this
we need two more ingredients .
the first is a family of noise channels , denoted by @xmath16 and acting only on the quantum register , that should model the effect of the noise per unit time ( see also appendix [ sec : noisemodels ] ) , and we think of the index @xmath17 $ ] as a noise strength parameter that is _ not _ known to the agent executing the algorithm .
classical registers are assumed noise - free .
for our results on noisy quantum seach in the following sections , we will for example choose the noise models in eqs .
( [ partiallydepolnoise ] ) or ( [ mostgeneralnoisemodel ] ) .
the second ingredient is the specification of the set @xmath18 of operations that an algorithm can perform per unit time .
@xmath18 may be any subset of all quantum channels acting jointly on the quantum and classical register .
actually , for oracular algorithms , each element @xmath19 also depends on the oracle index @xmath20 that is not directly accessible to the algorithm ; rather , when an algorithm `` calls '' the operation @xmath21 , then the quantum channel @xmath22 is executed .
again for the oracular search case below , we allow any @xmath23 that calls the oracle @xmath24 from eq .
( [ oracleaction ] ) once , possibly preceded and followed by any quantum channels @xmath25 acting on the quantum and classical registers ( and not depending on @xmath20 ) . for oracular problems ,
such a specification of the allowed operations per unit time is quite natural , but other choices are possible ; for computational problems one can for example impose locality restrictions on the operations per time step , such as in example [ binaryadditionexample ] in appendix [ sec : faultignorance ] .
the specification of what constitutes one time step will be implicit in eq .
( [ algorithmicsequenceia ] ) below , saying that the noise channel @xmath16 is to be applied once between any two operations from @xmath18 .
an _ algorithm _ is now simply a sequence @xmath26 of operations @xmath27 . algorithms may depend on an accuracy parameter @xmath28 , denoting the maximum probability with which we allow a wrong answer to be output .
crucially however , for a _ fault - ignorant algorithm _
@xmath29 , we _ do not _ allow the operations @xmath30 to depend on the above noise level @xmath2 , but still require that , for _ each _ noise level @xmath17 $ ] , after the execution of any number @xmath31 of time steps @xmath32 the classical output register holds the correct output @xmath33 , up to failure probability @xmath1 ( see also fig . [
fig : ignorance ] in appendix [ sec : faultignorance ] for illustration ) .
we call an algorithm @xmath29 _ fault - ignorant _ if @xmath34 can be chosen finite for all @xmath17 $ ] , and we will actually denote by @xmath34 the smallest time such that the previous requirement is satisfied for all @xmath31 ( see definition [ definefaultignorantalgo ] in appendix [ sec : faultignorance ] for an exact statement ) .
in other words , a fault - ignorant algorithm is ignorant of the noise level @xmath2 , but should nevertheless output the correct answer @xmath33 irrespective of the noise @xmath2 , within a time @xmath35 that may depend on the unknown noise level @xmath2 .
note that any computation which can be performed in a noiseless classical register of sufficient size alone constitutes a fault - ignorant algorithm , as we assume classical memory to be unaffected by the noise . as a further instructive example of a fault - ignorant algorithm ,
assume the following : _ ( a ) _
@xmath36 is a finite - step algorithm such that , after executing the sequence ( [ algorithmicsequenceia ] ) for @xmath37 time steps , there is for _ any _ noise level @xmath17 $ ] a nonzero success probability @xmath38 of obtaining the correct output @xmath33 in the classical register , _ ( b ) _ the set @xmath18 contains operations that allow to check whether a tentative result @xmath39 in the classical register is correct , _ ( c ) _ the input encoding @xmath7 is `` classical '' in the sense that there are operations in @xmath18 which can extract from @xmath7 the index @xmath40 onto a classical register , and conversely allow to prepare the quantum state @xmath7 given the classical value of @xmath40 ( requirement _ ( c ) _ is easy when @xmath41 , as in the search task ) .
if these three conditions hold , one can construct a fault - ignorant algorithm that solves the task for any fixed accuracy parameter @xmath42 , as follows : * repeatedly run the @xmath37-step sequence @xmath36 with noise interspersed as in ( [ algorithmicsequenceia ] ) on the initial state @xmath43 ( note that , after the value @xmath40 has been saved in a classical register at the start , this initial state can be reproduced before each iteration due to condition _ ( c ) _ ) , * after each iteration check for the correctness of the answer proposed in this iteration according to _ ( b ) _ , and store the correct result in an overall output register once it has been found . if the actual noise level is @xmath2 , then after @xmath44 iterations of @xmath36 , the failure probability equals @xmath45 , which will become smaller than any specified @xmath28 if only @xmath46 is large enough .
more precisely , since each round including verification consumes @xmath47 time steps , the composite check - and - repeat algorithm has a runtime @xmath48 for example , in searches using the quantum oracle ( [ oracleaction ] ) , one can satisfy the above conditions _ ( a)__(c ) _ by randomly selecting in each iteration an index @xmath49 ( see also the beginning of section [ firstsearchwithexclusionsubsection ] ) and then using the oracle @xmath24 once to check whether @xmath50 . in this example , @xmath51 and @xmath52 for all @xmath2 . for another similar example , see the beginning of section [ subsectionmemorylessunknown ] .
in fact , all the constructive algorithms we present in this paper will be variations of the above check - and - repeat algorithm , with possibly varying numbers @xmath53 of oracle uses in each round @xmath54 ( section [ subsectionmemorylessunknown ] , see fig .
[ fig : alg1 ] for illustration ) and possibly leaving out previously falsified items in future rounds ( section [ sec : withmemory ] , see also fig .
[ fig : alg2 ] ) . due to
this simple way of constructing fault - ignorant algorithms , the meaningful question , which we investigate in the following sections for noisy quantum search , is about the efficiency of fault - ignorant algorithms .
notice for example that the first algorithm proposed in the previous paragraph has a runtime of @xmath55 , proportional to the problem size @xmath0 and _ independent _ of @xmath17 $ ] .
however , one might hope that there exist fault - ignorant algorithms which for small actual noise level @xmath2 need fewer oracle calls , because at least when the noise is _ known _ to vanish ( i.e. @xmath56 ) then grover s algorithm solves the problem with at most roughly @xmath57 oracle calls ( see section [ setupnoiselessandmodels ] ) .
furthermore , both these runtimes diverge in the limit of perfect accuracy , i.e. @xmath58 , whereas the classical algorithm checking the @xmath0 items one after another needs only a finite number @xmath0 of steps for perfect accuracy .
in fact , in the following sections we will develop a fault - ignorant quantum search algorithm which , under the noise models ( [ mostgeneralnoisemodel ] ) and up to a constant overall factor , for any @xmath0 , @xmath2 , @xmath1 requires fewer oracle calls than the ones just mentioned ( see theorem [ thmknownpwithoutexcluding ] ) . when given a fault - ignorant algorithm solving one specific task ( e.g. one specified problem of size @xmath0 , of specified quantum register size @xmath59 , noise model @xmath60 , and accuracy goal @xmath1 ) , one can compare its runtime @xmath34 to the runtime of other algorithms that can be implemented ( e.g. to the classical search algorithm above , or to any algorithm that `` knows '' the noise level @xmath2 as one of its inputs , or to any algorithm that may use a quantum register of some larger size , etc . ) . on the basis of this comparison one
can then decide whether to consider this fault - ignorant algorithm useful w.r.t .
the competitor . instead of
solely the runtime , one may take into account also other factors in this comparison , such as the size of the quantum register used by either algorithm .
it does not seem possible to give general criteria for such a decision .
however , due to the prefactors given in the runtime bounds for our concrete algorithms , one can for example use our theorems [ thmunknownpwithoutexcluding ] and [ thmknownpwithoutexcluding ] to compare these fault - ignorant algorithms in such a way to other algorithms ( which we for example do below theorem [ thmunknownpwithoutexcluding ] and in section [ moregeneralmemoryalgos ] ) . instead of performing such a comparison for one specific task , one may consider a whole family of tasks ( e.g. one for each problem size @xmath61 and accuracy @xmath28 , possibly also allowing the size of the quantum register to vary independently of @xmath0 , etc . ) and
fault - ignorant algorithms solving them .
in this situation one can then investigate the _ scaling _ of the runtime @xmath34 with these parameters @xmath0 , @xmath1 , etc .
, as is usual in complexity theory , and investigate various tradeoffs , e.g. between the runtime and the size of the quantum register .
again , the concrete questions seem to depend highly on the tasks at hand . nevertheless , since the main feature of fault - ignorant algorithms is to find the correct answer _
without knowing _ the noise level @xmath2 in advance , we can introduce a distinguished notion of _ efficient fault - ignorant algorithm _ :
a family of fault - ignorant algorithms ( each solving a task from a given family of tasks ) is called _ efficient _ if there exists a constant @xmath62 such that for each fixed @xmath17 $ ] the runtime @xmath34 ( which depends on the member of the family ) is at most a factor of @xmath63 larger than the runtime of any algorithm that solves the same task while _ knowing _ the noise value @xmath2 as one of its inputs ( i.e. need _ not _ be fault - ignorant ) . in other words , we call a fault - ignorant algorithm _ efficient _ if knowing the actual noise level @xmath2 would shorten its runtime by at most an overall factor @xmath64 . our theorem [ thmlowerboundwithmemory ]
can thus be seen as a statement that algorithm [ algounknownpexcluding ] is efficient w.r.t . to the class of algorithms considered and , furthermore , an upper bound on
the constant @xmath63 is apparent together with theorem [ thmknownpwithoutexcluding ] .
independently of this efficiency notion , for low enough ( but unknown ) noise level @xmath2 , the runtime of algorithm [ algounknownpexcluding ] compares favorably with noiseless classical search ( see section [ moregeneralmemoryalgos ] ) .
the quantum search problem @xcite asks for an algorithm of short runtime to identify ( up to some small error probability @xmath1 ) one out of @xmath0 oracles , i.e. to return the index @xmath20 of the `` black box '' implementing the unitary transformation @xmath65 where @xmath9 and @xmath66 denotes addition modulo 2 . here , we assume the oracle @xmath20 to have been chosen uniformly at random ( often referred to as unstructured database ) .
it is customary to take the input of the oracle @xmath67 at each step to be of the product form @xmath68 so that the output is also of this form with the _ sign _ of the coefficient of @xmath69 flipped in the first tensor factor @xmath70 . in this case
one can neglect the second subsystem and concentrate on the effective unitary transformation @xmath71 as usual , we measure the runtime of an oracular algorithm by counting the number of queries ( oracle uses ) , which are viewed as the expensive or time - consuming operations , and disregard all the other quantum channels which are independent of the oracle .
grover @xcite found a solution to this problem which makes use of the equal superposition state @xmath72 this state reflects the initial lack of knowledge about the oracle and will be used frequently in the following . at the beginning of the algorithm the state @xmath73
is prepared and then the oracle black box @xmath24 and the unitary @xmath74 are applied alternately .
( in this description , and in our whole paper , we disregard any subsystem structure of @xmath75 ; if , for example , @xmath76 , then the `` inversion about the mean '' @xmath74 can be implemented efficiently in the number @xmath77 of qubits @xcite ) .
after @xmath37 applications of both operators , a von neumann measurement is performed in the standard basis . the outcome of this measurement will give the correct index @xmath20 of the oracle with probability @xcite @xmath78 independently of which oracle @xmath9 was implemented by the black box . in particular ,
if we choose @xmath79 , the success probability is @xmath80 .
alternatively , if we fix some small maximal error probability @xmath81 , with which the algorithm may return an incorrect oracle index , then we may stop after @xmath82 iterations .
in fact , it can be shown , for any @xmath83 , that grover s algorithm yields the highest probability of success which can be achieved by any quantum algorithm using the oracle @xmath37 times @xcite .
the above analysis is valid when the unitaries and measurements etc .
can be implemented perfectly and the quantum computer is not subject to noise . in more realistic settings , however , these idealizations have to be lifted , and some such extensions have been considered in the literature before , cf
. section [ sec : intro ] . in this work
, we consider the specific case where the quantum computer is continually affected by noise , e.g. coming from the environment .
our aim is not to approach this problem by implementing quantum error correction , which may be expensive in terms of the required control and size of the quantum computer .
rather , we aim to find ( optimal ) algorithms which succeed even under the influence of known , or ideally even _ unknown _ noise , in such a way that their runtime may depend on the noise level ; see sec . [
sec : intro ] . throughout this paper
the term `` noise '' will mean the application of a certain quantum channel to the state of the quantum register in discrete time steps .
this is supposed to model , within the discrete - time setting of oracle algorithms , that the quantum computer is continually affected by noise .
more precisely , we will impose that the noise channel has to act _ once between any two invocations of the oracle_. our paradigmatic example of noise will be the family of _ partial depolarizing _ channels @xmath84 acting on states @xmath85 on a @xmath86-dimensional hilbert space .
we can interpret these channels as acting on the system in a completely depolarizing way if a biased coin toss yields heads , which happens with probability @xmath17 $ ] , and otherwise leaving the quantum computer undisturbed .
intuitively speaking , the partially depolarizing noise ( [ partiallydepolnoise ] ) discards the whole quantum register with probability @xmath2 between any two successive oracle invocations .
in particular , quantum error correction can not be applied to this noise model ( cf .
appendix [ sec : noisemodels ] ) ; but so it serves to illustrate our idea of _ fault - ignorant computing _ , one of whose rationales actually is to avoid costly error correction procedures . in appendix
[ sec : noisemodels ] we will further introduce partially dephasing noise ( [ definepartialdephasing ] ) , which has an additional interpretation as modelling the transitioning from quantum to classical algorithms , and we relate the different noise models .
the lower bounds on the runtime of noisy quantum search algorithms which we prove ( theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] ) rely on partial depolarizing ( [ partiallydepolnoise ] ) , which is a very drastic and in some implementations quite pessimistic noise model , as it acts in a strongly correlated way across the whole quantum register ( somewhat similar to a noisy oracle @xcite , see also sec . [ moregeneralmemoryalgos ] ) . for initial implementations of quantum computing
this may in some cases indeed be a reasonable assumption , e.g. when @xmath2 denotes the occurrence probability of a noise event requiring the restarting of the whole quantum computer .
the noise strength @xmath2 could for example be related to the timescale of a drifting laser or of collective hits by external stray magnetic fields .
nevertheless , it would be desirable to prove similar lower runtime bounds for weaker noise models , in particular incorporating some kind of locality , but for now our bounds provide at least a ( pessimistic ) point of comparison .
on the other hand , the concrete algorithms we provide will function with the guaranteed upper runtime bounds given in theorems [ thmunknownpwithoutexcluding ] and [ thmknownpwithoutexcluding ] even under any more general noise of the form @xmath87 with @xmath21 being an arbitrary quantum channel .
i.e. it is necessary only that with probability @xmath88 at each step no fatal noise event occurs .
partial depolarizing and partial dephasing are special cases of this .
let us see how the probability of a successful measurement , i.e. returning the correct oracle index @xmath20 , looks when we include an application of the noise channel after each query . as a preparation ,
we first consider grover s algorithm under noise . introducing the channel @xmath89 , the final state can be written as @xmath90 , so that the success probability is then @xmath91 where we took an average over the @xmath0 possible oracles , since the search is unstructured and we assume equal a priori probabilities . in this paper
, we choose to consider the _ average success probability _ of algorithms , i.e. averaged over all possible oracles with equal weight ( see e.g. @xcite ) , as opposed to the minimal success probability , i.e. minimized over all oracles ( e.g. @xcite )
. both figures of merit agree for `` symmetric algorithms '' , e.g. for grover s algorithm @xcite and for the constructive algorithms we propose in this paper .
but our choice strengthens the lower bounds derived in the following on the required number of oracle invocations .
now , for all above noise models @xmath92 ( see eq .
( [ mostgeneralnoisemodel ] ) ) , the evolution @xmath93 can be written as a sum of @xmath94 histories . since each term
gives a nonnegative contribution to the sum , we can find a lower bound by keeping only the noise - free term @xmath95 in each factor : @xmath96 which is quite sharp unless @xmath97 , cf .
appendix [ sec : noisemodels ] ; compare this also to noiseless case , eq .
( [ noiselesssuccessprobab ] ) .
( the convention @xmath98 is understood throughout this paper . )
we are now interested in how well this simple algorithm , and other algorithms that we shall consider below ( i.e. not necessarily consisting of grover steps ) , perform .
that is , we would now like to derive _ upper _ bounds on the success probability @xmath99 depending on @xmath0 , @xmath37 , and @xmath2 .
as the starting point we use the implicit bound on @xmath99 derived by zalka @xcite for the average success probability @xmath99 after @xmath37 oracle calls : @xmath100 this bound has been established in @xcite for the following situation : we start from any pure state @xmath101 ; the oracle @xmath24 ( @xmath102 ) inverts the coefficients of the basis states within a subset @xmath103 where @xmath104 for @xmath105 ; we let arbitrary unitaries @xmath106 act after each oracle use ; in the end we perform a von neumann measurement in some basis , and our guess for @xmath20 is an arbitrary function of the measurement outcome .
the same bound ( [ zalkaimplicit ] ) holds thus when we start from any mixed state over @xmath107 ( @xmath108 ) , apply arbitrary channels between the oracle uses , and our guess for @xmath20 comes from measuring a povm @xmath109 .
this holds because mixed states , quantum channels and povm measurements can be dilated to pure states , unitary evolutions and von neumann measurements on a larger system @xcite , and the oracles ( [ oracleaction ] ) tensored by the identity on all other subsystems still invert coefficients of disjoint sets of basis states . even for the more general algorithms described above
, we can thus use inequality ( [ zalkaimplicit ] ) together with lemma [ lemmaxyineq ] ( see appendix [ app : technical ] ) , to prove the following _ explicit upper bound _ on the average success probability @xmath99 of any quantum algorithm using @xmath37 oracle calls ( with or without noise ) : @xmath110 this bound does not depend on the noise strength @xmath2 , and thus gives no further restrictions on _ noisy _ search compared to the noiseless case .
but it does enable us to prove a result on the limitations of algorithms employing a noisy quantum register @xmath111 for computation , with the oracle acting on the first subsystem .
note that the computational steps @xmath112 , @xmath113 in any algorithm covered by the following upper bounds on the success probability are nowhere assumed to be necessarily unital .
if they all were unital , then after the occurrence of a noise hit of partial depolarizing ( corresponding to the term @xmath114 in ( [ partiallydepolnoise ] ) ) the success probability of finding the marked item would be fixed at @xmath115 , whereas non - unital actions might try to correct a partial depolarizing error and increase the success probability ( we will comment on a particular non - unital error - detection - and - correction strategy below eq .
( [ eq : firstbound ] ) ) . and , actually , the following result holds even for noise channels @xmath116 that may be somewhat more general than the partial depolarizing given in eq .
( [ partiallydepolnoise ] ) : [ thm8kplus1overp ] let @xmath117 be the initial state of the algorithm , @xmath118 defined by @xmath119 be the quantum channels implementing the oracles on the first subsystem , and @xmath120 the noise channel that is to be applied between any two oracle calls , with @xmath17 $ ] and @xmath121 any quantum state .
let @xmath122 denote steps in the algorithm , such that the state after @xmath37 uses of the oracle @xmath123 is @xmath124 and let the final measurement be given by the povm @xmath125 .
then the average success probability of this algorithm is upper bounded as follows : @xmath126~\leq~\frac{1}{n}+\frac{8}{np^2}~,\label{moretrivialimplication}\ ] ] and @xmath127 introduce the following states : @xmath128 for @xmath129 , and @xmath130 . with these we can write @xmath131 and hence @xmath132~=~\sum_{i=1}^{k}p(1-p)^{i-1}\sum_x\operatorname{tr}[\sigma_i^xe_x]\,+\,(1-p)^k\sum_x\operatorname{tr}[\sigma_k^xe_x]~.\ ] ] as the `` computation '' of @xmath133 involved @xmath40 oracle calls , from ( [ psboundfromzalka ] ) we have @xmath134\le ( 2i+1)^2 $ ] , and thus @xmath135 which trivially leads to ( [ moretrivialimplication ] ) .
furthermore , we can use @xmath136 by the bernoulli inequality to obtain ( [ 8kplus1overp ] ) : @xmath137 algorithm 0 , performing @xmath37 steps of quantum search .
noise is acting between any two oracle invocations @xmath24 ( see eq .
( [ oracleaction ] ) ) , before and after which one may however apply arbitrary channels @xmath112 , @xmath138 .
the computation uses a quantum register of dimension @xmath0 , on which the oracle acts , and an ancillary ( quantum ) register @xmath139 , both noisy .
a measurement , described by some povm @xmath140 , is applied to the final state @xmath141 to guess the marked element @xmath20 . ] in the following text we shall address a sequence of operations as in ( [ statefrombasicbuildingblock ] ) acting on an initial state @xmath142 as * algorithm 0 * ( or @xmath143 ) , which is also depicted in fig .
[ fig : alg0 ] .
it is worthwhile to mention two ways of using bounds as in theorem [ thm8kplus1overp ] : 1 .
one can consider the noise level @xmath2 to be fixed and examine the scaling behaviour ( e.g. of the algorithm runtime ) with respect to the number @xmath0 of search items , or 2 .
one can consider @xmath2 to scale in some way with @xmath0 .
point ( ii ) provides a way to compare the results with other works ( e.g. @xcite ) where this kind of scaling was analyzed .
our results , however , apply to any values of @xmath0 and @xmath2 ( and later , of @xmath1 ) in the stated ranges , but we are implicitly often imagining the case @xmath144 , @xmath145 ( and @xmath146 ) , which is a sensible limit as explained in appendix [ sec : noisemodels ] .
theorem [ thm8kplus1overp ] has two important implications .
the first is that , for any fixed @xmath147 , the growth of the success probability is at most _ linear _ in the number @xmath37 of queries , as opposed to the quadratic growth in the noiseless case ( cf .
( [ noiselesssuccessprobab ] ) for @xmath148 ) .
this may come as a surprise as one might have guessed that the quadratic speedup of grover s algorithm may persist for small enough noise levels @xmath147
( i.e. fixed @xmath2 and @xmath144 ) . in other words ,
theorem [ thm8kplus1overp ] says that there exists _ no _ algorithm with success probability @xmath149 whenever partial depolarizing acts .
the inequality ( [ 8kplus1overp ] ) in particular implies that quantum error correction _ can not _ be done for partially depolarizing noise @xmath16 with @xmath147 .
the second implication is that the success probability is bounded by @xmath150 _ independently of @xmath37 _ , cf .
( [ moretrivialimplication ] ) . for growing @xmath144 , this goes to @xmath151 unless @xmath152 ; in general we can not reach a prescribed success probabilty @xmath153 with an algorithm as described in theorem [ thm8kplus1overp ] .
the straightforward solution to this problem is that we repeat the algorithm ( including the final measurement ) until the probability of failure in all the repetitions combined drops below @xmath1 .
this strategy is detailed in the following subsection and shows the basic structure for all further algorithms .
now we consider algorithms that consist of a number of repetitions ( rounds ) of the basic building block described in the previous subsection : repeatedly preparing states like ( [ statefrombasicbuildingblock ] ) , using the oracle @xmath24 a number of times , and trying to infer the oracle index @xmath20 by a measurement . in our concrete constructions
we shall , as in grover s algorithm @xcite , specify the channels @xmath138 in ( [ statefrombasicbuildingblock ] ) to perform unitary inversions ( @xmath74 ) about the mean , and @xmath112 to equal the identity channel ( whereas for our lower bounds @xmath112 , @xmath138 remain arbitrary ) .
note that these repetitions require a _ noiseless _ classical memory in order to reliably store the correct result @xmath20 after one of the rounds has been successful ( any noisy classical register may be part of the subsystem @xmath139 in theorem [ thm8kplus1overp ] and is subject to noise @xmath16 ) .
furthermore , in order to test whether the measurement after any one of the repetitions has returned the correct index @xmath20 and in order not to overwrite the correct result in subsequent rounds , one needs to perform a verification of the measurement outcome using _ one _ oracle call or classical table look - up immediately after each measurement . in this subsection
we will first develop and analyze an algorithm that finds , except for some specified failure probability @xmath42 , the marked element @xmath20 among @xmath0 elements , on a noisy quantum computer .
secondly , we will prove ( theorem [ thmsymmalgomemoryless ] ) that , up to a constant , the runtime of this algorithm is optimal among a certain class of algorithms , when the noise level @xmath2 is known in advance .
our algorithm will consist of @xmath154 rounds of the procedure described in theorem [ thm8kplus1overp ] , each time checking whether the concluding measurement gave the correct @xmath20 and , if so , storing the result in a noiseless classical register .
specifically , in each round we perform grover s algorithm for some number @xmath37 ( to be determined ) of steps this has been described at the beginning of subsection [ basicbuildingblocksubsection ] .
we now first give a motivating `` derivation '' of the algorithm .
clearly , the events `` the noise did not hit and the measurement was unsuccessful in the @xmath40th round '' and `` the noise did not hit and the measurement was unsuccessful in the @xmath155th round '' are independent for @xmath156 when we use grover iterations , and similarly for more than two rounds , since grover s algorithm is symmetric with respect to permutation of the oracles ( the probability for the complement of each such event is given by ( [ lowerboundprobs ] ) , which we will substitute below for @xmath157 ) .
this means that if we perform @xmath154 rounds with @xmath37 grover iterations in each round , then the probability of failure and the total number of queries will be @xmath158 and @xmath159 , respectively ( including the verification step after each of the @xmath154 measurements ) .
if we are to reach the target error bound @xmath1 in the least number of steps , we need to minimize @xmath159 subject to the condition @xmath160 .
the latter gives a lower bound on @xmath154 depending on @xmath0 , @xmath37 and @xmath2 , namely @xmath161 this means that for given values of @xmath0 and @xmath2 one needs to minimize @xmath162 i.e. the number of queries per decrease of error probability by a factor of @xmath163 , where the factor @xmath164 is included for normalization .
let the optimal @xmath37 be denoted by @xmath165 .
note that minimizing the rate function @xmath166 from ( [ definitionf ] ) originates from minimizing the number of oracle calls from independent rounds to get the failure probability below @xmath1 .
this is different than optimizing the _ expected _ number of oracle calls , cf .
@xcite for a comparison .
then our new algorithm consists of @xmath167 repetitions of rounds with @xmath165 grover steps each , measurement in the standard basis plus verification step and storing the result in the classical output register when successful .
the discussion in the following example will motivate an easy - to - compute quantity to be used instead of @xmath165 .
[ examplewherenequalsinfty ] if we are interested in large databases , we can simplify by taking @xmath144 , yielding , with ( [ lowerboundprobs ] ) instead of ( [ exactsuccessprobabfordepol ] ) ( giving virtually identical results ) : @xmath168 to find the optimal @xmath169 , we compare ( @xmath170 ) : @xmath171 this is a decreasing function of @xmath2 , so @xmath172 and @xmath173 intersect at @xmath174 which is appropriate for large @xmath37 , and we have @xmath175 whenever @xmath176 with the convention @xmath177 . by inverting the series expansion around @xmath178 one
can get an _
explicit _ approximate expression for @xmath179 : the inverse is @xmath180 , and thus @xmath181 is a good approximation especially for small @xmath2 . intuitively , this behaviour of the optimal length @xmath182 of a quantum round can be understood by noting that the quantum register remains undisturbed with reasonably high probability ( of order @xmath183 ) for time @xmath184 , whereas with a probability approaching @xmath185 exponentially the quantum register will be disturbed by noise when @xmath186 .
this is because the noise will hit at each step independently with probablity @xmath2 ( see eq .
( [ partiallydepolnoise ] ) ) .
thus , roughly @xmath184 grover steps provide an advantage ( see also , e.g. , @xcite ) . plugging @xmath187 into @xmath173 and considering small noise level @xmath188
, we see @xmath189 so @xmath190 vanishes linearly in @xmath2 around @xmath56 . the fact that this is non - zero and finite for @xmath147 , means that the normalization by @xmath115 in ( [ definitionf ] ) was the `` correct '' one .
this suggests that the number of steps @xmath191 necessary is proportional to @xmath192 , i.e. linear in @xmath0 for any fixed @xmath147 and @xmath28 , meaning that the quadratic speedup is lost under depolarizing noise ( [ partiallydepolnoise ] ) ; see theorem [ thmsymmalgomemoryless ] for a more rigorous and general lower bound on the runtime .
example [ examplewherenequalsinfty ] motivates a more rigorous analysis of the case of finite @xmath193 . in this finite case
, one has to be careful for small values of @xmath2 , since it is clearly not a good idea to do more than @xmath194 grover iterations in one round .
let us first assume that @xmath195 , and suppose we perform @xmath196 rounds with @xmath197 grover steps in each round , for some @xmath198 ( this ansatz is motivated by example [ examplewherenequalsinfty ] , and will below turn out to be good ) .
then the probability of failing in all rounds can be bounded as ( see [ lowerboundprobs ] ) @xmath199^{\lceil cnp^2\log\frac{1}{\varepsilon}\rceil}~ & \le~ \left[1-(1-p)^{\left\lfloor\frac{1}{p}\right\rfloor}\sin^2\left(\left(2\left\lfloor\frac{1}{p}\right\rfloor+1\right)\arcsin\frac{1}{\sqrt{n}}\right)\right]^{\lceil cnp^2\log\frac{1}{\varepsilon}\rceil } \nonumber \\ & \le~ \left[1-(1-p)^{\frac{1}{p}}\sin^2\left(\left(\frac{2}{p}-1\right)\arcsin\frac{1}{\sqrt{n}}\right)\right]^{\lceil cnp^2\log\frac{1}{\varepsilon}\rceil } \nonumber\\ & \le~\left[1-(1-p)^{\frac{1}{p}}(1-\delta)\left(\frac{2}{p}-1\right)^2\frac{1}{n}\right]^{\lceil cnp^2\log\frac{1}{\varepsilon}\rceil } \nonumber \\ & \le~\exp\left\{-c(1-p)^\frac{1}{p}(1-\delta)\left(\frac{2}{p}-1\right)^2p^2\log\frac{1}{\varepsilon}\right\}\nonumber\\ & = ~ \varepsilon^{c(1-\delta)(1-p)^{\frac{1}{p}}(2-p)^2}\label{boundfailingforknownp}\end{aligned}\ ] ] for some @xmath200 depending on @xmath0 and @xmath2 . as we want to guarantee a failing probability of at most @xmath1 , we need to choose @xmath201 such that the exponent in the final expression ( [ boundfailingforknownp ] ) is at least @xmath185 independently of @xmath2 for the following statements to be valid .
however , for large values of @xmath2 the exponent goes to @xmath151 , which is a consequence of vanishing ( [ lowerboundprobs ] ) for large @xmath2 and @xmath202 grover steps ; this can be avoided by introducing a cutoff @xmath203 into the specification of the algorithm , such that for @xmath204 we use @xmath51 iterations in each round , i.e. only perform verification steps on randomly chosen elements .
the failure probability in this range is @xmath205 numerically one finds that viable values for @xmath201 and @xmath206 in the specification of a concrete algorithm , i.e. such that both ( [ boundfailingforknownp ] ) and ( [ failureproblargep ] ) do not exceed @xmath1 , are , for example , @xmath207 and @xmath208 ( even when setting @xmath209 ) .
we have not optimized these constants , as our main interest for now is in the scaling for large @xmath0 , small @xmath1 , and all noise levels @xmath2 .
the number of oracle invocations during such an algorithm is upper bounded by @xmath210 where we used @xmath195 . for small noise levels @xmath211 ( where @xmath212 )
, one can do @xmath213 grover iterations in each of the @xmath154 rounds , i.e. before each measurement , which gives a success probability of at least roughly @xmath214 by the lower bound ( [ lowerboundprobs ] ) ; therefore , @xmath215 rounds are sufficient to get the failure probability below @xmath1 , putting an upper bound on the number of oracle calls similar to ( [ upperboundnumsteps ] ) with a term proportional to @xmath216 instead of the term @xmath217 . summing up
, our algorithm finds the marked element @xmath20 , except with probability @xmath218 $ ] , on a quantum computer with noise level @xmath17 $ ] using the oracle at most @xmath219 times , e.g. for ( non - optimal ) constants @xmath220 .
we omit here a formal statement of the algorithm , which should have become resonably clear from the description above , but will remedy this in subsection [ subsectionmemorylessunknown ] for the more general case of unknown noise level .
the algorithm just described performs a number @xmath154 of quantum rounds , each of identical length @xmath37 given by an ansatz that is based on example [ examplewherenequalsinfty ] .
a cleverer algorithm might try to detect when a noise event has happened and then immediately abort the present round in such a case and start a fresh round in order not to `` waste '' oracle uses .
one way to accomplish this would be to adjoin to the quantum register @xmath75 used above another quantum register @xmath221 of @xmath44 qubits that is initialized to @xmath222 at the beginning of each round and is left untouched by the above algorithm . in case of a partial depolarizing noise event , given by the term @xmath223 in ( [ partiallydepolnoise ] ) , the @xmath44-qubit register will then be reset to a computational basis state other than @xmath222 which can be detected by a projective measurement on this auxilliary system with probability @xmath224 .
thus , by expending a small number @xmath44 of extra qubits ( e.g. a number @xmath44 that is constant in the problem size @xmath0 , or chosen as @xmath225 ) one can detect a noise hit with high probability and abort the present round to gain a saving in the number of oracle calls compared to the algorithm outlined above . while this is a viable strategy in the noise model used above
, it is actually extremely dependent on the noise model ( [ partiallydepolnoise ] ) .
if , for example , the noise would replace the whole quantum state with probability @xmath2 by @xmath226 ( instead of @xmath227 ) , then the exact strategy would not work anymore .
in particular , any such strategy relies strongly on the fact that the noise is correlated across the whole quantum register .
while we allow such strongly correlated noise as a pessimistic assumption from the outset , in particular to prove our _ lower _ runtime bounds , one would probably not want the actual _ constructive _ algorithms to rely on this assumption .
in contrast , our algorithm developed below example [ examplewherenequalsinfty ] as well as the upper runtime bound ( [ eq : firstbound ] ) work for any noise model @xmath228 with @xmath21 _ any _ quantum channel @xmath21 ( see app . [
sec : noisemodels ] ) .
this is because we only use eq .
( [ lowerboundprobs ] ) , which merely relies on the fact that with probability @xmath229 the quantum register remains undisturbed . furthermore ,
even when relying on an exactly known noise model as e.g. in eq .
( [ partiallydepolnoise ] ) , one would at most save a constant factor of order @xmath185 by the error - detection - strategy compared to the runtime ( [ eq : firstbound ] ) of the algorithm outlined above .
this is due to the exponential first factor in ( [ lowerboundprobs ] ) , which implies that only with small probability @xmath230 will the noise hit occur before executing @xmath231 steps in one round ( where @xmath232 here , such that there would be a saving ) .
we would like to point out that there are at least two different interpretations of runtime complexity results like eq .
( [ eq : firstbound ] ) .
firstly , one can run the algorithm indefinitely long ( i.e. without any a priori bound on the number of rounds ) until the marked element is found .
then we can guarantee that the algorithm gives the correct result with probability @xmath185 , and the number of oracle calls required is at most @xmath233 except with probability @xmath1 .
alternatively , one can decide in advance to use the oracle @xmath233 times before terminating the above algorithm , and after any successful measurement store the result in a classical memory ; then , in the end , the marked element will have been found with probability at least @xmath153 . with the runtime bound ( [ eq : firstbound ] ) at hand , one can look at the case where @xmath2 is fixed and independent of @xmath0 .
then we see that , unless @xmath56 , the leading term is @xmath234 , i.e. proportional to @xmath0 . on the other hand ,
if one supposes that @xmath2 scales decreasingly with @xmath0 , the other terms may dominate .
in particular , if @xmath235 , the leading term is @xmath236 . next we show that our algorithm presented above is essentially optimal within a certain class of algorithms ( a wider class of algorithms will be considered in section [ sec : withmemory ] ) .
namely , we assume that the algorithms employ a quantum register @xmath111 ( as above in theorem [ thm8kplus1overp ] ) , consist of several `` rounds '' where in each round we prepare some state , apply arbitrary channels and use the oracle an arbitrary number of times ( possibly different for each round , but applying the noise channel between any two consecutive queries ) , do any measurement yielding an element of @xmath237 , and verify the result with one oracle use , writing it into a ( noiseless ) classical register reserved for storing the output if correct .
crucially , we assume that the events of success in each round are independent of each other .
this assumption is valid in particular if the noise is symmetric ( under permutations of the basis vectors @xmath69 of @xmath75 , which partial depolarizing from eq .
( [ partiallydepolnoise ] ) satisfies ) and if the steps between measurements are grover iterations ( as for example in our algorithm above ) .
[ thmsymmalgomemoryless ] consider a sequence of algorithms , one for each size of the search space @xmath238 , satisfying the assumptions just stated and subject to partial depolarizing noise ( eq . ( [ partiallydepolnoise ] ) ) .
if the success probabilities are @xmath239 , then , asymptotically , the number of queries @xmath240 in the @xmath0th algorithm is lower bounded by the level @xmath17 $ ] of depolarizing noise : @xmath241 more precisely , for any @xmath242 $ ] and any finite @xmath243 , the number of queries @xmath240 satisfies : @xmath244^{-1}~.\ ] ] suppose that the @xmath0th algorithm consists of @xmath245 rounds with @xmath246 queries in each round ( abbreviating @xmath247 ) , with failure probability @xmath248 in the @xmath40th round .
then @xmath249 , and by the independence condition @xmath250 which is a weighted average of expressions of type @xmath251 .
lower - bounding this expression thus automatically lower - bounds ( [ weightedaverageinproof ] ) , and therefore it is enough to consider the @xmath252 case only .
since , by ( [ moretrivialimplication ] ) , @xmath253 as @xmath144 ( for any @xmath147 ) , one has @xmath254 for some positive sequence @xmath255 . using that , by ( [ 8kplus1overp ] ) ,
also @xmath256 we get @xmath257 the finite-@xmath0 bound follows from @xmath258 and setting @xmath259 equal to the quantity inside the square brackets in ( [ finitenbound ] ) .
theorem [ thmsymmalgomemoryless ] shows that , under any nonzero noise @xmath147 ( and @xmath28 ) , our algorithm from above has asymptotically optimal runtime , up to a constant factor : in our algorithm , @xmath260 was chosen independent of @xmath0 and ( [ eq : firstbound ] ) shows that asymptotically @xmath261 , which matches ( [ eq : asymptotics ] ) up to a factor of @xmath262 . in the noiseless case @xmath56 , our algorithm reduces to repeated grover searches , whose optimality for @xmath56 was shown in @xcite .
one other implication is noteworthy : on the one hand , theorem [ thmsymmalgomemoryless ] says that , at fixed positive noise @xmath147 and asymptotically for @xmath144 , the number of oracle queries @xmath263 _ has _ to grow at least linearly in @xmath0 , so that the quadratic speed of noiseless grover search is lost ( at least for the class of algorithms considered above , and for the depolarizing noise model , eq .
[ partiallydepolnoise ] ) . on the other hand , however , the prefactor in this linear growth is @xmath264 , which is actually achieved by the explicit algorithm above , see eq .
( [ eq : firstbound ] ) ; thus , for small enough noise @xmath2 , the number of oracle calls to solve the search task by a quantum algorithm is _ much less _ than the minimal number @xmath265 of oracle calls required by any classical algorithm , even in a noiseless environment .
in the following subsection , we will extend the above algorithm so that it works in a noisy environment even when its noise level @xmath2 is _ not _ known beforehand ( algorithm [ algounknownpnotexcluding ] and theorem [ thmunknownpwithoutexcluding ] ) .
we are now ready to turn to the _ `` fault - ignorant '' setting _ the algorithm should be ignorant of the actual noise level under which the quantum computer operates . more precisely , the goal is to find an algorithm for which not the ability to give the correct result depends on the level of noise , but rather only its runtime may depend on the actual noise level .
actually , the algorithm described in the previous subsection has this property for any fixed number @xmath37 of oracle calls per round ; however , the runtime can then become large unless @xmath266 .
for example , if we choose @xmath267 in order to get a quadratic speedup for @xmath56 , then for @xmath268 the number of oracle calls grows as fast as @xmath269 , which is clearly unsatisfactory . in order to overcome this problem
we allow the number of queries to change from round to round .
thus , for each @xmath0 and @xmath1 , we need to find a sequence @xmath270 , where @xmath271 denotes the number of grover iterations performed in the @xmath40th round . again , for our constructive algorithm , we employ the usual grover iterations ; and again , theorem [ thmsymmalgomemoryless ] will later show that this algorithm is nearly optimal .
one idea can be as follows . in the first round
, we do a grover search with @xmath272 oracle calls ( for the definition of @xmath182 see below ( [ definitionf ] ) ) . for @xmath273 this is enough to get the error probability below @xmath1 as long as @xmath56 ; the set @xmath274\,\big|\,p_s(n , k_{opt}(n,0),p)>1-\varepsilon\right\}$ ] is open and therefore @xmath275\,\big|\,1-p_s(n , k_1,p)\geq\varepsilon\right\}~\ ] ] exists and is larger than @xmath151 ( if the set is empty , e.g. when @xmath276 , we set the infimum to @xmath277 ) .
suppose that the measurement after the first round fails to find the marked element @xmath20 .
there are now two possibilities : either the actual noise level was below @xmath278 , in which case the probability of this failure was at most @xmath1 ( i.e. as required ) ; or the actual noise level exceeded @xmath278 , in which case the function @xmath182 gives an upper bound on the optimal number of grover iterations to perform in the next round , so we set @xmath279 .
we then proceed similarly by iteratively setting @xmath280\,\big|\,\prod_{j=1}^{i-1}\left(1-p_s(n , k_j , p)\right)\ge\varepsilon\right\}$ ] and @xmath281 , giving the number of grover iterations to be performed in the @xmath40th round .
the fault - ignorant algorithm [ algounknownpnotexcluding ] searches the marked element @xmath20 in consecutive rounds of @xmath282 grover steps plus one verification step each .
each round starts by preparing the equal superposition state @xmath283 , which is follwed by grover steps ( a special case of algorithm @xmath151 , see fig .
[ fig : alg0 ] ) , and a measurement in the computational basis ; finally , the output is checked against the oracle ( possibly by list look - up ) and , in case of success , stored ( by @xmath284 ) in a noiseless classical register with @xmath0 states , so that the result is ready for readout by an external agent ( the algorithm may however continue ) .
no ancillary system @xmath139 is used by the algorithm ( @xmath285 ) . ]
the sequences @xmath286 obtained this way are difficult to analyze , but by examining the results of numerical simulations for various values of @xmath0 and @xmath1 one can get an idea about their behaviour .
this turns out to be enough to get an approximation which still achieves the asymptotically optimal performance , up to a multiplicative factor in the runtime ( see below ) . specifically , we arrive at the following algorithm ( see also fig .
[ fig : alg1 ] ) for _ fault - ignorant quantum search _ : [ algounknownpnotexcluding ] for suitably chosen @xmath198 , define @xmath287 repeat the following steps for @xmath288 : 1 .
prepare the equal superposition state @xmath289 on a quantum register @xmath75 , 2 .
perform @xmath290 grover iterations , 3 .
measure in the standard basis , verify the result using one oracle invocation ; if correct then store in classical output register .
the following theorem proves that algorithm [ algounknownpnotexcluding ] is fault - ignorant , i.e. finds the marked element independently of the actual noise level ( in particular , the algorithm does not need @xmath2 as an input ) , and gives a bound on its runtime which , however , depends on the actual noise level .
[ thmunknownpwithoutexcluding ] let @xmath17 $ ] be the actual noise level ( i.e. noise @xmath291 with any quantum channel @xmath21 ) acting on the quantum register when executing algorithm [ algounknownpnotexcluding ] , and let @xmath292 and @xmath293 $ ] .
then algorithm [ algounknownpnotexcluding ] , with @xmath294 , finds the marked element after at most @xmath295 oracle queries except with probability at most @xmath1 .
if one wants algorithm [ algounknownpnotexcluding ] to be fault - ignorant only for noise levels @xmath296 $ ] , and if one presupposes reasonable values @xmath297 and @xmath298 $ ] , then one can guarantee the constant prefactor to be @xmath299 instead of @xmath300 as in ( [ stepsinfirstfaulttolerantalgo ] ) , when using @xmath301 ; cf .
appendix [ mainthm ] .
as the proof of theorem [ thmunknownpwithoutexcluding ] is rather technical , we present here only a sketch ; for details , see appendix [ mainthm ] .
as the noise acts symmetrically with respect to the different oracles and since the grover steps of algorithm [ algounknownpnotexcluding ] are symmetric as well , the success events for different rounds @xmath302 are independent , so that we will be able to upper bound the failure probability after round @xmath303 : @xmath304 when the sum @xmath305 in the last expression is greater than @xmath306 then we can guarantee the failure probability to be at most @xmath1 , as desired . to get the statement about the number of oracle calls
, we upper bound it by @xmath307 the proof of theorem [ thmunknownpwithoutexcluding ] now consists in showing that there exists a number @xmath303 ( of rounds ) such that the failure probability ( [ failprobabinfirstlowerbound ] ) is at most @xmath1 , while the number of oracle calls ( [ numoforaclecallsfirstlowerbound ] ) is at most ( [ stepsinfirstfaulttolerantalgo ] ) . similar to our analysis leading up to ( [ eq : firstbound ] ) , this argument will be split into three different cases : for @xmath308 the first few rounds ( @xmath288 ) are the important ones ; for @xmath309 we take into account only the rounds with @xmath310 ( i.e. large @xmath302 ) ; and for @xmath311 our proof relies on an intermediate regime of @xmath302 .
details are given in appendix [ mainthm ] .
theorem [ thmsymmalgomemoryless ] actually shows that the runtime of algorithm [ algounknownpnotexcluding ] ( which we just proved to be at most eq .
( [ stepsinfirstfaulttolerantalgo ] ) ) is _ optimal up to a constant _ : to see this , note that algorithm [ algounknownpnotexcluding ] is contained in the class of algorithms to which the bound from theorem [ thmsymmalgomemoryless ] on the number of oracle queries @xmath312 applies . for any fixed noise level @xmath147 and up to a constant factor
, this equals the upper bound ( [ stepsinfirstfaulttolerantalgo ] ) on the number of queries needed by algorithm [ algounknownpnotexcluding ] .
in particular , even if one does _ not _ know the actual noise level in advance , one only loses a constant factor in the number of queries , compared to the runtime in case of known @xmath2 ( given in eq .
( [ eq : firstbound ] ) ) .
we re - emphasize here the last point from subsection [ subsectrepeatbuildingblockknownp ] , that for small enough but constant noise levels @xmath2 and in the limit @xmath144 , the quantum algorithm [ algounknownpnotexcluding ] needs _ fewer _ oracle calls than even the best classical algorithm in a noiseless environment .
as the runtime depends on an unknown parameter , it is necessary to have the ability to stop the algorithm as soon as the result is found .
theorem [ thmunknownpwithoutexcluding ] then states a `` probabilistic bound '' on the number of oracle uses up to the point when the element is found ; this bound is probabilistic in the sense that in a fraction of at most @xmath1 of all runs of algorithm [ algounknownpnotexcluding ] , the actual runtime may exceed this bound . when considering more general algorithms , namely for which the events of failure in different rounds are not independent , the derivation of the lower bound on the necessary number of queries from theorem [ thmsymmalgomemoryless ] is no longer valid .
this dependency can arise either from asymmetric noise or from an asymmetry in the algorithm itself .
indeed , it is useful to consider such `` asymmetric algorithms '' : already classically one can find the marked element using @xmath313 queries , except with error probability @xmath1 , by simply testing a subset of @xmath313 elements using one oracle call each .
this feature of not considering previously falsified items again is absent from algorithm [ algounknownpnotexcluding ] whose runtime may therefore exceed that of classical search , through the factor @xmath306 in theorem [ thmsymmalgomemoryless ] .
the algorithms considered in section [ sec : withmemory ] will make use of this asymmetry , which can also be conceived of as conditioning the actions in future rounds on previous measurement outcomes that are being stored in a classical memory .
this will be done by incorporating a noiseless classical memory which we will allow the algorithms to use in a limited way , namely by excluding oracle indices that have been falsified in previous rounds .
classical search algorithms can find the marked element with maximal failure probability @xmath1 using at most @xmath313 steps , by excluding falsified oracle indices .
here we aim to achieve an upper bound of @xmath0 on the runtime independently of @xmath1 and of @xmath2 for our quantum algorithms as well , whereas in section [ sec : memoryless ] we have only presented algorithms whose runtime may exceed @xmath0 parametrically due to the factor @xmath306 , e.g. in ( [ stepsinfirstfaulttolerantalgo ] ) . on a quantum computer a random choice may be implemented by preparing the equal superposition state @xmath73 over a subset of basis states followed by a measurement in that basis .
this in turn can be viewed as a grover search with zero iterations ( cf .
subsection [ setupnoiselessandmodels ] ) .
this leads to the idea of replacing the uniform random choices by proper grover searches ( each including several grover steps plus a concluding measurement ) over decreasing subsets , i.e. @xmath314 after round @xmath315 . for this
, the classical noiseless memory that previously stored only the correct search outcome , will be exanded by a register of @xmath316 states ( @xmath0 bits ) to mark the previously excluded items
. we shall now develop and sketch a search algorithm based on this idea of excluding previously tested elements ; the following procedure is applicable if the noise level @xmath2 is known beforehand .
if one fixes the number @xmath0 of database entries , the noise parameter @xmath2 and the target error bound @xmath1 , then the question is how to choose the number of iterations in each round in order to consume the least number of queries .
suppose that in the @xmath40th round we perform @xmath317 queries and we do @xmath154 rounds in total .
then the number of queries is @xmath318 , while the probability of not finding the marked element is at most @xmath319 ; thus , the minimal number of queries for which one can guarantee success ( up to failure probability @xmath1 ) in the general noise model is @xmath320 e.g. letting @xmath321 equal the lower bound in ( [ lowerboundprobs ] ) ( alternatively , ( [ exactsuccessprobabfordepol ] ) ) . for given @xmath322 , the minimum ( [ mindynprogramming ] ) and the corresponding sequence @xmath323 of grover steps can be found by dynamic programming .
clearly , @xmath313 is an upper bound on the number of oracle calls and for any fixed @xmath324 we can bound this as @xmath325 for all @xmath326 .
similarly , by ( [ eq : firstbound ] ) or theorem [ thmunknownpwithoutexcluding ] , for @xmath327 there is also an upper bound of the form @xmath328 , since @xmath329 where @xmath201 is determined by @xmath330 .
hence , an upper bound on the runtime of the form @xmath331 holds for the complement of any neighbourhood of @xmath332 ^ 2 $ ] , at least asymptotically for @xmath144 . in following subsection
we simplify the above algorithm , based on typical behaviour of the sequences @xmath286 found in numerical experiments .
in this subsection we present a more explicit algorithm to solve the search problem in the _ fault - ignorant setting _ , i.e. an algorithm which can be specified and works even for _ unknown _ noise level @xmath2 , using the exclusion described above to obtain faster runtime ( cf .
also fig .
[ fig : alg2 ] ) : [ algounknownpexcluding ] for suitably chosen @xmath198 , define @xmath333 and @xmath334 repeat the following steps for @xmath288 : 1 .
prepare the equal superposition state @xmath335 over the set @xmath336 , 2 .
perform @xmath53 grover iterations ( with @xmath337 as reflection ) , where @xmath338 3 .
measure in the standard basis , verify the result @xmath339 using one oracle invocation , store if correct , 4 .
let @xmath340 .
algorithm [ algounknownpexcluding ] uses exclusion in searching for an element in consecutive rounds of @xmath282 grover steps each , supplemented by one verification query .
each round starts by preparing the equal superposition state @xmath335 of the previously not excluded elements , noted in the classical memory @xmath341 $ ] , and is concluded by a measurement in the computational basis .
the output is then verified against the oracle ( list look - up ) and stored in the classical noiseless memory @xmath342 $ ] if the element is found and marked in the memory @xmath341 $ ] if the round was unsuccessful ( @xmath343 ) . ]
similarly as theorem [ thmunknownpwithoutexcluding ] for algorithm [ algounknownpnotexcluding ] , the following theorem proves fault - ignorance of algorithm [ algounknownpexcluding ] and provides a bound on its runtime : [ thmknownpwithoutexcluding ] let @xmath17 $ ] be the actual noise level ( i.e. noise @xmath291 with any quantum channel @xmath21 ) acting on the quantum register when executing algorithm [ algounknownpexcluding ] , and let @xmath344 and @xmath293 $ ] .
then algorithm [ algounknownpexcluding ] , with @xmath294 , finds the marked element after at most @xmath345 oracle queries except with probability at most @xmath1 .
the success probability in each round of algorithm [ algounknownpexcluding ] is at least as large as in algorithm [ algounknownpnotexcluding ] because we are excluding elements ; thus , the runtime of algorithm [ algounknownpnotexcluding ] puts an upper bound on the runtime by ( [ stepsinfirstfaulttolerantalgo ] ) . before that , however , algorithm [ thmunknownpwithoutexcluding ] may switch to testing ( and excluding ) elements in random order ( the second selector in ( [ selectstep2ofalgo2 ] ) ) ; this switch happens after at most @xmath346 oracle calls , and after the switch algorithm [ algounknownpexcluding ] needs at most @xmath347 additional calls to find the marked element except with probability @xmath1 .
the constant @xmath300 in ( [ faultignorantbound ] ) can be improved to @xmath299 for a restricted range of parameters , following the remark below theorem [ thmunknownpwithoutexcluding ] .
again , similar to theorem [ thmsymmalgomemoryless ] , we can show that , despite restricting to grover s specific steps , algorithm [ algounknownpexcluding ] is essentially optimal within a wider class of algorithms .
namely , we extend the class of algorithms considered in and before theorem [ thmsymmalgomemoryless ] in such a way that , instead of requiring independence of the failure probabilites in different rounds , we assume that after each unsuccesful round we exclude the tested element and thereby reduce the search space as well as the state space of the computation .
we also need to ensure that failure probabilities are multiplicative , which is the case e.g. if both the noise and the algorithm treat the search elements uniformly ( this in particular applies to grover iterations and partial depolarizing noise , see eq .
[ partiallydepolnoise ] ) . this wider class of algorithms is qualitatively different from the algorithms considered in theorem [ thmsymmalgomemoryless ] , as it now contains algorithms succeeding with @xmath348 oracle calls , _ independently _ of @xmath2 , for example the classical verification - and - exclusion algorithm described at the beginning of sec .
[ firstsearchwithexclusionsubsection ] .
this is reflected in the fact that the lower bound in the following theorem never exceeds @xmath0 , unlike the bounds on @xmath240 in the memoryless setting from theorem [ thmsymmalgomemoryless ] .
[ thmlowerboundwithmemory]for any quantum search algorithm ( that may or may not have the noise level @xmath2 as an input ) satisfying the above constraints and whose quantum register is subject to depolarizing noise ( see eq .
( [ partiallydepolnoise ] ) ) with fixed strength @xmath349 , the number @xmath240 of queries to find the marked element up to fixed failure probability @xmath28 is lower bounded as @xmath350 for some sequence @xmath351 with @xmath352 .
we can assume that @xmath353 , because there does exist an algorithm with this limit being @xmath153 ( see above ) .
for now , fix @xmath0 , and by letting @xmath144 later we will make sure that all following expressions are well - defined ( e.g. no logarithms of negative arguments occur , etc . ) .
let the number of queries in round @xmath302 ( @xmath354 ) be @xmath355 ( i.e. including the verification - exclusion step ) , so @xmath356 . for the success probabilty in round @xmath302 we have by ( [ moretrivialimplication ] ) and ( [ 8kplus1overp ] ) @xmath357 where the latter inequality implies that @xmath358 is bounded independently of @xmath37 ( and of @xmath0 ) , since @xmath2 is given .
thus we can lower bound the failure probability ( using @xmath359 ) : @xmath360^{-1}\right\}~\ge~\exp\left\{-\frac{8}{p}\frac{q_n}{n - q_n}c_n\right\}~,\label{longchaininmemoryopimalthem}\end{aligned}\ ] ] where we used @xmath361 from above and defined @xmath362^{-1}$ ] , which converges to @xmath185 as @xmath144 . inverting ( [ longchaininmemoryopimalthem ] ) to get an explicit bound on @xmath240 finally gives ( [ eqninmemorythmlowerbound ] ) . with the usual conventions in treating @xmath363 and @xmath364 , theorem [ thmlowerboundwithmemory ] is valid for all
@xmath365 $ ] .
similar to eq .
( [ finitenbound ] ) in theorem [ thmsymmalgomemoryless ] , one could explicitly specify a sequence @xmath366 in the bound ( [ eqninmemorythmlowerbound ] ) which would however complicate the expression . in summary , algorithm [ algounknownpexcluding ] , which uses the exclusion strategy and grover iterations , is _ fault - ignorant _ and theorem [ thmknownpwithoutexcluding ] provides an upper bound on its runtime .
conversely , theorem [ thmlowerboundwithmemory ] provides a lower bound on the number of oracle calls for any symmetric fault - ignorant algorithm using the exclusion strategy . and
the inequalities from lemma [ memoryoptimaliylemma ] ( see appendix [ app : technical ] ) show that algorithm [ algounknownpexcluding ] is basically optimal within this class of algorithms , in the sense that for any @xmath367 its runtime is at most a constant factor ( independent of @xmath2 and @xmath1 ) above the lower bound from theorem [ thmlowerboundwithmemory ] .
and even stronger , the lower bound on the runtime in theorem [ thmlowerboundwithmemory ] is proven for algorithms that may `` know '' the noise level @xmath2 as one of their inputs ( such as the algorithm resulting from ( [ mindynprogramming ] ) ) , whereas our algorithm [ algounknownpexcluding ] basically saturates this lower bound _ without _ actually being dependent on the actual noise level @xmath2 ; the latter feature is the characteristic of _ fault - ignorant algorithms_. thus , not knowing the noise level inflicting upon the quantum computation extends the runtime at most by a constant factor , which was observed in the memoryless setting following theorem [ thmunknownpwithoutexcluding ] as well . in section [ sec : memoryless ]
we did not allow for a classical memory ( except to store the correct output ) , whereas in section [ sec : withmemory ] we allowed a noiseless classical register in order to exclude falsified items from future search rounds .
this is obviously not the most general class of algorithms .
one may for example perform non - projective measurements which could result in a non - uniform distribution over oracles ( cf .
@xcite ) after the measurement . or one may abandon the division into `` rounds '' altogether , and rather use the noisy quantum register and noiseless classical memory in a more general way ( cf .
appendix [ sec : faultignorance ] ) .
while these possibilities are rather vague , at least in the noiseless case ( @xmath56 ) grover s algorithm is exactly optimal @xcite .
a similar general proof eludes us in the noisy case considered in this paper ; theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] give such a bound under more restrictive qualifications .
nevertheless , the results obtained here may suggest that any nonzero noise level @xmath147 ( in our noise models , cf .
appendix [ sec : noisemodels ] ) prolongs the runtime beyond the noiseless lower bound ( in @xcite ) , necessitating it to be proportional to the number of search items as @xmath144 .
however , for small but constant noise level @xmath147 , the runtime bound @xmath368 on our algorithms ( cf .
theorems [ thmunknownpwithoutexcluding ] and [ thmknownpwithoutexcluding ] ) can be far below the @xmath369 oracle calls required by the best noiseless classical algorithm . in this regard ,
see @xcite for a treatment of _ locally _ acting noise and questions about optimality in this case .
similar in spirit to our theorems [ thm8kplus1overp ] , [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] , a lower bound of @xmath370 on the runtime of general noisy quantum search algorithms was obtained in @xcite , whose faulty oracle model is somewhat similar to our partial depolarizing noise ( [ partiallydepolnoise ] ) ( with roughly the same noise parameter @xmath2 ; they fixed @xmath371 ) ; see also @xcite for a continuous - time analogue of this result .
one difference is that , in theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] , we allow error - free ( e.g. classical ) verification steps , whereas every oracle use in @xcite is potentially faulty , leading to a diverging bound as @xmath372 . also , @xcite does not include a noiseless classical memory . due to these extensions ,
our lower bounds are restricted to `` symmetric '' algorithms consisting of `` rounds '' , whereas @xcite applies to _ all _ algorithms within their memoryless setting .
these works do not consider achievability of the bound .
the work @xcite specifically investigated grover s algorithm under phase noise ( see also @xcite ) , again somewhat analogous to our noise model ( [ partiallydepolnoise ] ) .
it was observed that grover s algorithm gives an advantage only if it runs for @xmath373 steps , and it was hinted that at this time one may perform a measurement and start a new grover round . in sections
[ basicbuildingblocksubsection ] and [ subsectrepeatbuildingblockknownp ] , we give more rigorous arguments ( and prefactors ) for the scaling @xmath374 , also for algorithms not necessarily consisting of grover steps . our algorithms [ algounknownpnotexcluding ] and [ algounknownpexcluding ] do indeed use the division into grover rounds , but they even function _ fault - ignorantly_. the avoidance of active error correction @xcite is advocated by @xcite as well .
a more technical difference of our work to most of the literature is that we consider the _ average _ success probability of an algorithm , i.e. averaged over all @xmath0 oracles with equal weight , whereas the literature most often only investigates the _ minimum _ success probability of any of the @xmath0 oracles .
this makes our lower bounds stronger than the ones obtained in the literature .
( as our constructive algorithms are symmetric , the minimum and average success probabilities coincide for those . )
in this paper we have investigated the idea of _ fault - ignorant quantum algorithms_. such algorithms should output the correct result even in the presence of noise of potentially unknown strength , in such a way that the actual noise level @xmath2 may affect the runtime it takes to arrive at the correct answer ( up to some specified failure probability @xmath1 ) , but should not affect that fact that the correct answer is found eventually .
this approach allows to reduce the required spatial circuit sizes , for example in comparison to using full - scale quantum error correction , however at the expense of increased runtime . following this general idea ,
we have provided fault - ignorant algorithms for quantum searching that function under depolarizing or dephasing noise of unknown strength @xmath2 .
we find the `` quadratic speedup '' to be achievable only for low decoherence rates @xmath375 .
otherwise , our best algorithm s runtime scales asymptotically like @xmath376 as @xmath144 .
this is linear in @xmath0 , but for low enough noise levels @xmath2 it nevertheless outperforms the optimal classical search algorithm .
our algorithms may thus be useful for initial uses of quantum computing , when unlimited scalability of the size of quantum computers is not yet be achievable due to technological limitations .
we moreover proved that , up to a constant factor , our algorithms runtimes are optimal within wide classes of noisy quantum search algorithms .
remarkably , for the searching task , it turned out that ignorance of the actual noise level will extend the runtime by only a constant factor compared to the case of known noise level @xmath2 .
due to the novelty of the approach , our algorithms and lower bounds leave questions for further research . on the side of concrete algorithms , one may ask for them to be independent not only of the noise level @xmath2 but also of the desired accuracy @xmath1 ; then one could continue running the algorithm for longer time to increase the success probability or accuracy . concerning lower bounds on the complexity of noisy quantum search , it would be worthwhile to establish an analogue of theorem [ thm8kplus1overp ] for the case of local noise models or even partial dephasing or general partial entanglement - breaking noise .
the latter would immediately extend the validity of our lower bound in theorem [ thmlowerboundwithmemory ] to the class of quantum algorithms that use a noiseless classical register in an arbitrary way and need not be divided into `` quantum rounds '' . in a similar vein , it may also be possible to prove that the essentially optimal runtime for quantum searching under partial depolarizing noise , which we mainly investigated , can always be achieved by an algorithm divided into such rounds ( see sec . [ moregeneralmemoryalgos ] ) . if not , it would be very interesting to find fault - ignorant algorithms that are not of this simple check - and - repeat form .
finally , it would be desirable to investigate whether and how the fault - ignorant idea could possibly be applied to computational models other than the quantum circuit model .
this would in particular be desirable in computational models for which quantum error correction techniques are less developed , such as adiabatic quantum computing , and where other methods to achieve fault tolerance are needed .
generally , we hope that , beyond unstructured search , the fault - ignorant idea will be fruitfully applied to algorithmic tasks , such as sampling algorithms .
* acknowledgments : * we thank the two anonymous referees for their insightful comments which helped to improve the paper .
this research was initiated at a workshop of the fp7 project coquit , which also supported p. vrana and d. reitzner .
d. reeb was supported by the marie curie intra - european fellowship quintyl .
m. wolf acknowledges support from the chist - era / bmbf project cqc and the alfried krupp von bohlen und halbach - stiftung .
the aim of this appendix is to provide a rigorous mathematical definition of fault - ignorant computing ( see section [ moreformaldescription ] for a less formal discussion ) .
we do this in a way which enables to include a fairly broad class of algorithmic problems into this framework in a unified manner , while keeping the definition reasonably simple .
the definitions are supposed formalize algorithms that do not need to know the actual noise level in order to accomplish their task they should be _ ignorant _ of the noise .
a fault - ignorant algorithm should be robust enough to provide the answer ( up to some specified failure probability ) under any level of noise , the latter affecting only its runtime . in our formalization , we want to allow the desired and the actual output of the algorithm to be probabilistic ( as is usual in sampling and quantum simulation problems ) , and to depend on an input ( as for example in computational problems ) and on an oracle ( as in search problems ) .
given the discrete - time nature of the computation as well as of our noise models ( cf .
appendix [ sec : noisemodels ] ) , it is necessary to explicitly refer to an allowed class of quantum operations ( the set @xmath18 in the following definition ) .
this can be done most conveniently if one also specifies the ( spatial ) resources available for performing the computation , i.e. the size of the quantum computer available or of any accompanying classical register .
we thus view the specification of the size of the available quantum register as part of the task to be solved ; and indeed , since early realizations of quantum computers will be limited in the number of qubits , it will be a part of the challenge to solve a desired task on the _ available _ hardware , esp . under noise influence because full - scale quantum error correction may be prohibited .
further , we consider only the quantum register to be noisy , whereas noiseless classical memory is today a reasonable technological assumption . in light of this
, we propose the following definitions , which we explain and supplement by examples afterwards .
[ definitionnoisytask ] a _ noisy quantum computational task _ is a tuple @xmath377 where * @xmath378 , @xmath379 , @xmath380 are sets , * @xmath381 is a stochastic matrix , i.e. has nonnegative entries and for any @xmath5 and @xmath6 we have @xmath382 , * @xmath8 is a hilbert space , * @xmath383 is a function with density operators as values , * @xmath384\to\operatorname{cpt}(\operatorname{\mathcal{b}}(\mathcal{h}))$ ] is a function with quantum channels on @xmath385 as values , * @xmath386 , * @xmath387 . a fault - ignorant algorithm ( definition [ definefaultignorantalgo ] ) , specified to solve a noisy quantum computational task ( definition [ definitionnoisytask ] ) : an input state @xmath388 is affected in turn by devised operations @xmath389 ( which may include an oracle indexed by @xmath20 , or other coherent operations , or measurement / verification procedures that store information in noiseless classical registers @xmath390 , @xmath391 , @xmath392 ) and noise @xmath16 acting on the quantum register . after some number of steps
, the probability distribution of the output @xmath4 should approximate the desired distribution @xmath393 , up to error @xmath1 . ]
we interpret @xmath378 as the set labelling the different oracles , @xmath379 and @xmath380 as the sets of possible inputs and outputs , respectively ( see also fig . [
fig : ignorance ] ) . for a task which does not make use of an oracle ,
we let @xmath378 be any singleton set , and similarly , if the computation does not need an input , we let @xmath41 .
the stochastic matrix @xmath394 describes the desired distribution on the output set depending on the input and the oracle .
the computation is performed using the hilbert space @xmath8 and a classical memory of @xmath395 states , with the output being written into an additional classical register with @xmath396 states , corresponding to the possible output states in @xmath380 .
the additional classical bit @xmath397 is to have value @xmath185 iff the algorithm wants to signal that the result is available in the register @xmath398 .
the reason for this is that in the fault - ignorant setting the runtime depends on an unknown parameter ( namely @xmath2 , see below ) , and therefore the algorithm needs a way to tell whether the computation is already done , without destroying the quantum state . the map @xmath383 plays the role of input encoding in the sense that the physical initial state @xmath388 on the register @xmath385 represents the abstract input value @xmath6 .
the quantum register @xmath8 is subject to noise modeled by the quantum channels @xmath16 ( as specified in eq .
( [ defineoutputstate ] ) below ) depending on a parameter @xmath17 $ ] , which we think of as a strength parameter .
finally , the set @xmath399 represents the set of allowed elementary steps .
an element in this subset is understood as a quantum channel depending on the oracle @xmath5 , whereby the quantum channel acts on the quantum register @xmath385 as well as on the classical ( diagonal ) registers @xmath390 , @xmath400 and @xmath392 described above ( our specification below will be such that all these classical registers are initialized in the state @xmath401 at the start of an algorithm ) .
this gives a way to impose conditions on how `` complicated '' the elementary operations are , e.g. in terms of oracle use or locality requirements ( see examples below ) , and at the same time it maps the abstract oracle @xmath5 to actual physical transformations @xmath22 it may perform .
[ definefaultignorantalgo ] a _ fault - ignorant algorithm _ solving the noisy quantum computational task @xmath377 is a family @xmath402 of finite or infinite sequences with @xmath403 such that for all @xmath28 and for all @xmath17 $ ] the value @xmath404 is finite , where @xmath405 is defined by @xmath406 and @xmath407 for @xmath5 , @xmath6 , @xmath4 , and @xmath408\label{defineoutputstate}\ ] ] is a probability distribution on @xmath409 .
thus , the sequence of operations in ( [ defineoutputstate ] ) models a @xmath410-step noisy quantum computation , in the sense that between any two elementary operations from @xmath18 a noise channel @xmath16 is to be applied on the quantum register @xmath385 ( fig .
[ fig : ignorance ] ) .
the sequence @xmath411 itself describes the algorithmic operations , which may depend on the required accuracy @xmath1 , i.e. on the maximally tolerable distance from the desired output distribution @xmath412 , cf .
( [ defineminimalsteps ] ) .
the requirement for @xmath413 to be finite for _ any _ @xmath2 , even though the algorithm @xmath402 does _ not _ depend on @xmath2 , justifies the term _ fault - ignorant algorithm_. the condition `` @xmath414 '' in ( [ defineminimalsteps ] ) requires the result to be available in the classical memory at any later time when the outside agent , ignorant of the noise level @xmath2 and thus of the necessary computation time @xmath413 , may check the @xmath397 flag to see whether the computation has already finished and want to read out the result .
note that definition [ definefaultignorantalgo ] does _ not _ put any requirements on the efficiency of the algorithm , which however in some circumstances may be quantified by @xmath413 , i.e. the minimal number of invocations of @xmath415 ( e.g. oracle calls ) ; see section [ moreformaldescription ] .
we now illustrate the definitions above by two examples . as an example
we now show how the noisy quantum search problem considered in sections [ sec : memoryless ] and [ sec : withmemory ] fits into this framework . in this case
we have a set of @xmath0 oracles @xmath416 , and the algorithm is required to identify the oracle , so @xmath417 . since no input is needed , we set @xmath12 .
now the function to be computed is deterministic , so @xmath394 will be a @xmath151-@xmath185 matrix , more specifically @xmath418 .
the hilbert space we use is @xmath419 for some @xmath420 setting the size of the ancillary quantum system and @xmath392 standing for the ancillary system used by the oracle , cf .
eq . ( [ oraclebasic ] ) .
the noise acting on it is for example partial depolarizing , @xmath421 .
since there is no input , @xmath142 is just any fixed initial state , e.g. we may take @xmath422 . in the version without classical memory we set @xmath423 ( section [ sec : memoryless ] ) , while if we are to exclude previously tested elements , we may set @xmath424 corresponding to an @xmath0-bit classical memory ( section [ sec : withmemory ] ) .
the set of allowed elementary operations to be applied between two noise hits is @xmath425 where @xmath123 first prepares the pure state @xmath426 on the @xmath392-subsystem of @xmath8 , and then acts as @xmath427 on @xmath8 ( cf .
( [ oraclebasic ] ) and ( [ oracleaction ] ) ) and as the identity on the classical registers .
this choice of @xmath18 means that an elementary step consists of a single use of the oracle , possibly applying an arbitrary ( but oracle - independent ! ) channel before and afterwards .
finally , half of the trace - distance in ( [ defineminimalsteps ] ) gives , when the ready - flag @xmath392 has been set to @xmath185 , exactly the probability of not outputting the correct oracle index in the classical output register , and it is this failure probability which we wanted to be smaller than @xmath1 in sections [ sec : memoryless ] and [ sec : withmemory ] .
[ binaryadditionexample]this example illustrates the possibility to introduce some kind of `` locality structure '' .
the task is the addition of two @xmath77-bit numbers given their binary representation using local gates on a @xmath428-bit quantum register with local dephasing noise .
such a task is given by @xmath429 , @xmath430 , @xmath431 , @xmath432 , the hilbert space is @xmath433 , @xmath434 ( with @xmath435 considered as a @xmath428-bit string ) , @xmath436 with @xmath437 the partial dephasing with strength @xmath2 , @xmath423 , and @xmath438 consisting of @xmath185- and @xmath439-(qu)bit gates , i.e. channels which act as the identity on all but at most two bits ( quantum or classical ) , remembering the subsystem structure of @xmath440 .
an algorithm that works only for _ known _ noise level @xmath2 is _ not _ fault - ignorant ; such algorithms may be formalized by assuming a @xmath2-dependence @xmath441 in the family of sequences in definition [ definefaultignorantalgo ] . on the other hand ,
if these sequences do _ not _ depend on the desired accuracy @xmath1 , i.e. @xmath442 , then the algorithm does have another feature : the level of accuracy @xmath1 need not be specified before starting the algorithm ; when higher accuracy is desired ( i.e. smaller @xmath1 ) , one only needs to continue running the algorithm for longer time .
returning to efficiency questions , one may call a fault - ignorant algorithm ( or rather , a family of fault - ignorant algorithms , parametrized by some `` problem size '' @xmath0 ) _ efficient _ if , for any @xmath2 , any @xmath1 and any @xmath0 , its runtime is within a constant factor times the runtime of the best algorithm that may depend on @xmath1 _ and _ on @xmath2 ( see section [ moreformaldescription ] ) . in this sense ,
our theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] can be seen as statements that algorithms [ algounknownpnotexcluding ] and [ algounknownpexcluding ] are efficient ( within restricted classes of algorithms ) .
it should be clear that there is nothing special about the set @xmath443 $ ] parametrizing the noise channels apart from the possibility to interpret it as `` strength '' or to use it directly as a coefficient in a convex combination .
one could instead consider a family @xmath444 of noise channels indexed by an arbitrary set @xmath445 parametrizing wider classes of noise , and so allowing for `` more '' ignorance about the faults .
another obvious extension of definition [ definitionnoisytask ] would be to allow for time - dependent noise .
here we elaborate on different kinds of noise which may be acting on the quantum computer during its runtime , and in particular on the noise models to which our results apply . _
partial depolarizing _ , @xmath446 for noise level @xmath17 $ ] , has been defined in eq .
( [ partiallydepolnoise ] ) , and corresponds to erasing the state of the quantum register with probability @xmath2 ( between any two oracle calls ) .
somewhat similar is _ partial dephasing _ , @xmath447 acting on states @xmath85 on a @xmath86-dimensional hilbert space equipped with a distinguished orthonormal basis @xmath448 ( for these , we imagine the basis states with respect to which the oracles ( [ oraclebasic ] ) act ) . for @xmath449
, all quantum coherence is lost between any two oracle calls , but one can still perform a classical algorithm ( on the basis states @xmath69 ) ; in this sense , the noise level @xmath2 of partial dephasing parametrizes how `` quantum '' a search algorithm may be .
our constructive algorithms also work with the runtimes guaranteed by theorems [ thmunknownpwithoutexcluding ] and [ thmknownpwithoutexcluding ] under the more general noise model @xmath450 where @xmath21 may be any quantum channel , see discussion below eq .
( [ partiallydepolnoise ] ) .
our formalization of noisy search algorithms ( sections [ sec : memoryless ] and [ sec : withmemory ] ) does allow to _ noiselessly check _ whether a given index @xmath451 equals the marked element @xmath20 , since immediately before and after an oracle call one may perform any quantum operation without noise ( cf .
( [ statefrombasicbuildingblock ] ) ) and thus one action of @xmath67 from ( [ oraclebasic ] ) on a suitably prepared quantum register can accomplish this check and write the result into the ( noiseless ) classical memory .
this fact is important , as it allows the verification / falsification step at the conclusion of each round ( cf .
algorithms [ algounknownpnotexcluding ] and [ algounknownpexcluding ] ) .
alternatively , such a noiseless check may be implemented by a classical table lookup .
the above noise models are formulated in discrete time , but our prescription for the noise @xmath16 to act between any two oracle calls is supposed to model the continuous action of noise in a real - world situation .
for example , since in ( [ partialdepolsecond ] ) the probability to `` lose '' the quantum computer between any two consecutive oracle calls is @xmath2 , its lifetime is roughly @xmath452 ( measured in the time between two oracle calls ) ; and indeed , the time scale @xmath374 appeared often in the analysis in subsection [ subsectrepeatbuildingblockknownp ] .
note that quantum error correction @xcite does not work for partial depolarizing ( [ partialdepolsecond ] ) or dephasing ( [ definepartialdephasing ] ) , as these noises affect the whole quantum computer `` collectively '' .
this means that the whole quantum computer is subjected to a `` flash '' of noise , such as drifting lasers or an external hit by a magnetic field .
these may be reasonable noise models for not - too - large quantum computers .
discussing the noise models more quantitatively , we first notice that the lower bound ( [ lowerboundprobs ] ) on the success probability after @xmath37 steps of grover s algorithm under noise applies to all three noise models ( [ partialdepolsecond])([gentnoise ] ) . for partial depolarizing ( [ partialdepolsecond ] ) one can compute the success probability in ( [ lowerboundprobs ] ) exactly : the @xmath453 omitted terms are of the form @xmath454 where @xmath154 is the number of noise hits .
these terms correspond to events when the maximally mixed state is prepared at some point due to noise acting and since both @xmath16 and @xmath455 ( see before eq .
( [ firststatementsofpsinmaintext ] ) ) are unital .
as the coefficients of these terms sum up to @xmath456 , the exact success probability for this model is @xmath457 using this exact success probability for partial depolarizing improves the runtime bounds for this specific noise model ( e.g. theorem [ thmunknownpwithoutexcluding ] for large noise level @xmath2 ) , but the lower bound ( [ lowerboundprobs ] ) is quite tight unless @xmath458 .
the drawbacks of relying on a too specific noise model are furthermore discussed below eq .
( [ eq : firstbound ] ) .
an exact computation of the success probability can also be done for partial dephasing ( [ definepartialdephasing ] ) , but is much more involved .
furthermore , one can prove that the success probability for partial dephasing is not smaller than for depolarizing at the _ same _ noise level : @xmath459 .
this inequality is , however , not immediate , as for identical noise parameters @xmath349 , partial depolarizing @xmath460 _ can not _ be obtained by post - processing @xmath461 , i.e. @xmath462 for all quantum channels @xmath445 .
our proofs of the general lower bounds on the number of oracle calls ( theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] ) require partial depolarizing ( [ partialdepolsecond ] ) , as theorem [ thm8kplus1overp ] was proved only for generalized partial depolarizing noise @xmath463 and the proofs ( and presuppositions ) of theorems [ thmsymmalgomemoryless ] and [ thmlowerboundwithmemory ] require furthermore a symmetry between the oracle indices , limiting further to @xmath464 .
finally , we argue that it makes sense in sections [ sec : memoryless ] and [ sec : withmemory ] to perform efficiency analyses by keeping the noise parameter @xmath2 fixed while the size of the quantum register @xmath0 ( or @xmath465 ) varies , possibly even tending to infinity . phrased another way
, we ask whether , for example , the strength of partial depolarizing ( [ partialdepolsecond ] ) with parameter @xmath2 on an @xmath86-dimensional quantum system is comparable to the strength of partial depolarizing with the same parameter @xmath2 in @xmath466 dimensions , even if @xmath86 and @xmath466 are widely different .
first , both partial depolarizing ( [ partialdepolsecond ] ) and partial dephasing ( [ definepartialdephasing ] ) ( the latter with respect to a tensor product of bases ) are compatible under tracing out subsystems when the _ same _ parameter @xmath2 is used on the tensor product system and on the subsystem : @xmath467~=~d^{i,\varphi}_p\left({\rm tr}_b[\varrho_{ab}]\right)\qquad\forall p\in[0,1]~.\ ] ] with this parametrization of the noise , it does therefore not help for algorithm performance to introduce larger and larger ancillary systems or `` innocent bystanders '' : the noise on the `` grover part '' of the algorithm can not be made small in this way , which is a reasonable requirement . secondly , both for partial depolarizing and dephasing , one can obtain the noise in @xmath468 dimensions by post - processing the noise on a @xmath86-dimensional system : @xmath469\,\forall\varrho\in\operatorname{\mathcal{b}}(\mathbb{c}^{d-1})~,\ ] ] where @xmath470 for dephasing noise ( and the additional dimension @xmath471 has to correspond to one of the basis vectors in ( [ definepartialdephasing ] ) ) , and @xmath472 for depolarizing .
this compatibility under restrictions of the hilbert space to subspaces is important and sensible in the context of exclusion algorithms ( algorithm [ algounknownpexcluding ] , and proof of theorem [ thmknownpwithoutexcluding ] ) , where the effective dimension of the quantum register is reduced by @xmath185 in each round .
as we assume the noise to act symmetrically with respect to the different oracles ( which both partial depolarizing and dephasing do ) and since the grover steps of algorithm [ algounknownpnotexcluding ] are symmetric as well , the success events in different rounds @xmath302 are independent . thus , with ( [ lowerboundprobs ] ) , we can upper bound the failure probability after round @xmath303 by @xmath473 to show that the failure probability is at most @xmath1 , as desired , below we will lower bound the sum @xmath305 and adjust parameters such that it is at least @xmath306 .
the sum can be further bounded by assuming it to start at some @xmath474 with @xmath475 : @xmath476 the number of oracle calls can be upper bounded as follows : @xmath477 the proof of the theorem now consists in showing that there exists a number @xmath303 ( of rounds ) such that the failure probability ( [ repeatfailureprobabinappendix ] ) is at most @xmath1 , while the number of oracle calls ( [ repeatupperboundonoraclecallsinappendix ] ) does not exceed the value given in ( [ stepsinfirstfaulttolerantalgo ] ) .
this argument will be split into three cases , as sketched in the main text .
we make abundant use of the fact that @xmath478 , since @xmath479 and @xmath294 .
we also define @xmath480 and assume @xmath481 throughout , in accord with the statement of theorem [ thmunknownpwithoutexcluding ] .
* case 1 : * @xmath482 . in this case , the actual decoherence rate @xmath2 is small , and we take only the first few rounds @xmath302 into account to obtain an upper bound on @xmath483 . by using @xmath484 , @xmath485 for @xmath486 , and setting @xmath487}{\left[\left(\frac{\pi}{2}\sqrt{n_0}+1\right)\arcsin(1/\sqrt{n_0})\right]^2}~,\ ] ]
we continue in bounding ( [ eq : sumfail ] ) : @xmath488 where we used @xmath489 for @xmath490 .
choosing @xmath491 and @xmath492 , we evaluate the integral in ( [ eq : sumfailadvanced ] ) to find @xmath493 where we used @xmath494 .
when the prefactor of @xmath306 is at least @xmath185 , then the failure probability @xmath483 will be at most @xmath1 by eq .
( [ repeatfailureprobabinappendix ] ) ; this happens e.g. for the choice @xmath495 .
the number of oracle calls from ( [ repeatupperboundonoraclecallsinappendix ] ) is then @xmath496 which is less than @xmath497 due to @xmath498 .
we notice that the term linear in @xmath0 ( cf .
( [ stepsinfirstfaulttolerantalgo ] ) ) is absent from the runtime ( [ upperboundonruntimeincase1 ] ) in case 1 ; intuitively speaking , such small noise levels @xmath2 still allow for quadratic speedup in the quantum search .
* case 2 : * @xmath499 , where we define @xmath500 . in this intermediate region of the actual decoherence rate ( the need for
@xmath501 will become evident later ) , we define @xmath502 and @xmath503 with @xmath504 to be determined later .
our choice @xmath505 will in particular imply @xmath506 , so that we can continue lower - bounding ( [ eq : sumfailadvanced ] ) by bounding the integrand , @xmath507 where we used @xmath508 ( due to @xmath204 ) and @xmath509 ( due to @xmath510 ) .
again , the prefactor of @xmath306 can be made larger than @xmath185 by choosing @xmath511 and @xmath512 .
the number of oracle calls from ( [ repeatupperboundonoraclecallsinappendix ] ) is then @xmath513 which again is less than @xmath514 .
* case 3 : * @xmath204 . for large actual noise levels
@xmath2 , it is enough to consider only those rounds @xmath302 for which @xmath515 ; in each such round , a measurement is performed on the equal superposition state , leading to a success probability of exactly @xmath115 .
this leads to the choice @xmath516 and @xmath517 , and we can lower - bound ( [ eq : sumfail ] ) : @xmath518 by choosing @xmath519 , the prefactor of @xmath306 exceeds @xmath185 .
the number of oracle calls from ( [ repeatupperboundonoraclecallsinappendix ] ) is then @xmath520 which is again less than @xmath514 .
so far we have proved that the algorithm is fault - tolerant with runtime at most @xmath514 .
due to @xmath521 , ( [ stepsinfirstfaulttolerantalgo ] ) is an upper bound on the runtime . for the proof of the prefactor
@xmath299 mentioned below theorem [ thmunknownpwithoutexcluding ] , the case 3 in the proof above can be neglected , and we set @xmath522 , and alter the lower bound ( [ eq : sumfailadvanced ] ) a bit , such that constants @xmath523 , @xmath524 , @xmath525 etc . can be found to yield the lower guaranteed runtime .
by the following lemma , we convert zalka s implicit bound , eq . ( [ zalkaimplicit ] ) , into an explicit one ( see after the proof below ) : the left hand side is a concave function of @xmath530 ^ 2 $ ] , smooth in the interior , and hence its graph stays under its tangent plane drawn at any point @xmath531 for @xmath532 and @xmath533 .
the partial derivatives of @xmath534 at @xmath535 are @xmath536 writing @xmath537 for @xmath528 , we have @xmath538 now taking the limit @xmath539 yields ( [ lemmaxyineqeqn ] ) .
we apply this lemma to zalka s bound @xcite ( eq .
( [ zalkaimplicit ] ) above ) with @xmath540 and @xmath541 : @xmath542 \\ & \ge~2n\left[1-\left(1+\frac{1}{2}(\alpha-1 ) p_s+\frac{1}{2}\left(\frac{1}{\alpha}-1\right)\frac{1}{n}\right)\right ] \\ & = ~(1-\alpha)np_s-\left(\frac{1}{\alpha}-1\right)~. \end{split}\ ] ] one can easily see that the sharpest bound on @xmath99 is obtained for @xmath543 , yielding @xmath544 .
the following lemma shows that algorithm [ algounknownpexcluding ] in section [ sec : withmemory ] is optimal within the class of algorithms considered in theorem [ thmlowerboundwithmemory ] , up to a constant factor in the runtime : now let @xmath550 for @xmath551 . then @xmath552 and @xmath553 , so @xmath554 is concave , and the tangent at @xmath555 is @xmath556 .
this gives @xmath557 , which implies @xmath558 concluding the left inequality .
for the right inequality , if @xmath559 , then @xmath560
lastly , if @xmath561 then in particular @xmath562 . thus , finally , @xmath563 99 m.a .
nielsen , i.l .
chuang , _ quantum computation and quantum information , _ cambridge university press , cambridge ( 2000 ) .
p. shor , _ scheme for reducing decoherence in quantum memory , _ phys .
rev . a * 52 , * 2493 ( 1995 ) .
a.m. steane , _ error correcting codes in quantum theory , _ phys .
* 77 , * 793 ( 1996 ) .
p. shor , _ fault - tolerant quantum computation , _ proc . of the 37th annual symposium on fundamentals of computer science , ieee press , los alamitos , ca ( 1996 ) .
e. knill , r. laflamme , w.h .
zurek , _ resilient quantum computation : error models and thresholds , _ proc .
r. soc .
a * 454 , * 365 ( 1998 ) .
j. preskill , _ reliable quantum computers , _ proc .
r. soc .
a * 454 , * 385 ( 1998 ) .
d. gottesman , _ fault - tolerant quantum computation with constant overhead , _
[ quant - ph ] ( 2013 ) .
zanardi , _ dissipative dynamics in a quantum register , _ phys .
a * 56 , * 4445 ( 1997 ) .
lidar , i.l .
chuang , k.b .
whaley , _ decoherence - free subspaces for quantum computation , _ phys . rev .
* 81 , * 2594 ( 1998 ) .
grover , _ quantum mechanics helps in searching for a needle in a haystack , _ phys .
* 79 , * 325 ( 1997 ) .
bennett , e. bernstein , g. brassard , u. vazirani , _ strengths and weaknesses of quantum computing , _ siam j. comput .
* 26 , * 1510 ( 1997 ) .
m. boyer , g. brassard , p. hoyer , a. tapp , _ tight bounds on quantum searching , _ fortsch .
* 46 , * 493 ( 1998 ) . c. zalka , _
grover s quantum searching algorithm is optimal , _ phys .
rev . a * 60 , * 2746 ( 1999 ) .
b. pablo - norman , m. ruiz - altaba , _ noise in grover s quantum search algorithm , _ phys .
rev . a * 61 , * 012301 ( 1999 ) .
long , y.s .
li , w.l .
zhang , c.c .
tu , _ dominant gate imperfection in grover s quantum search algorithm , _ phys .
rev . a * 61 , * 042305 ( 2000 ) .
h. azuma , _ decoherence in grover s quantum algorithm : perturbative approach , _ phys .
a * 65 , * 042311 ( 2002 ) .
d. shapira , s. mozes , o. biham , _ effect of unitary noise on grover s quantum search algorithm , _ phys .
rev . a * 67 , * 042301 ( 2003 ) .
j. hsieh , c. li , d. chuu , _ an improved phase error tolerance in quantum search algorithm , _ chin .
j. phys .
* 42 , * 585 ( 2004 ) .
n. shenvi , k.r .
brown , k.b .
whaley , _ effects of a random noisy oracle on search algorithm complexity , _ phys .
a * 68 , * 052313 ( 2003 ) .
o. regev , l. schiff , _ impossibility of a quantum speed - up with a faulty oracle , _ proc . of the 35th international colloquium on automata , languages and programming * 1 , * 773 ( 2008 ) .
k. temme , _ a note on the runtime of a faulty hamiltonian oracle ,
_ arxiv:1404.1977 [ quant - ph ] ( 2014 ) .
a. montanaro , _ quantum search with advice , _ proc . of the 5th conference on theory of quantum computation ,
communication and cryptography , pg .
77 , springer , berlin ( 2010 ) .
m. ben - or , d. gottesman , a. hassidim , _ quantum refrigerator , _
arxiv:1301.1995 [ quant - ph ] ( 2013 ) . | we investigate the problem of quantum searching on a noisy quantum computer . taking a _ fault - ignorant _ approach ,
we analyze quantum algorithms that solve the task for various different noise strengths , which are possibly unknown beforehand .
we prove lower bounds on the runtime of such algorithms and thereby find that the quadratic speedup is necessarily lost ( in our noise models ) .
however , for low but constant noise levels the algorithms we provide ( based on grover s algorithm ) still outperform the best noiseless classical search algorithm . |
the bilateral filter of tomasi and maduchi @xcite is a particular instance of an edge - preserving smoothing filter .
the origins of the filter can be traced back to the work of lee @xcite and yaroslavsky @xcite .
the susan framework of smith and brady @xcite is also based on a similar idea .
the relation between the bilateral and other closely related filters is surveyed in @xcite .
the bilateral filter has turned out to be a versatile tool that has found widespread applications in image processing , computer graphics , computer vision , and computational photography .
a detailed survey of some of these applications can be found in @xcite .
more recently , the bilateral filter has received renewed attention in the context of image denoising @xcite . the original bilateral filter @xcite has a straightforward extension to signals of arbitrary dimension and , in particular , to video and volume data @xcite .
thus , while we will limit our discussion to images in this paper , the ideas that we present next can also be extended to higher - dimensional signals .
consider a discrete image @xmath3 where @xmath4 is some finite rectangular domain of @xmath5 .
the gaussian bilateral filtering of this image is given by @xmath6 where both the _ spatial _ and _ range _ kernels are gaussian , @xmath7 in practice , the domain of the spatial kernel @xmath8 is restricted to some neighbourhood of the origin .
typically , @xmath8 is a square neighbourhood , @xmath9 \times [ -w , w]$ ] where @xmath10 @xcite .
we refer the reader to @xcite for a detailed exposition on the working of the filter .
it is clear that a direct implementation of requires @xmath11 operations per pixel . in general , the directly computed bilateral filter is slow for practical settings of @xmath12 @xcite . to address this issue ,
researchers have come up with several fast algorithms @xcite that are based on some form of approximation and yield various levels of speed and accuracy .
we refer the interested reader to @xcite for a survey of algorithms for fast bilateral filtering .
the ultimate goal in this regard is to reduce the complexity to @xmath2 per pixel , that is , the run time of the implementation should not depend on @xmath12 .
this is commonly referred to as a constant - time implementation .
the constant - time algorithms in @xcite are particularly relevant to the present work .
the authors here proceed by approximating the gaussian range kernel using polynomial and trigonometric functions , and demonstrate how the bilateral filter can be decomposed into a series of spatial gaussian filters as result .
now , since a gaussian filter can be implemented in constant - time using separability and recursion @xcite , the overall approximation can therefore be computed in constant - time . in this paper
, we propose a fast @xmath2 algorithm for computing which is motivated by the line of work in @xcite .
in particular , we present a novel approximation for the range term in that allows us to decompose the bilateral filter into a series of fast gaussian convolutions .
the fundamental difference between the above papers and the present approach is that instead of approximating the gaussian and then translating the approximation , we directly approximate the translated gaussians using the so - called gauss - polynomials .
the advantages of the proposed approximation are the following : @xmath13 it is generally much more accurate than the polynomial approximation in @xcite .
@xmath13 for a fixed approximation degree ( to be defined shortly ) , it leads to exactly half the number of gaussian filterings than that required by the approximation in @xcite , and hence has a smaller run time .
@xmath13 it does not involve transcendental functions such as @xmath14 and @xmath15 which are used in @xcite .
it only involves polynomials ( and just a single gaussian ) and hence can be efficiently implemented on hardware @xcite .
this is partly what motivated the present work .
moreover , we also show how the proposed approximation can be improved by first centering the range data , then applying the approximation algorithm , and finally adding back the centre to the processed range data .
the main idea in @xcite was to approximate the range kernel in using appropriate polynomials and trigonometric functions . by using these approximations in place of the gaussian kernel
, it was shown that the numerator and denominator of can be approximated using a series of gaussian filtering .
the present idea is to consider the translates of the range kernel @xmath16 that appear in , where @xmath17 and @xmath18 take values in some intensity range , say , @xmath19 $ ] .
for example , @xmath20 and @xmath21 for an @xmath22-bit grayscale image . we can write @xmath23 for a fixed translation @xmath24 , this is a function of @xmath25 .
notice that the first term is simply a scaling factor , while the second term is a gaussian centered at the origin .
in fact , the second term essentially contributes to the `` bell '' shape of the translated gaussian .
the third term is a monotonic exponential function which is increasing or decreasing depending on the sign of @xmath24 .
this term helps in translating the gaussian to @xmath26 . the decomposition in plays an important role in the rest of the discussion and is illustrated in figure [ decomposition ] .
+ consider the taylor series of the exponential function , @xmath27 by dropping the higher - order terms , we obtain the following approximation of : @xmath28.\ ] ] being the product of a gaussian and a polynomial of degree @xmath29 , we will henceforth refer to as a `` gauss - polynomial '' of degree @xmath29 . at this point , we note that one of the proposals in @xcite was to approximate @xmath16 using its taylor polynomial .
the fundamental difference with our approach is that instead of approximating the entire gaussian , we approximate one of its component , namely the monotonic exponential in .
the intuition behind this is that a polynomial eventually goes to infinity as one moves away from the origin .
this makes it difficult to approximate the gaussian function on its asymptotically - decaying tails . as against this ,
the exponential function in is monotonic and hence can be more accurately approximated using polynomials .
this is explained with an example in figure [ comparison ] .
in particular , notice in figure [ comparison ] ( b ) that the gauss - polynomial approximation is fairly accurate over the entire range of interest and is comparable to the raised - cosine approximation @xcite .
we note that for a fixed degree @xmath29 , the accuracy of the gauss - polynomial approximation in depends on @xmath24 .
indeed , when @xmath30 , there is nothing to approximate since the exponential function reduces to a constant in this case . on the other hand , we see from that the magnitude of the higher - order terms increases with increase in @xmath31 , and the approximation accuracy drops as a result .
this is demonstrated with an example in figure [ comparisontau ] ( a ) .
of course , the approximation can be improved by using gauss - polynomials of higher degree .
however , we note that for a fixed degree , the approximation accuracy can be improved simply by reducing the maximum @xmath31 that appears in .
we propose the following `` centering '' trick in which we set , for example , @xmath32 we next translate each pixel intensity by @xmath33 , which has the effect of centering the transformed intensity range at the origin . for example , when @xmath20 and @xmath21 , the maximum @xmath31 equals @xmath34 . however ,
if we center the intensity range @xmath35 $ ] , say at @xmath36 , then we can effectively reduce the maximum @xmath31 to @xmath37 .
the gauss - polynomial approximations obtained after the centering are shown in figure [ comparisontau ] ( b ) .
the above idea of centering is compatible with the bilateral filter precisely because of the following property of the bilateral filter .
if @xmath38 , then @xmath39 this is a simple consequence of the fact that the range kernel depends only on the intensity difference , and that for a fixed range term , preserves constant functions . in other words , we can first centre the intensity range , apply the bilateral filter , and finally add back the centre to the output .
we now present the constant - time implementation of using gauss - polynomials .
suppose that @xmath29 is the degree of the polynomial in . for @xmath40 , define the images @xmath41 denote the gaussian filtering of @xmath42 by @xmath43 ,
that is , @xmath44 substituting @xmath17 and @xmath18 , and using the gauss - polynomial approximation in place of @xmath16 , it can be verified that ( after interchanging summations ) we can express the numerator of as @xmath45 where @xmath46 similarly , we can express the denominator of as @xmath47 where @xmath48 in other words , we can approximate by @xmath49 note that we have effectively transferred the non - linearity of the bilateral filter to the intermediate images in , which are obtained from the input image using pointwise non - linear transforms .
the main leverage that we get from the above manipulation is that , for any arbitrary @xmath12 , can be computed using @xmath2 operations per pixel @xcite .
the overall cost of computing is therefore @xmath2 per pixel . in other words
, we have a constant - time approximation of the bilateral filter . in this regard , we note that the above analysis holds if we replace the spatial gaussian filter by any other filter ( e.g. , a box filter ) that has a constant - time implementation .
@xmath50 @xmath38 @xmath51 @xmath52 @xmath53 @xmath54 @xmath55 @xmath56 @xmath57 @xmath58 the overall algorithm is summarized in algorithm [ gpf ] .
we will henceforth refer to this as the gauss - polynomial - based bilateral filter ( ` gpf ` ) .
notice that we use centering and to improve the accuracy .
moreover , we efficiently implement steps to .
in particular , we recursively compute the images in and the factorials in and .
notice that steps @xmath59-@xmath60 , @xmath61-@xmath62 , @xmath63 , and @xmath64 are applied to each pixel ( cheap pointwise operations ) . to avoid confusion , we note that the specification of the some of quantities in algorithm [ gpf ] are somewhat different from the corresponding definitions in - .
it is clear that the main computations in ` gpf ` are the gaussian filterings in step @xmath65 ( and the initial filtering in step @xmath22 ) .
that is , the overall cost is dominated by the cost of computing @xmath66 gaussian filterings . in this regard ,
we note that for the same degree @xmath29 , the number of gaussian filterings required in @xcite is @xmath67 .
indeed , we will see in section [ sec : results ] for a fixed @xmath29 , the overall run - time of ` gpf ` is about a third of that of @xcite .
finally , we note that ` gpf ` involves the evaluation of a transcendental function just once in step 4 .
thus , ` gpf ` is better suited to hardware implementation @xcite compared to the algorithm in @xcite which involves the repeated evaluation of cosine and sine functions .
we now present some results concerning the accuracy and run - time of the proposed ` gpf ` algorithm . in particular , we compare it with some of the fast algorithms @xcite . the experiments were performed using matlab on an intel quad - core 2.7 ghz machine with 8 gb memory .
we implemented the gaussian filtering in ` gpf ` and @xcite using deriche s constant - time algorithm @xcite .
the average run - times of the various fast algorithms are reported in table [ table ] for a @xmath68 .
we do not redundantly report the run - times for different image sizes , since this can roughly be estimated from the run - times in table [ table ] ( the algorithms scale linearly with the number of pixels ) .
notice that the run time of ` gpf ` and @xcite does not change appreciably with @xmath12 .
to evaluate the accuracy , we also report the mean - squared - error ( mse ) between the exact implementation of and the result obtained using the fast algorithms in table [ table ] .
in particular , the mse between two images @xmath69 and @xmath70 is defined to be @xmath71 db , where @xmath72 .
notice that ` gpf ` is competitive with the existing algorithms in terms of accuracy and run - time .
in particular , ` gpf ` has the smallest run - time , and its mse is in general better than the rest of the algorithms and comparable to that of the raised - cosine - based approximation in @xcite .
the degree of the raised - cosine and the gauss - polynomial filter is @xmath73 for all the experiments ( this gives a good tradeoff between accuracy and run - time ) . in this regard ,
an open question is how the accuracy of ` gpf ` varies with the degree , as a function of @xmath12 and @xmath74 .
this will be addressed in future work .
note that the run - time of the polynomial approximation in @xcite is almost identical to that of the proposed algorithm and is hence not reported . for a visual comparison ,
we report the result obtain on the _ peppers _ image in figure [ outputimages ] .
notice the visible distortions in subfigure ( d ) which arises on account of the poor approximation of the gaussian kernel using taylor polynomials .
l*6cr + & @xmath59 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 + exact & 1.5s & 3.2s & 5.3s & 8.4s & 32.5s & 73.2s + @xcite & 93ms & 134ms & 191ms & 261ms & 847ms & 1.92s + @xcite & 112ms & 118ms & 115ms & 116ms & 118ms & 120ms + @xcite & 210ms & 215ms & 220ms & 225ms & 230ms & 250ms + ` gpf ` & 74 ms & 82ms & 88ms & 89ms & 95ms & 98ms + + @xcite & 5.9 & 7.8 & 9.1 & 9.8 & 12.2 & 13.1 + @xcite & -3.3 & -1.1 & 0.5 & 1.8 & 6.2 & 9.2 + @xcite & -10.5 & -6.4 & -3.8 & -1.7 & 4.4 & 7.8 + ` gpf ` & -9.6 & -5.6 & -3.1 & -1.1 & 5.1 & 8.4 + [ table ] + +
we presented a fast algorithm for bilateral filtering based on gauss - polynomial decompositions of the translations of the range kernel .
we presented some preliminary results which demonstrated the accuracy and speed of the algorithm in comparison to some of the existing fast algorithms for bilateral filtering .
in particular , we saw that the algorithm is comparable to the one in @xcite , but has a smaller run time ( about a third ) .
moreover , as remarked in the introduction , the proposed algorithm has an edge over @xcite in the context of hardware implementation it is based on polynomials and does not involve the computation of multiple trigonometric functions @xcite .
we note that the algorithm has a direct extension to other variants of the bilateral filter including the joint and guided filter @xcite , and can also be extended for handling volume and video data @xcite .
g. petschnigg , r. szeliski , m. agrawala , m. cohen , h. hoppe , and k. toyama , `` digital photography with flash and no - flash image pairs , '' _ acm transactions on graphics _ , vol .
664 - 672 , 2004 . | the bilateral filter is a versatile non - linear filter that has found diverse applications in image processing , computer vision , computer graphics , and computational photography .
a common form of the filter is the gaussian bilateral filter in which both the spatial and range kernels are gaussian .
a direct implementation of this filter requires @xmath0 operations per pixel , where @xmath1 is the standard deviation of the spatial gaussian . in this paper
, we propose an accurate approximation algorithm that can cut down the computational complexity to @xmath2 per pixel for any arbitrary @xmath1 ( constant - time implementation ) .
this is based on the observation that the range kernel operates via the translations of a fixed gaussian over the range space , and that these translated gaussians can be accurately approximated using the so - called gauss - polynomials .
the overall algorithm emerging from this approximation involves a series of spatial gaussian filtering , which can be efficiently implemented ( in parallel ) using separability and recursion .
we present some preliminary results to demonstrate that the proposed algorithm compares favorably with some of the existing fast algorithms in terms of speed and accuracy . bilateral filter , approximation , gauss - polynomial , convolution , fast algorithm . |
since their introduction by atiyah @xcite , segal @xcite and witten @xcite , topological quantum field theories ( tqfts ) have been studied intensely using a wide range of techniques . the first construction in 2 + 1 dimensions was given by reshetikhin and turaev @xcite using representation theory of quantum groups at roots of unity to construct link invariants and in turn derive invariants of 3-manifolds through surgery and kirby calculus .
shortly thereafter , a combinatorial construction was given by blanchet , habegger , masbaum and vogel @xcite in the language of skein theory .
a geometric realization was proposed by witten @xcite , suggesting the use of quantum chern - simons theory or conformal field theory to construct the 2-dimensional part .
the gauge theoretic approach was studied independently by axelrod , della pietra and witten @xcite and hitchin @xcite , proving that the quantum spaces arising from geometric quantization of chern - simons theory for compact gauge group are indeed independent of the conformal structure on the surface , in the sense that they are identified by parallel transport of a projectively flat connection over the teichmller space of the surface .
these constructions have been expressed and generalized in purely differential geometric terms in @xcite and @xcite , and we shall be mainly concerned with this description in the present paper . the other construction proposed by witten , through conformal field theory , was provided by tsuchiya , ueno and yamada @xcite , and the link to the gauge theoretic construction was established by laszlo @xcite .
only recently has the relation to reshetikhin and turaev been fully demonstrated . in a series of papers @xcite , the first author of this paper and ueno obtain a modular functor , from a twist of the conformal field theory construction , and identify it with the modular functor constructed from skein theory , and hence with the original construction of reshetikhin and turaev
this has paved the way for studying the tqft through geometric quantization of moduli space and in particular the application of toeplitz operator theory , see e.g. @xcite . for non - compact gauge group ,
the situation is very different . in the paper @xcite
, witten initiated the study of chern - simons topological quantum field theory for complex gauge groups from a physical point of view .
he proposed that the smooth sections of the chern - simons line bundle over the moduli space of flat connections in the corresponding compact real form of the group should be the appropriate pre - hilbert space of this theory .
he reduced the description to this model space by considering a real polarization , which we review in section [ qcst ] , on the space of connections with values in the complex gauge group .
although this model space itself does not depend on a choice of complex structure on the surface , the polarization , and hence the interpretation as the quantum space , does .
witten argued that the needed infinitesimal change of polarization , under infinitesimal change of the complex structure on the surface , can be encoded as a connection in the trivial bundle with the fixed model space as fiber .
furthermore , witten provided infinite - dimensional gauge theory arguments for the projective flatness of this connection . in this paper , we review the general differential geometric construction of the hitchin connection for a rigid family of complex structures on a symplectic manifold with vanishing first betti number and first chern class represented essentially by the symplectic form . inspired by witten s considerations
, we also consider a certain one - parameter family of connections on the prequantum spaces . for any value of the parameter
, the resulting connection will be called the hitchin - witten connection .
the main feature of both connections is projective flatness , at least when the khler structures have few symmetries , and we shall establish this fact by direct curvature calculations .
let us briefly introduce the setting and state the main results .
the basic assumptions and their implications will be explored in greater detail in the following sections .
consider a symplectic manifold @xmath0 , with vanishing first betti number and first chern class given by @xmath1 $ ] , for some integer @xmath2 .
futhermore , let @xmath3 be a complex manifold parametrizing a holomorphic family @xmath4 of integrable almost complex structures on @xmath5 , none of which admit non - constant holomorphic functions on @xmath6 .
the variation of the khler structure , along the holomorphic part of a vector field on @xmath7 on @xmath3 , is encoded by a section @xmath8 , defined by @xmath9 = g(v ) { \smash{\cdot}}\omega$ ] , of the second symmetric power of the holomorphic tangent bundle , and we will assume that the family @xmath10 is _ rigid _ in the sense that the bivector field @xmath11 defines a holomorphic section @xmath12 . by the assumption on the chern class , the symplectic manifold @xmath5 admits a hermitian line bundle @xmath13 with a compatible connection of curvature @xmath14 .
for any @xmath15 , the space @xmath16 of holomorphic setions is the quantum space , at level @xmath17 , arising from geometric quantization using the complex structure @xmath18 .
these spaces sit inside the prequantum space @xmath19 of smooth sections , and as @xmath20 varies , they form a sub - bundle of the trivial bundle @xmath21 . as proved in @xcite , this sub - bundle is preserved by the explicitly given connection , @xmath22 ) + v'[f],\end{aligned}\ ] ] where @xmath23 is a second - order operator with symbol @xmath11 , and @xmath24 is the ricci potential , expressing the relation between the ricci form @xmath25 and the symplectic form through @xmath26 .
this connection will be called the hitchin connection , and generalizes the connections studied by @xcite and @xcite in the setting of quantum chern - simons theory for compact gauge group . inspired by the work of witten @xcite , we may also use the family of khler structures to define another connection on @xmath27 by the expression , @xmath28 ) - \frac{1}{2\bar t } ( \delta_{\bar g(v ) } \ ! + \!2 \nabla _ { \bar g(v ) { \smash{\cdot}}df } \!- \ ! 2{\lambda}v''[f ] ) \!+ \ !
v[f],\end{aligned}\ ] ] for any complex number @xmath29 with real part equal to @xmath30 .
we shall refer to this as the hitchin - witten connection at level @xmath30 .
of course it also depends on the imaginary part of @xmath31 , so for each level @xmath30 we get family of theories , parametrized by one real parameter . unlike the hitchin connection
, the hitchin - witten connection will not preserve the sub - bundle of quantum spaces .
both connections do , however , share another meritorious feature , captured by the main theorem of the paper .
[ thm : mt ] if all complex structures in the family have zero - dimensional symmetry group , then both the hitchin and hitchin - witten connections are projectively flat .
since the moduli spaces have no holomorphic vector fields , which is equivalent to the condition stated in the theorem , we get the following immediate corollary [ thm : mtcs ] the hitchin and hitchin - witten connections for the moduli spaces of flat connections on a closed oriented surface are projectively flat .
this result on the hitchin connection , for the moduli spaces of flat @xmath32 connections , is due to hitchin , but his proof uses algebraic geometric properties of these moduli spaces . as mentioned above , witten gave an infinite - dimensional gauge - theoretic argument for this result on the hitchin - witten connection , for the same moduli spaces . in this paper , we provide a purely differential geometric finite - dimensional argument for projective flatness , which applies in the more general setting we have described . as explained above , the quantum representation of the mapping class groups from the reshetikhin - turaev tqft for @xmath32 coincides with the representation obtained from the hitchin connection .
we expect that the same will hold for the hitchin - witten connection and quantum chern - simons theory for the complex gauge group @xmath33 . in the paper @xcite ,
the first author has computed explicitly the resulting representation of the mapping class group for genus @xmath34 , the gauge group @xmath35 and all integer levels @xmath30 .
only recently has the whole tqft been rigorously constructed by the first author of this paper and kashaev in @xcite and @xcite for the case of @xmath36 and level @xmath37 , using quantum teichmller theory and the faddeev quantum dilogarithm . in the paper @xcite ,
the same authors have constructed quantum chern - simons theory for @xmath36 and all non - negative integer levels @xmath30 and further understood how it relates to the geometric quantization of the @xmath36-moduli spaces .
in fact , they have proposed a very general scheme which just requires a pontryagin self - dual locally compact group , which we expect will lead to the construction of the @xmath33 for all non - negative integer levels @xmath30 .
this should be seen in parallel to the developments on indeces @xcite , which should be related to the level @xmath38 theory . in the physics literature
, the complex quantum chern - simons theory has been discussed from a path integral point of view in a number of papers @xcite and latest by dimofte @xcite using the more advanced 3d-3d correspondence .
let us briefly outline the organization of the paper .
section [ cha : famil - kahl - struct ] discusses general aspects of families of khler structures on a fixed symplectic manifold .
in particular , a rather serious holomorphicity condition , called rigidity ( ) , on the infinitesimal deformation of the khler structure will be discussed .
we derive a symmetry result ( ) for certain tensor fields associated with such families , which will prove crucial in the calculation of the curvature of the hitchin connection . in section [ cha : geom - quant ]
, we briefly recall the basics of geometric quantization , described as a two stage process where the prequantum space @xmath39 is first constructed as sections of a line bundle , and the quantum space @xmath40 is then defined as the subspace of polarized sections with respect to some choice of auxiliary khler polarization .
one way of understanding the influence of this choice on quantization is by studying the infinitesimal behaviour of a family of such polarizations , and this is exactly the approach employed with the hitchin connection , which relates the quantum spaces through parallel transport .
following the initial discussion , we calculate the commutators of general second - order differential operators acting on the prequantum spaces .
these results will be useful when calculating the curvature of the hitchin connection .
the hitchin connection is the subject of section [ cha : hitchin - connection ] , which contains the main results .
after briefly reviewing the differential geometric construction of the hitchin connection , we move on to the straightforward but rather lengthy calculation of its curvature , culminating with the first major result in .
the fact that the curvature acts as a differential operator of order at most one is a crucial point .
building on these computations and inspiration from witten s work on quantum chern - simons theory for complex gauge group @xcite , we then consider the hitchin - witten connection defined in as well as above .
the second major result is the calculation of its curvature in .
the expressions turn out to be essentially equal to the curvature of the hitchin connection in , and in particular it acts as a differential operator of order at most one .
this property , shared by both connections , entails projective flatness if , for instance , the family of complex structures does not admit holomorphic vector fields .
this is the content of , leading ultimately to .
the final section applies the results to quantum chern - simons theory , both for compact and complex gauge groups .
the hitchin connection was originally studied in this setting , with compact gauge group @xmath32 , by hitchin @xcite and axelrod , della pietra and .
our results provide another proof of projective flatness , which still relies on the absence of holomorphic vector fields , but uses general properties of rigid families of khler structures to prove the vanishing of higher - order symbols .
as mentioned , the case of complex gauge group was studied by witten in @xcite , where he used a real polarization to reduce the quantum space from complex to unitary connections .
the real polarization , and hence the reduction , depends on the conformal structure on the surface , and witten derived a formula for an analogue of the hitchin connection in this model , arriving at exactly the expression .
we recall the necessary theory and connect it with the results of previous sections , providing a differential geometric and purely finite - dimensional construction of a projectively flat connection , the hitchin - witten connection , in witten s model of quantum chern - simons theory with complex gauge group .
before we recall the construction of the hitchin connection and calculate its curvature , we will explore the properties of families of khler structures on a symplectic manifold .
such families are central to the notion of a hitchin connection , and the results obtained will play a fundamental role in subsequent parts .
the section serves to introduce notation , establish conventions and provide a number of basic results for later reference .
the result is somewhat lengthy and can be read swiftly on first reading .
let @xmath41 be a symplectic manifold .
if @xmath3 is a manifold , we say that a smooth map , @xmath42 is a family of khler structures on @xmath41 if it defines an integrable and @xmath43-compatible almost complex structure for every point @xmath15 .
smoothness of @xmath10 means that it defines a smooth section of the pullback bundle @xmath44 over @xmath45 , where @xmath46 denotes the projection .
for any point @xmath15 , the almost complex structure @xmath18 induces a splitting , @xmath47 of the complexified tangent bundle of @xmath6 into the two eigenspaces of @xmath18 , with associated subspace projections @xmath48 and @xmath49 , explicitly given by @xmath50 we denote by @xmath51 the associated splitting of a vector field @xmath52 on @xmath53 . in general
, the subscript @xmath54 indicates dependence on the complex structure , but we shall typically ommit it when the dependence is obvious and the formula is valid for any point in @xmath3 .
the khler metric associated with the complex structure @xmath10 is given by @xmath55 where the dot denotes contraction of tensors .
the inverses of @xmath56 and @xmath43 are denoted by @xmath57 and @xmath58 , respectively , and they are the unique symmetric , respectively anti - symmetric , bivector fields satisfying @xmath59 as above , a dot will be used to denote contraction of tensors , and the placement of the tensors relative to each other indicates which entries to contract .
we will , however , also encounter more complicated expressions , where the entries to be contracted can not be indicated by simply placing the tensors next to each other . in such cases , we will use abstract indices to denote the entries of each tensor .
the indices only name the entries of a tensor , and do not represent a choice of local coordinates . as usual ,
subscript indices refer to covariant entries of a tensor , whereas superscript indices refer to contravariant entries , and following the einstein convention , repeated indices will indicate contraction . if the two contracted indices are both either subscript or superscript , the khler metric is used for contraction . with these conventions ,
the above identities become @xmath60 in a few places , we will also need to apply the projections @xmath61 and @xmath62 to the entries of the tensor . in the index notation ,
composition with @xmath61 will be indicated by a prime on the index , whereas composition with @xmath62 will be indicated by two primes . as an example we can write @xmath63 for the symplectic form , which is of type ( 1,1 ) with respect to a compatible complex structure .
associated with the khler metric @xmath56 , we have the levi - civita connection @xmath64 , and as usual its curvature is defined by @xmath65 } z,\end{aligned}\ ] ] for any vector fields @xmath66 and @xmath67 on @xmath6 . in abstract index notation , the curvature writes @xmath68 , and we can use the metric to lower the upper index and get the curvature tensor @xmath69 the ricci curvature is the symmetric @xmath10-invariant tensor @xmath70 defined by @xmath71 and its corresponding skew - symmetric two - form is the ricci form @xmath72 , which would correspond to @xmath73 in index notation .
finally , the scalar curvature @xmath74 is the metric trace of the ricci curvature @xmath75 if @xmath76 is a vector bundle over @xmath6 and @xmath77 is a differential operator of order at most @xmath78 , we can assign the principal symbol @xmath79 , which is a symmetric section of the @xmath78th tensor power of the tangent bundle . if the principal symbol vanishes , then @xmath80 is of order at most @xmath81 . in general
, there is no good notion of lower order symbols of differential operators , but a connection on @xmath76 and can be combined with the levi - civita connection on @xmath6 to define symbols of all orders . for vector fields
@xmath82 on @xmath6 , we consider the inductively defined differential operator on sections of @xmath76 , @xmath83 with the obvious induction start given by the covariant derivative .
it is easily verified that this expression is tensorial in the vector fields , so we get a map @xmath84 for any tensor field @xmath85 , the symbol of @xmath86 is given by the symmetrization @xmath87 of @xmath88 . if @xmath77 is an operator , of order at most @xmath78 , with principal symbol @xmath89 , then the operator @xmath90 is of order at most @xmath81 , since its principal symbol vanishes .
inductively , it follows that the operator @xmath80 can be written uniquely in the form @xmath91 where @xmath92 is called the _ symbol of order @xmath93 _ and gives rise to a map @xmath94 any finite order differential operator on @xmath76 is uniquely determined by the values of these symbol maps . in fact , through the expression , a choice of symbols specifies a differential operator on any vector bundle with connection , and in particular on functions .
we will also need the notion of divergence of vector fields and more general contravariant tensors .
recall that the divergence @xmath95 of a vector field @xmath52 on @xmath6 is defined in terms of the lie derivative and volume form by the equation @xmath96 .
although the divergence of a vector field only depends on the symplectic volume , and not on the khler metric itself , a simple computation reveals that the divergence can be calculated using the levi - civita connection by the formula @xmath97 in which the independence of the khler structure is perhaps not so evident .
the laplace - de rham operator on functions can be expressed in terms of the divergence by @xmath98 where @xmath99 denotes the ( 1,0)-part of the hamiltonian vector field associated with the function @xmath100 .
the formula generalizes to tensors of higher degree . for vector fields @xmath82 om @xmath6 , we define @xmath101 this defines a map @xmath102 , also called the divergence , which does depend on the khler structure
. the generalization of divergence to sections of the endomorphism bundle of the tangent bundle will also be convenient .
if @xmath103 is a one - form and @xmath52 is a vector field , we define @xmath104 which gives a map @xmath105 .
finally , for any bivector field @xmath106 , we introduce the second - order differential operator , @xmath107 which will appear repeatedly throughout the paper . for a smooth family of khler structures
, we can take its derivative along a vector field @xmath7 on @xmath3 to obtain a map @xmath108 \colon \mathcal{t } \to c^\infty(m , \operatorname{end}(tm)).\end{aligned}\ ] ] differentiating the identity @xmath109 , we see that @xmath110 $ ] and @xmath10 anti - commute , @xmath111j + j v[j ] = 0,\end{aligned}\ ] ] so @xmath110_\sigma$ ] interchanges types on the khler manifold @xmath53 .
therefore , it splits as @xmath112 = v[j ] ' + v[j]'',\end{aligned}\ ] ] where @xmath110'_\sigma \in c^\infty(m , t'\!m_\sigma { \varotimes}t''\!m^*_\sigma)$ ] and @xmath110''_\sigma \in c^\infty(m , t''\!m_\sigma { \varotimes}t'\!m^*_\sigma)$ ] is its conjugate . notice that this splitting occurs for any infinitesimal deformation of an almost complex structure on @xmath6 and defines an almost complex structure on the space of almost complex structures on @xmath6 .
differentiating the integrability condition on @xmath10 , expressed through the vanishing of the nijenhuis tensor , reveals that @xmath110 ' \in \omega^{0,1}(m , t'\!m)$ ] satisfies the holomorphicity condition @xmath113 ' = 0 $ ] , and the associated cohomology class in @xmath114 is the kodaira - spencer class of the deformation ( see @xcite ) .
define a bivector field @xmath115 by the relation @xmath108 = \tilde g(v ) { \smash{\cdot}}\omega,\end{aligned}\ ] ] for any vector field @xmath7 on @xmath3 .
differentiating the identity @xmath116 along @xmath7 , we get @xmath117 = - v[j ] { \smash{\cdot}}\tilde \omega = - \tilde g(v),\end{aligned}\ ] ] and since @xmath57 is symmetric , this implies that @xmath118 is a symmetric bivector field .
furthermore , the combined types of @xmath110 $ ] and @xmath43 yield a decomposition , @xmath119 where @xmath120 and @xmath121 . in other words
, the real symmetric bivector field @xmath118 has no ( 1,1)-part .
the variation of the khler metric is obtained by differentiating the identity @xmath122 , which yields @xmath123 = \omega { \smash{\cdot}}v[j ] = \omega { \smash{\cdot}}\tilde g(v ) { \smash{\cdot}}\omega = g { \smash{\cdot}}\tilde g(v ) { \smash{\cdot}}g.\end{aligned}\ ] ] we shall also need the variation of the levi - civita connection , which is the tensor field @xmath124 \in c^\infty(m , s^2(tm^ * ) { \varotimes}tm)\end{aligned}\ ] ] given by ( see @xcite theorem 1.174 ) @xmath125_{x } y , z ) = \nabla_{x } ( v[g])(y , z ) + \nabla_{y } ( v[g])(x , z ) - \nabla_{z } ( v[g ] ) ( x , y),\end{aligned}\ ] ] for any vector fields @xmath66 and @xmath67 on @xmath6 . in index notation
, this translates to @xmath126_{ab}^c = \nabla_a \tilde g(v)^{cu } g_{ub } + g_{au } \nabla_b \tilde g(v)^{uc } - g_{au } \tilde g^{cw } \nabla_w \tilde g(v)^{uv } g_{vb},\end{aligned}\ ] ] and we remark that the trace @xmath127_{xb}^x$ ] of this tensor vanishes .
indeed , we get that @xmath128_{xb}^x = \nabla_x \tilde g(v)^{xu } g_{ub } + g_{xu } \nabla_b \tilde g(v)^{ux } - g_{xu } \tilde g^{xw } \nabla_w \tilde g(v)^{uv } g_{vb } = 0,\end{aligned}\ ] ] where the first and last term cancel , and the middle term vanishes because @xmath118 has no part of type ( 1,1 ) , which is the type of the metric .
we will also need to know the variation of the ricci curvature , which will result from the bianchi identity of a certain line bundle associated with the family of complex structures . for a family @xmath10 of khler structures
, we can consider the vector bundle , @xmath129 with fibers @xmath130 given by the holomorphic tangent spaces of @xmath6 . throughout the paper
, we shall generally use a hat in the notation to indicate that we are working over the product @xmath45 . following this convention , the exterior differential on @xmath45
is denoted by @xmath131 , whereas the differential on @xmath3 is denoted by @xmath132 and by @xmath93 on @xmath6 .
the khler metric induces a hermitian structure @xmath133 on @xmath134 , and the levi - civita connection gives a compatible partial connection along the directions of @xmath6 .
we can extend this partial connection to a full connection @xmath135 on @xmath136 in the following way .
if @xmath137 is a smooth family of sections of the holomorphic tangent bundle , and @xmath7 is a vector field on @xmath3 , then we define @xmath138.\end{aligned}\ ] ] in other words , we regard @xmath67 as a smooth family of sections of the complexified tangent bundle @xmath139 , and then we simply differentiate @xmath67 along @xmath7 in this bundle , which does not depend on the point in @xmath3 , and project the result back onto the holomorphic tangent bundle
. clearly , the connection @xmath135 preserves the hermitian structure in the directions of @xmath6 , since it is induced by the levi - civita connection .
moreover , if @xmath7 is a vector field on @xmath3 , and @xmath52 and @xmath140 are sections of @xmath134 , we get that @xmath141 = v[g(x,\bar y ) ] & = v[g](x , \bar y ) + g(v[x ] , \bar y ) + g(x , \overline{v[y ] } ) \\ & = h(\hat \nabla_v x , y ) + h(x , \hat \nabla_v y),\end{aligned}\ ] ] since the @xmath142-part of @xmath143 $ ] vanishes .
it follows that @xmath144 preserves the hermitian structure on @xmath134 .
now consider the line bundle @xmath145 which will be referred to as the canonical line bundle of the family of khler structures .
as usual , the hermitian structure and connection on @xmath134 induce a hermitian structure @xmath146 and a compatible connection @xmath147 on @xmath148 .
the curvature of @xmath147 was calculated in @xcite and will be recalled below , but before stating it , we introduce the following important notation . for any vector fields @xmath7 and @xmath149 on @xmath3 , we define @xmath150 by @xmath151 where @xmath152 denotes symmetrization
. we also give a name to the metric trace of the symmetric bivector field @xmath153 and define @xmath154 clearly , this defines a real two - form @xmath155 on @xmath3 with values in smooth functions on @xmath6 .
we note that the two - form @xmath156 over @xmath3 is exact . to see this
, we take the variation of @xmath157 along @xmath149 to get @xmath158 & = i g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) - i \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) - \pi^{2,0 } ( wv[\tilde g ] ) \\ & = 2i \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) - \pi^{2,0 } ( wv[\tilde g ] ) , \end{aligned}\end{aligned}\ ] ] which in turn shows that @xmath159 - w[g(v ) ] & = -i \operatorname{\mathcal{s}}(\bar g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}g(w ) ) - i\operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) = -i \theta ( v , w),\end{aligned}\ ] ] for commuting vector fields @xmath7 and @xmath149 on @xmath3 .
this can be rephrased as @xmath160 where @xmath161 is viewed as a one - form in @xmath162 .
the following proposition gives the curvature of the canonical line bundle of a family of khler structures and is proved in @xcite .
[ prop:6 ] the curvature of @xmath147 is given by @xmath163 for any vector fields @xmath66 on @xmath6 and @xmath164 on @xmath3 . by applying the bianchi identity to the connection @xmath147 , and using the formulas of
, we get three useful results .
the first is the fact that the two - form @xmath155 is closed , which is a trivial reformulation of the bianchi identity for three vector fields on @xmath3 . by applying the bianchi identity to two vector fields @xmath7 and @xmath149 on @xmath3 , and one vector field on @xmath6
, we get @xmath165 { \smash{\cdot}}\omega - \frac{1}{2 } v[\delta \tilde g(w ) ] { \smash{\cdot}}\omega}.\end{aligned}\ ] ] finally , the bianchi identity for two vector fields on @xmath6 and one on @xmath3 gives following important formula for the variation of the ricci form , @xmath166 = \smash{\frac{1}{2 } d ( \delta\tilde g(v ) { \smash{\cdot}}\omega),}\end{aligned}\ ] ] for any vector field @xmath7 on @xmath3 . as an immediate consequence of , we get the following simple formula for the variation of the scalar curvature @xmath167 & = v[\rho]_{uv } \tilde \omega^{vu } = \frac{1}{2 } \nabla_u ( \delta \tilde g(v))^w \omega_{wv } \tilde \omega^{vu } - \frac{1}{2 } \nabla_v ( \delta \tilde g(v))^w \omega_{wu } \tilde \omega^{vu}= \delta \delta \tilde
g(v ) , \end{aligned}\end{aligned}\ ] ] for any vector field @xmath7 on @xmath3 . in case
the manifold @xmath3 is itself a complex manifold , we can require the family @xmath10 to be a holomorphic map from @xmath3 to the space of complex structures .
this is made precise by the following definition , which uses the splitting of @xmath110 $ ] .
[ def:3 ] suppose that @xmath3 is a complex manifold , and that @xmath10 is a family of complex structures on @xmath6 , parametrized by @xmath3 .
then @xmath10 is _ holomorphic _ if @xmath168 = v[j ] ' \qquad \text{and } \qquad v''[j ] = v[j ] '' , \end{aligned}\ ] ] for any vector field @xmath7 on @xmath3 .
if @xmath169 denotes the integrable almost complex structure on @xmath3 induced by its complex structure , then we get an almost complex structure @xmath170 on @xmath171 defined by @xmath172 the following proposition gives another characterization of holomorphic families @xcite .
[ prop:9 ] the family @xmath10 is holomorphic if and only if @xmath170 is integrable . by this proposition , a holomorphic family induces a complex structure on the product manifold @xmath45 .
clearly , the projection @xmath173 is a holomorphic map , and its differential is the projection @xmath174 , where @xmath175 is the pullback of @xmath176 by @xmath177 .
since the bundle @xmath134 over @xmath45 is the kernel of this map , it has the structure of a holomorphic vector bundle , and it is easily verified that the connection @xmath135 is compatible with this holomorphic structure . since the connection also preserves the hermitian structure , it must be the chern connection .
holomorphicity has several useful consequences .
first of all , it implies that @xmath178 { \smash{\cdot}}\tilde \omega = v[j ] ' { \smash{\cdot}}\tilde \omega = g(v),\end{aligned}\ ] ] and similarly @xmath179 .
this means that @xmath161 , viewed as a one - form over the complex manifold @xmath3 , has type ( 1,0 ) and that @xmath156 and @xmath180 have type ( 1,1 ) .
these important facts will be used without reference going forward .
finally , for commuting vector fields @xmath181 and @xmath182 , the identity reduces to @xmath183 = \frac{i}{2 } g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) - \frac{i}{2 } \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) = i \theta(v',w'').\end{aligned}\ ] ] this expression for the second - order variation of the complex structure will prove very useful in later calculations , but we emphasize the fact that it only holds for commuting vector fields .
the following rather serious assumption on a family of khler structures turns out to be crucial to the construction of the hitchin connection as well as the calculation of its curvature .
[ def:4 ] a family of khler structures is called _ rigid _ if @xmath184 for all vector fields @xmath7 on @xmath3 and @xmath52 on @xmath6 . in other words , the family @xmath10 is rigid if @xmath11 is a holomorphic section of @xmath185 , for any vector field @xmath7 on @xmath3 . for examples of rigid families in a basic setting ,
we refer to @xcite . by differentiating the rigidity condition along @xmath3
, we get the following crucial result .
[ prop:14 ] any rigid family of khler structures satisfies the symmetry property @xmath186 for any vector fields @xmath7 and @xmath149 on @xmath3 . throughout the proof ,
let @xmath7 and @xmath149 be commuting vector fields on @xmath3 . by differentiaing the holomorphicity condition on the bivector field @xmath11 along @xmath149 ,
we obtain @xmath187 \\ & = w[\pi^{0,1}]^u_a\nabla_{u } g(v)^{bc } + \nabla_{a '' } w[g(v)]^{bc } \\ & \qquad + w[\nabla]^b_{a''u } g(v)^{uc } + w[\nabla]^c_{a''u } g(v)^{bu}. \end{aligned } \end{aligned}\ ] ] let us work out each of these terms individually .
using rigidity , the first term reduces to @xmath188^u_a\nabla_{u } g(v)^{bc } = i \tilde g(w)^{uv } \omega_{va } \nabla_u g(v)^{bc } = - g_{av } g(w)^{uv } \nabla_u g(v)^{bc}. \end{aligned}\ ] ] for the second term of , we simply apply and rigidity to obtain @xmath189^{bc } ) \\ & = - \nabla_a ( \bar g(w)^{bu } ) g_{uv } g(v)^{vc } - g(v)^{bu } g_{uv } \nabla_a ( \bar g(w)^{vc } ) - \nabla_{a '' } ( ( \pi^{2,0})^{bc}_{uv } wv[\tilde g^{uv } ] ) .
\end{aligned}\ ] ] finally , for the last two terms of , we can apply in combination with rigidity of the family to obtain @xmath190^b_{a''u } g(v)^{uc } = g_{av}\nabla_u ( g(w)^{vb } ) g(v)^{uc } + \nabla_a ( \bar g(w)^{bv } ) g_{vu } g(v)^{uc}. \end{aligned}\ ] ] substituting these expressions into , and cancelling terms , we get that @xmath191 ) .
\end{aligned}\ ] ] raising an index , this implies that @xmath192 ) , \end{aligned}\ ] ] and clearly this is symmetric in @xmath7 and @xmath149 because these were chosen to commute .
this proves the proposition . repeatedly taking the divergence of the identity
yields similar identities for contravariant tensors of lower degree
. these will play a crucial role in the calculation of the curvature of the hitchin connection , so we introduce special notation for them . inspired by ,
we define a two - form @xmath193 on @xmath3 , with values in sections of the third symmetric power of the holomorphic tangent bundle over @xmath6 , by @xmath194 for any vector fields @xmath7 and @xmath149 on @xmath3 .
then ensures that @xmath193 actually defines a symmetric two - form on @xmath3 .
clearly , the symmetry of the two - form over @xmath3 is not affected by taking the divergence of the tri - vector field part of @xmath193 over @xmath6 .
in other words , @xmath195 defines a symmetric two - form on @xmath3 with values in the second symmetric power of the holomorphic tangent bundle on @xmath6 .
it is given by @xmath196 for any vector fields @xmath7 and @xmath149 on @xmath3 .
in contrast to the first two terms of this expression , the last term is obviously symmetric in @xmath7 and @xmath149 .
this leads us to define @xmath197 by @xmath198 which is also a symmetric two - form on @xmath3 and encodes the interesting part of the symmetry statement for @xmath199 . repeatedly taking the divergence , and removing obviously symmetric parts
, we get the following important proposition [ prop:2 ] the four two - forms @xmath200 on @xmath3 , with values in symmetric contravariant tensors on @xmath6 , defined by @xmath201 are all symmetric in the vector fields @xmath7 and @xmath149 on @xmath3 .
suppose that the first chern class of @xmath41 is represented by the symplectic form , that is @xmath202,\end{aligned}\ ] ] for some integer @xmath203 .
now the first chern class is also represented by the ricci form @xmath204 , so the difference between the forms @xmath25 and @xmath205 is exact .
a smooth real function @xmath206 , which can be viewed as a smooth map @xmath207 , is called a _
family of ricci potentials _ if it satisfies @xmath208 for every point @xmath15 .
any two ricci potentials differ by a global pluriharmonic real function . if we assume that @xmath209 vanishes , such a function is globally the real part of a holomorphic function , so a family of ricci potential is then uniquely determined up to a function on @xmath3 if the family of khler structures does not allow any non - constant holomorphic functions on @xmath6 .
the existence of a ricci potential is clearly a global issue over @xmath6 , as the local @xmath210-lemma ensures local existence around any point on @xmath6 when holds .
if the manifold @xmath6 is compact , the global @xmath210-lemma from hodge theory ensures the existence of a ricci potential , using the fact that the khler form @xmath43 is harmonic . in this case , the ricci potential is certainly unique up to a function on @xmath3 , and by imposing zero average over @xmath6 , @xmath211 we can fix it uniquely .
the following proposition gives an important identity , involving the variation of a family of ricci potentials .
[ prop:10 ] suppose that @xmath6 is a symplectic manifold with @xmath212 and @xmath213 $ ] , and let @xmath10 be a holomorphic family of khler structures on @xmath6 , none of which admit non - constant holomorphic functions on @xmath6 .
then @xmath214 = \delta g(v ) { \smash{\cdot}}\omega + 2 df { \smash{\cdot}}g(v ) { \smash{\cdot}}\omega , \end{aligned}\ ] ] for any family of ricci potentials @xmath215 and any vector field @xmath7 on @xmath3 . by differentiating the identity in the direction of @xmath181 ,
we get @xmath216 & = - d ( df { \smash{\cdot}}g(v ) { \smash{\cdot}}\omega ) + 2i { \partial}\bar { \partial}v'[f ] , \end{aligned}\ ] ] and by using on the left - hand side , this yields @xmath217 = 0 . \end{aligned}\ ] ] on one hand , it follows that the one - form @xmath218 \end{aligned}\ ] ] is closed , and hence exact by the assumption @xmath212 . on the other hand ,
it is of type ( 0,1 ) , so it can not be exact unless it is zero , because we assumed that none of the khler structures admit non - constant holomorphic functions .
this proves the lemma .
another equivalent form of is the following , @xmath219 = \delta g(v ) { \smash{\cdot}}\omega,\end{aligned}\ ] ] and if we combine this with , for commuting vector fields @xmath181 and @xmath182 on @xmath3 , we get @xmath220 - v'[\delta \bar g(w ) { \smash{\cdot}}\omega ] \\ & = 4i w''v'[\bar { \partial}f ] - 4i v'w''[{\partial}f ] \\ & = 4i v'w''[d f ] \\ & = 4i { \partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f ( v',w'').\end{aligned}\ ] ] since the family is holomorphic , the form @xmath180 has type ( 1,1 ) on @xmath3 , so we have shown [ prop:1 ] in the setting of , any family of ricci potentials satisfies @xmath221 in other words , the form takes values in constant functions on @xmath6 .
this ends the general discussion of families of khler structures . in the next section ,
we discuss general aspects of geometric quantization , and in particular the need for a choice of auxiliary polarization in the construction .
understanding the effects of this choice naturally leads us to consider families of khler structures , and ultimately to the hitchin connection relating the different choices .
the results from this section will play a fundamental role in the discussion .
in very broad terms , geometric quantization concerns the passage from classical mechanics to quantum mechanics .
it aims to produce a hilbert space of quantum states from a classical phase space , in the form of a symplectic manifold , and a self - adjoined operator from a classical observable , in the form of a function on the classical phase space . in the following , we shall briefly review the basic notions from geometric quantization relevant to us .
a key role is played by an auxiliary choice of polarization , which is often chosen to be khler . for a broader treatment of geometric quantization ,
the reader is referred to @xcite and @xcite . after reviewing the elements of geometric quantization
, we calculate the commutators of certain differential operators acting on the prequantum spaces
. these will be relevant in later discussions of the hitchin connection and its curvature . in geometric quantization
, the hilbert space of quantum states arises as sections of a certain hermitian line bundle over the classical phase space . as a model for this classical phase space
, we consider a symplectic manifold @xmath41 of dimension @xmath222 .
a prequantum line bundle over the @xmath6 is a complex line bundle @xmath13 endowed with a hermitian metric @xmath223 and a compatible connection @xmath224 of curvature @xmath225 a symplectic manifold admitting a prequantum line bundle is called prequantizable .
evidently , this is not the case for every symplectic manifold .
indeed , the real first chern class of a prequantum line bundle is given by @xmath226 $ ] , leading us to the following necessary condition for prequantizability , called the prequantum condition , @xmath227 \in \operatorname{im}\big ( h^2(m , { { \mathbbm{z } } } ) \to h^2(m , { { \mathbbm{r } } } ) \big).\end{aligned}\ ] ] this is , in fact , also sufficient to ensure the existence of a prequantum line bundle , and the inequivalent prequantum line bundles over @xmath6 are parametrized by @xmath228 .
for any natural number @xmath30 , called the _ level _ , we consider the _ prequantum space _
, @xmath229 of smooth sections of the @xmath30th tensor power of the line bundle @xmath13 .
these sections play the role of wave functions in the quantum theory . if @xmath230 is a function on @xmath6 , the corresponding _ prequantum operator _ , acting on @xmath39 , is defined by @xmath231 where @xmath232 is the hamiltonian vector field of the function @xmath233 .
the virtue of is that the prequantum operators satisfy the correspondence principle @xmath234 } = \frac{i}{k } p_k(\{f , g\}),\end{aligned}\ ] ] which is one of the distinctive features of a viable quantization . from a physical perspective ,
the wave functions in @xmath39 depend on twice the number of variables they should . a standard way to remedy
this is to pick an auxiliary polarization on @xmath6 and consider the space of polarized sections of the line bundle
. the polarization can be by real or complex lagrangian subspaces . in the following ,
we will focus on the case of complex khler polarizations . from now on
, we assume that the symplectic manifold admits a khler structure , in the form of an integrable almost complex structure @xmath10 on @xmath6 which is compatible with the symplectic structure .
since the khler form @xmath43 has type @xmath142 , it follows that the ( 0,1)-part of the connection on the prequantum line bundle @xmath13 defines a holomorphic structure .
therefore , we can define the _ quantum space _ to be the space of holomorphic ( or polarized ) sections , @xmath235 which is a subspace of the prequantum space @xmath39 of smooth sections .
if the manifold @xmath6 is compact , then @xmath236 is a finite - dimensional space by standard theory of elliptic operators .
unfortunalely , the prequantum operators do not in general preserve the space of holomorphic sections .
a function @xmath230 is polarized if the ( 1,0)-part of its corresponding hamiltonian vector field is holomorphic .
in other words , the space of polarized functions is given by @xmath237 now , the operator @xmath238 preserves @xmath236 if and only @xmath233 is a polarized function . in fact
, it is easily verified that a first - order differential operator of the form @xmath239 on @xmath39 preserves the subspace @xmath240 if and only if @xmath233 is a polarized function and @xmath241 .
this gives another justification for the choice of prequantum operators .
in fact , the ( 0,1)-part of the vector field makes no difference to the action of @xmath238 on @xmath236 , so if we define a polarized variation of the prequantum operators by @xmath242 then these are essentially the only first - order operators with a chance of preserving the subspace @xmath236 of @xmath39 . for a real polarized function @xmath243 , the hamiltonian vector field @xmath232
must be killing for the khler metric , effectively reducing the quantizable observables to an at most finite - dimensional , and often trivial , space ( see @xcite ) . to get more quantizable observables ,
their quantization is modified in the following way .
the space @xmath240 is in fact a closed subspace of @xmath39 , and therefore we have the orthogonal projection @xmath244 . for @xmath230 , we then define the corresponding quantum operator by @xmath245 these operators do not form an algebra , but they satisfy a weaker form of ( at least if @xmath6 is compact ) in the sense that @xmath246 - \frac{i}{k } q_k(\{f , g\ } ) \big \rvert = o(k^{-2 } ) \quad \text{as } \quad k \to \infty,\end{aligned}\ ] ] with respect to the operator norm on @xmath236 .
the proof of relies on the fact that these operators are toeplitz operators ( see @xcite ) .
although this quantization scheme gives a hilbert space of the right size , it still fails to produce the right answers on basic examples from quantum mechanics . in the end , what really matters is the spectrum of the operators , and if the above procedure is applied to the one - dimensional harmonic oscillator , the quantization yields a spectrum which differs from the correct one by a shift . to deal with this problem
, the so - called metaplectic correction can be introduced .
we shall not pursue this direction further in the present paper , but refer the interested reader to @xcite , where the constrution of the hitchin connection in the metaplectic setting is discussed .
the description of the hitchin connection , and the calculation of its curvature in particular , requires the calculation of commutators of a number of differential operators on sections on the prequantum line bundle @xmath13 and its tensor powers . since the commutators of even second - order operators are quite complicated , it will be convenient to encode the operators through their symbols , as in , and have general but explicit descriptions of the symbols of the commutators in terms of the symbols of the commutants .
the followings lemmas give exactly such descriptions .
the very definition of the curvature of the line bundle @xmath247 implies the basic relation @xmath248s = \nabla_{[x , y ] } s - ik \omega(x , y ) s\end{aligned}\ ] ] for any vector fields @xmath249 on @xmath6 and any smooth section @xmath250 .
things become a little more complicated when second - order operators are introduced .
[ lem:7 ] for any khler structure on @xmath6 , any vector field @xmath52 on @xmath6 , and any symmetric bivector field @xmath251 , we have the symbols @xmath252 } & = 2 \operatorname{\mathcal{s}}(b { \smash{\cdot}}\nabla x ) - \nabla_x b\\ \sigma_1 { \big [ \nabla^2_b , \nabla_x \big ] } & = \nabla^2_b x - 2ik b { \smash{\cdot}}\omega { \smash{\cdot}}x + b^{uv } r^a_{wuv } x^w\\ \sigma_0 { \big [ \nabla^2_b , \nabla_x \big ] } & = -ik\ , \omega ( b { \smash{\cdot}}\nabla x ) \end{aligned}\ ] ] for the commutator of the operators @xmath253 and @xmath254 acting on @xmath39 . by straightforward calculation
, we get @xmath255 where we used the fact that the symplectic form @xmath43 is parallel with respect to the levi - civita connection . to get an expression for the desired commutator
, we subtract @xmath256 and the stated symbols can easily be extracted from the result .
naturally , things get even more complicated for two second - order operators .
[ lem:8 ] for any khler structure on @xmath6 and any symmetric bivector fields @xmath257 , we have the symbols @xmath258 } & = 2 \operatorname{\mathcal{s}}(a { \smash{\cdot}}\nabla b ) - 2 \operatorname{\mathcal{s}}(b { \smash{\cdot}}\nabla a ) \\[2pt ] \sigma_2 { \big [ \nabla^2_a , \nabla^2_b \big ] } & = \nabla^2_a b - \nabla^2_b a - 4ik \operatorname{\mathcal{s}}(a{\smash{\cdot}}\omega { \smash{\cdot}}b ) \\ & \qquad + 2 \operatorname{\mathcal{s}}(a^{xy } r^a_{uxy } b^{ub } ) - 2 \operatorname{\mathcal{s}}(b^{uv } r^a_{xuv } a^{xb } ) \\[2pt ] \sigma_1 { \big [ \nabla^2_a , \nabla^2_b \big ] } & = - 2i k a^{xy } \omega_{yu } \nabla_x ( b^{ua } ) + 2ik b^{uv } \omega_{vx } \nabla_u ( a^{xa } ) \\[2pt ] & \qquad - a^{xy } \nabla_x ( r^a_{yuv } ) b^{uv } + b^{uv } \nabla_v ( r^a_{uxy } ) a^{xy } \\[2pt ] & \qquad - \smash{\frac{4}{3 } } a^{xy } r^a_{xuv } \nabla_y b^{uv } + \smash{\frac{4}{3 } } b^{uv } r^a_{uxy } \nabla_v a^{xy } \\[2pt ] \sigma_0 { \big [ \nabla^2_a , \nabla^2_b \big ] } & = \frac{ik}{2 } a^{xy } j_y^j r_{xuvj } b^{uv } - \frac{ik}{2 } b^{uv } j_v^j r_{uxyj } a^{xy } \end{aligned}\ ] ] for the commutator of the operators @xmath259 and @xmath254 acting on @xmath39 .
once again , the proof proceeds by straightforward calculation .
we get @xmath260 focusing on the first term , we commute the indices @xmath261 and @xmath262 past @xmath263 and @xmath264 .
@xmath265 where we used the fact that @xmath43 is parallel , but otherwise just added the curvature terms .
expanding by the leibniz rule , using symmetries , and collecting terms , this can be rewritten as @xmath266 where the order of differentiation was interchanged for the term @xmath267 .
the second term of can be rewritten as @xmath268 analagous to , the other term of the commutator @xmath269}$ ] yields @xmath270 where the second term can be rewritten as @xmath271 by subtracting from , substituting and , and collecting terms by order of covariant differentiation , one verifies the claimed symbols .
in this section , we study the hitchin connection and calculate its curvature .
we start by recalling the differential geometric construction of the hitchin connection in geometric quantization .
the results concerning this construction are all proved in @xcite , to which the reader is referred for further details .
consider a symplectic manifold @xmath0 , equipped with a prequantum line bundle @xmath13 , and assume that @xmath272 and that the real first chern class of @xmath41 is given by @xmath273,\end{aligned}\ ] ] for some integer @xmath203 .
further , assume that @xmath6 is of khler type , and let @xmath10 be a rigid and holomorphic family of khler structures on @xmath0 , parametrized by some complex manifold @xmath3 .
finally , assume that the family of khler structures admits a family of ricci potentials @xmath215 , and that it does not admit any non - constant holomorphic functions on @xmath6 .
the prequantum space @xmath19 forms the fiber of a trivial , infinite - rank vector bundle over @xmath3 , @xmath274 if @xmath275 denotes the trivial connection on @xmath27 , we consider a connection of the form @xmath276 where @xmath277 is a one - form on @xmath3 with values in the space of differential operators on sections of @xmath247 , and we seek an @xmath278 for which the connection @xmath279 preserves the quantum subspaces @xmath280 of holomorphic sections inside each fiber of @xmath27 .
[ def:11 ] a _ hitchin connection _ on the bundle @xmath27 is a connection of the form which preserves the fiberwise subspaces @xmath40 .
it turns out that , with the assumptions made above , an explicit construction of a hitchin connetion can be given . for any vector field @xmath7 on @xmath3 , the operator @xmath281
is of order two , with principal symbol @xmath11 and lower order symbols given in terms of @xmath11 and the ricci potential . the precise statement
is contained in the following theorem from @xcite .
[ thm:58 ] let @xmath41 be a prequantizable symplectic manifold with @xmath212 and @xmath282 $ ] .
further , let @xmath10 be a rigid , holomorphic family of khler structures on @xmath6 , parametrized by a complex manifold @xmath3 , admitting a family of ricci potentials @xmath215 but no non - constant holomorphic functions on @xmath6 .
then the expression @xmath283 ) \end{aligned}\ ] ] defines a hitchin connection in the bundle @xmath284 over @xmath3 . the characterizing feature of the operator - valued one - form @xmath278 is the fact that it satisfies @xmath285 } s = - \smash{\frac{i}{2 } } \omega { \smash{\cdot}}g(v ) { \smash{\cdot}}\nabla s,\end{aligned}\ ] ] for any section @xmath74 of @xmath27 .
in fact , this property , and the fact that the hitchin connection preserves the quantum subspaces @xmath40 inside @xmath39 , implies that these subspaces form a bundle @xmath284 over @xmath3 , and this is part of the statement in .
having reviewed the explicit differential geometric construction of a hitchin connection in geometric quantization , we turn to the calculation of its curvature .
suppose the assumptions of are satisfied , ensuring the existence of the hitchin connection , and let us calculate its curvature .
for this calculation , it will be convenient to rewrite the hitchin connection slightly as @xmath286 \quad \text{with } \quad b(v ) = \delta_{g(v ) } + 2 \nabla_{g(v ) { \smash{\cdot}}df } - 2{\lambda}v'[f],\end{aligned}\ ] ] essentially splitting the operator @xmath281 into orders of @xmath30 .
in particular , the one - form @xmath287 does not involve the level @xmath30 .
we shall divide the calculation of the curvature into a number of propositions .
the first relies on and , in combination with , to compute a commutator of fundamental importance to the curvature calculation .
[ prop:3 ] for any rigid family of khler structures , the commutator of @xmath288 and @xmath289 , acting on @xmath39 , has the symbols @xmath290 } & = 0 \\ \sigma_2 { \big [ \delta_{\tilde g(v ) } , \delta_{\tilde g(w ) } \big ] } & = -4ik\;\ !
\theta(v , w ) \\
\sigma_1{\big [ \delta_{\tilde g(v ) } , \delta_{\tilde g(w ) } \big ] } & = \delta_{g(v ) } \delta g(w ) - \delta_{g(w ) } \delta g(v ) + \delta_{\bar g(v ) } \delta \bar g(w ) - \delta_{\bar g(w ) } \delta \bar g(v ) \\[-1pt ] & \qquad \smash{- 4ik\;\ ! \delta ( \theta(v , w ) ) - \delta ( \tilde g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\tilde g(w ) + \delta ( \tilde g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}\tilde g(v ) } \\[1pt ] \sigma_0 { \big [ \delta_{\tilde g(v ) } , \delta_{\tilde g(w ) } \big ] } & = - ik\;\ ! \delta
\delta \theta ( v , w ) + ik\;\ !
r ( \theta(v , w ) ) , \end{aligned}\ ] ] for any vector fields @xmath7 and @xmath149 on @xmath3 .
we verify that the third - order symbol vanishes . using and its conjugated version , which rely heavily on rigidity on the family , we get @xmath291 } & = \sigma_3{\big [ \nabla^2_{\tilde g(v ) } , \nabla^2_{\tilde g(w ) } \big ] } \\ & = 2 \operatorname{\mathcal{s}}(\tilde g(v ) { \smash{\cdot}}\nabla \tilde g(w ) ) - 2 \operatorname{\mathcal{s}}(\tilde g(w ) { \smash{\cdot}}\nabla \tilde g(v ) ) \\ & = 2 \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\nabla g(w ) ) - 2 \operatorname{\mathcal{s}}(g(w ) { \smash{\cdot}}\nabla g(v ) ) \\ & \qquad + 2 \operatorname{\mathcal{s}}(\bar g(v ) { \smash{\cdot}}\nabla \bar g(w ) ) - 2 \operatorname{\mathcal{s}}(\bar g(w ) { \smash{\cdot}}\nabla \bar g(v ) ) \\ & = 2 \gamma_3(v , w ) - 2 \bar \gamma_3(w , v ) + 2 \gamma_3(v , w ) - 2 \bar \gamma_3(w , v ) \\ & = 0 .
\end{aligned}\ ] ] to calculate the second - order symbol , we first notice that @xmath292 } & = \ :
\sigma_2{\big [ \nabla^2_{g(v ) } , \nabla^2_{g(w ) } \big ] } + \sigma_2{\big [ \nabla^2_{g(v ) } , \nabla_{\delta g(w ) } \big ] } - \sigma_2{\big [ \nabla^2_{g(w ) } , \nabla_{\delta g(v ) } \big ] } \\ & = \nabla^2_{g(v ) } g(w ) + \nabla_{\delta g(v ) } g(w ) + 2 \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\nabla \delta g(w ) ) \\ & \qquad - \nabla^2_{g(w ) } g(v ) - \nabla_{\delta g(w ) } g(v ) - 2 \operatorname{\mathcal{s}}(g(w ) { \smash{\cdot}}\nabla \delta g(v ) ) \\ & = \gamma_2(v , w ) - \gamma_2(w , v ) \\ & = 0 .
\end{aligned}\ ] ] for the mixed - type terms of the second - order symbol , we first observe that @xmath293 using this , we calculate that @xmath294 } \\ & \qquad = \ : \sigma_2 { \big [ \nabla^2_{g(v ) } , \nabla^2_{\bar g(w ) } \big ] } + \sigma_2{\big [ \nabla^2 _ { g(v ) } , \nabla _ { \delta \bar g(w ) } \big ] } - \sigma_2{\big [ \nabla^2_{\bar g(w ) } , \nabla_{\delta g(v ) } \big ] } \\ & \qquad = - 4ik \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) + 2 \operatorname{\mathcal{s}}(g(v)^{xy } r^a_{uxy } \bar g(w)^{ub } ) - 2 \operatorname{\mathcal{s}}(\bar g(w)^{uv } r^a_{xuv } g(v)^{xb } ) \\ & \qquad \qquad + 2 \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\nabla \delta \bar g(w ) ) - 2 \operatorname{\mathcal{s}}(\bar g(w ) { \smash{\cdot}}\nabla \delta g(v ) ) \\ & \qquad = - 4ik \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) , \end{aligned}\ ] ] where the last equality follows by inserting and its conjugate . in total , this means that the second - order symbol is given by @xmath295 } & = - 4ik \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) - 4ik \operatorname{\mathcal{s}}(\bar g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}g(w ) ) \\ & = - 4ik \operatorname{\mathcal{s}}(\tilde g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\tilde g(w ) ) \\ & = - 4ik\ , \theta(v , w )
. \phantom{\tilde g(v ) } \end{aligned}\ ] ] next , we find the first - order symbol .
first of all , we get @xmath296 } & = \delta_{g(v ) } \delta g(w ) - \delta_{g(w ) } \delta g(v ) .
\end{aligned}\ ] ] for the terms of mixed type on @xmath6 , we get the following computation , @xmath297 } \\ & \quad = \sigma_1 \!\!\:{\big [ \nabla^2_{g(v ) } , \!\nabla^2_{\bar g(w ) } \big ] } \
! + \sigma_1 \!\!\ : { \big [ \nabla_{\delta g(v ) } , \!\nabla_{\delta \bar g(w ) } \big ] } \ ! + \sigma_1 \!\!\ : { \big [ \nabla^2_{g(v ) } , \!\nabla_{\delta \bar g(w ) } \big ] } \!-
\sigma_1 \!\!\ : { \big [ \nabla^2_{\bar g(w ) } , \!\nabla_{\delta g(v ) } \big ] } \\ & \quad = - { \ifmmode\sbox\cbox{$2ik g(v)^{xy } \omega_{yu } \nabla_x ( \bar g(w)^{ua})$}\else\sbox\cbox{2ik g(v)^{xy } \omega_{yu } \nabla_x ( \bar g(w)^{ua})}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } + { \ifmmode\sbox\cbox{$2ik \bar g(w)^{uv } \omega_{vx } \nabla_u ( g(v)^{xa})$}\else\sbox\cbox{2ik \bar g(w)^{uv } \omega_{vx } \nabla_u ( g(v)^{xa})}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad \qquad - g(v)^{xy } \nabla_x ( r^a_{yuv } ) \bar g(w)^{uv } + \bar g(w)^{uv } \nabla_v ( r^a_{uxy } ) g(v)^{xy } \\ & \qquad \qquad- { \ifmmode\sbox\cbox{$\frac{4}{3 } g(v)^{xy } r^a_{xuv } \nabla_y \bar g(w)^{uv}$}\else\sbox\cbox{\frac{4}{3 } g(v)^{xy } r^a_{xuv } \nabla_y \bar g(w)^{uv}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } + { \ifmmode\sbox\cbox{$\frac{4}{3 } \bar g(w)^{uv } r^a_{uxy } \nabla_v g(v)^{xy}$}\else\sbox\cbox{\frac{4}{3 } \bar g(w)^{uv } r^a_{uxy } \nabla_v g(v)^{xy}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad\;\ ! - \delta g(v ) { \smash{\cdot}}r { \smash{\cdot}}\bar g(w ) + { \ifmmode\sbox\cbox{$\delta g(v)^{x } r^a_{xuv } \bar g(w)^{uv}$}\else\sbox\cbox{\delta g(v)^{x } r^a_{xuv } \bar g(w)^{uv}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad \qquad + \delta \bar g(w ) { \smash{\cdot}}r { \smash{\cdot}}g(v ) - { \ifmmode\sbox\cbox{$ \delta \bar g(w)^{u } r^a_{uxy } g(v)^{xy}$}\else\sbox\cbox { \delta \bar g(w)^{u } r^a_{uxy } g(v)^{xy}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad\;\
! + \nabla^2_{g(v ) } \delta \bar g(w ) - 2ik g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) + { \ifmmode\sbox\cbox{$g(v)^{xy } r^a_{uxy } \delta \bar g(w)^u$}\else\sbox\cbox{g(v)^{xy } r^a_{uxy } \delta \bar g(w)^u}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad\;\ !
- \nabla^2_{\bar g(w ) } \delta g(v ) + 2ik \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta g(v ) - { \ifmmode\sbox\cbox{$\bar g(w)^{uv } r^a_{xuv } \delta g(v)^x$}\else\sbox\cbox{\bar g(w)^{uv } r^a_{xuv } \delta g(v)^x}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \quad = - { \ifmmode\sbox\cbox{$g(v)^{xy } \nabla_x ( r^a_{yuv } ) \bar g(w)^{uv}$}\else\sbox\cbox{g(v)^{xy } \nabla_x ( r^a_{yuv } ) \bar g(w)^{uv}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } + { \ifmmode\sbox\cbox{$\bar g(w)^{uv } \nabla_v ( r^a_{uxy } ) g(v)^{xy}$}\else\sbox\cbox{\bar g(w)^{uv } \nabla_v ( r^a_{uxy } ) g(v)^{xy}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } \\ & \qquad\;\ ! - \delta g(v ) { \smash{\cdot}}r { \smash{\cdot}}\bar g(w ) + \delta \bar g(w ) { \smash{\cdot}}r { \smash{\cdot}}g(v ) \\ & \qquad\;\!- g(v)^{xy } \nabla_x ( r_{yu } ) \bar g(w)^{ua } + { \ifmmode\sbox\cbox{$g(v)^{xy } \nabla_x ( r^a_{yuv } ) \bar g(w)^{uv}$}\else\sbox\cbox{g(v)^{xy } \nabla_x ( r^a_{yuv } ) \bar g(w)^{uv}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } - 2ik g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) \\ & \qquad\;\ ! + \bar g(w)^{uv } \nabla_u ( r_{vx } ) g(v)^{xa } - { \ifmmode\sbox\cbox{$\bar g(v)^{uv } \nabla_u ( r^a_{vxy } ) g(v)^{xy}$}\else\sbox\cbox{\bar g(v)^{uv } \nabla_u ( r^a_{vxy } ) g(v)^{xy}}\fi \makebox[0pt][l]{\usebox\cbox } \rule[0.3\ht\cbox-0.2pt/2]{\wd\cbox}{0.2pt } } + 2ik \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta g(v ) \\ & \quad = - \delta g(v ) { \smash{\cdot}}r { \smash{\cdot}}\bar g(w ) + \delta \bar g(w ) { \smash{\cdot}}r { \smash{\cdot}}g(v ) \\ & \qquad\;\
! - g(v)^{xy } \nabla_x ( r_{yu } ) \bar g(w)^{ua } - 2ik g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) \\ & \qquad\;\ ! + \bar g(w)^{uv } \nabla_u ( r_{vx } ) g(v)^{xa } + 2ik \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta g(v ) \\ & \quad = - \delta ( g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\bar g(w ) \ ! + \!\delta ( \bar g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}g(v)\
! + \!2ik \delta ( \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) ) \!- \ ! 2ik \delta ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) \\ & \quad = - \delta ( g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\bar g(w ) + \delta ( \bar g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}g(v ) - 4ik \delta \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) , \end{aligned}\ ] ] where we have indicated cancelling terms .
analagously , we compute @xmath298 } = \delta ( g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}\bar g(v ) - \delta ( \bar g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}g(w ) - 4ik \delta \operatorname{\mathcal{s}}(\bar g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}g(w ) ) , \end{aligned}\ ] ] so that finally @xmath299 } & = - 4i \delta ( \theta(v , w ) ) - \delta ( \tilde g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\tilde g(w ) + \delta ( \tilde g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}\tilde g(v ) \\ & \quad \;\ ! + \delta_{g(v ) } \delta g(w ) - \delta_{g(w ) } \delta g(v ) + \delta_{\bar g(v ) } \delta \bar g(w ) - \delta_{\bar g(w ) } \delta \bar g(v ) . \end{aligned}\ ] ] for the symbol of order zero , we first observe that @xmath300 } = 0 .
\end{aligned}\ ] ] for the mixed terms , we get @xmath301 } \\ & \quad = \sigma_0{\big [ \nabla^2_{g(v ) } , \!\nabla^2_{\bar g(w ) } \big ] } \!+ \sigma_0{\big [ \nabla_{\delta g(v ) } , \!\nabla_{\delta \bar g(w ) } \big ] } \!+ \sigma_0 { \big [ \nabla^2_{g(v ) } , \!\nabla_{\delta \bar g(w ) } \big ] } \
! - \sigma_0 { \big [ \nabla^2_{\bar g(w ) } , \!\nabla_{\delta g(v ) } \big ] } \\ & \quad = - kg(v)^{xy } r_{xuvy } \bar g(w)^{uv } - ik \delta g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) \\ & \qquad\;\ ! - ik \omega(g(v ) { \smash{\cdot}}\nabla \delta \bar g(w ) ) + ik \omega ( \bar g(w ) { \smash{\cdot}}\nabla \delta g(v ) ) \\ & \quad = - kg(v)^{xy } r_{xuvy } \bar g(w)^{uv } - ik \delta \delta ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) - ik \omega(g(v ) { \smash{\cdot}}\nabla \delta \bar g(w ) ) \\ & \quad = -ik\;\ ! \delta \delta ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) + ik\;\ ! r ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) , \end{aligned}\ ] ] where was applied for the last equation , and this finally gives @xmath302 } = - ik\;\ ! \delta
\delta \theta ( v , w ) + ik\;\ ! r ( \theta(v , w ) ) .
\end{aligned}\ ] ] this proves the proposition . before stating the next proposition , we introduce a one - form @xmath303 , with values in smooth functions on @xmath6 , which will play a central role in the calculations .
it is defined by the expression , @xmath304.\end{aligned}\ ] ] this one - form serves , in fact , as the zero - order part of the hitchin connection in metaplectic quantization , studied in @xcite , but as we will see , it also appears in the curvature of the hitchin connection in the setting considered here . in the metaplectic case , the crucial property satisfied by this one - form is the following relation , @xmath305 which will also be useful to us .
this is verified through the calculation @xmath306 - 2\bar { \partial}(df { \smash{\cdot}}g(v ) { \smash{\cdot}}df ) - i\delta ( g(v ) { \smash{\cdot}}\rho),\end{aligned}\ ] ] where we applied twice for the third equality and for the last equality . as hinted above
, the exterior derivative of the one - form @xmath307 over @xmath3 appears in the curvature of the hitchin connection . to see how , we must be able to recognize this derivative .
[ prop:24 ] the exterior derivative of the one - form @xmath307 defined by is given by @xmath308 - b(w ) v'[f ] \\
\bar { \partial}_{\scriptscriptstyle \mathcal{t } } { c}(v , w ) & = \frac{i}{4 } \delta\delta \theta(v , w ) - \frac{i}{4 } r(\theta ( v , w ) ) - i { \lambda}\theta(v , w ) - 2{\lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f ( v , w ) \end{aligned}\ ] ] for any vector fields @xmath7 and @xmath149 on @xmath3 . for the first statement , choose @xmath7 and @xmath149 so that @xmath181 and @xmath309 commute . using the identity @xmath310 and the fact that the divergence operator on vector fields does not depend on the khler structure
, we get @xmath311 - w'[{c}(v ) ] \\ & = - \delta_{v'[g(w ) ] } f - \delta_{g(w ) } v'[f ] - 2 df { \smash{\cdot}}g(w ) { \smash{\cdot}}dv'[f ] \\ & \qquad + \delta_{w'[g(v ) ] } f + \delta_{g(v ) } w'[f ] + 2 df { \smash{\cdot}}g(v ) { \smash{\cdot}}dw'[f ] \\ & = b(v ) w'[f ] - b(w ) v'[f ] , \end{aligned}\ ] ] where we used the the fact that @xmath312 = - w'v'[\tilde g ] = v'[g(w)]$ ] , since the vector fields were chosen to commute . for the second statement , choose @xmath7 and @xmath149 so that @xmath181 and @xmath182 commute . once again using @xmath313 , the identities and give us that @xmath314 \\ & \qquad= 4\delta(w''[g(v ) ] { \smash{\cdot}}df ) + 4\delta ( g(v ) { \smash{\cdot}}w''[f ] ) \\
& \qquad= i \delta ( \delta \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) ) - 2i \delta ( \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df ) \\ & \qquad= i \delta \delta ( \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) ) - 2i \delta \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df + 2i { \partial}\bar { \partial}f({\bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) } ) , \end{aligned}\ ] ] where rigidity of the family of khler structures was used for the last equality .
the ricci potential satisfies the equation @xmath315 , so we get that @xmath316 finally , the identities and can be used to verify that @xmath317 = 2i \delta \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df .
\end{aligned}\ ] ] combining the identities above , we get @xmath318 \\ & \qquad = i \delta\delta \theta(v',w '' ) - ir(\theta(v',w '' ) ) - 4i{\lambda}\theta(v ' , w '' ) - 8{\lambda}{\partial}_{\scriptscriptstyle \mathcal{t}}\bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v ' , w '' ) .
\end{aligned}\ ] ] for the first term of the last equality , we used the fact that repeated application of the divergence operator to a bivector field only depends on its symmetric part .
this finishes the proof of the proposition .
we will also need the following [ prop:8 ] the one - form @xmath307 defined in satisifes @xmath319 for any vector fields @xmath7 and @xmath149 on @xmath3 .
as usual , the hamiltonian vector field in the statement is determined by @xmath320 to calculate the right - hand side of this , we use the fact that @xmath307 satisfies to get @xmath321 = w ' [ \bar { \partial}{c}(v ) ] = \bar { \partial}w'[{c}(v ) ] - \frac{i}{2 } \omega { \smash{\cdot}}g(w ) { \smash{\cdot}}d { c}(v ) . \end{aligned}\ ] ] on the other hand , we calculate @xmath322 & = w'[\nabla_x ( g(v)^{xy } \rho_{xa } ) ] \\ & = w'[{{\nabla^{g}}}]^z_{zx } g(v)^{xy } \rho_{xa } - w'[{{\nabla^{g}}}]^z_{xa } g(v)^{xy}\rho_{yz } + \delta ( w'[g(v ) { \smash{\cdot}}\rho ] ) \\ & = \delta ( w'[g(v ) ] { \smash{\cdot}}\rho ) + \delta ( g(v ) { \smash{\cdot}}w'[\rho ] ) , \end{aligned}\ ] ] where the last equality uses and and type considerations .
applying , we get @xmath323 ) & = \delta ( g(v ) { \smash{\cdot}}d ( \delta g(w ) { \smash{\cdot}}\omega ) ) = \delta ( g(v ) { \smash{\cdot}}\nabla \delta g(w ) ) { \smash{\cdot}}\omega = \delta_{g(v ) } \delta g(w ) { \smash{\cdot}}\omega . \end{aligned}\ ] ] combining the previous three identities , we find @xmath324 - \bar { \partial}w'[{c}(v ) ] \\ & \qquad = \frac{i}{2 } \omega { \smash{\cdot}}g(v ) { \smash{\cdot}}d { c}(w ) - \frac{i}{2 } \omega { \smash{\cdot}}g(w ) { \smash{\cdot}}d { c}(v ) + \frac{i}{4 } \delta_{g(w ) } \delta g(v ) { \smash{\cdot}}\omega - \frac{i}{4 } \delta_{g(v ) } \delta g(w ) { \smash{\cdot}}\omega , \end{aligned}\ ] ] for commuting vector fields @xmath181 and @xmath182 .
raising the index with @xmath43 ends the proof . the curvature calculation for the hitchin connection proceeds with expressions for commutators involving the one - form @xmath287 defined in . [ prop:22 ]
the commutator of the operators @xmath325 and @xmath326 , acting on sections of @xmath27 , is a first - order operator with symbols given by @xmath327 } & = 4i x'_{{\partial}_{{{\scriptscriptstyle \mathcal{t } } } } { c}(v , w ) } \\
\sigma_0 { \big [ b(v ) , b(w ) \big ] } & = -2{\lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } c(v , w ) , \end{aligned}\ ] ] for any vector fields @xmath7 and @xmath149 on @xmath3 .
the vanishing of the third - order symbol is essentially , @xmath328 } = \sigma_3 { \big [ \delta_{g(v ) } , \delta_{g(w ) } \big ] } = -4ik\theta(v',w ' ) = 0 .
\end{aligned}\ ] ] vanishing of the second - order symbol is seen through the following calculation using and , @xmath329 } \\ & \qquad = \sigma_2{\big [ \delta_{g(v ) } , \delta_{g(w ) } \big ] } + 2 \sigma_2{\big [ \delta_{g(v ) } , \nabla_{g(w){\smash{\cdot}}df } \big ] } - 2 \sigma_2{\big [ \delta_{g(w ) } , \nabla_{g(v ) { \smash{\cdot}}df } \big ] } \\ & \qquad = 2\sigma_2 { \big [ \nabla^2_{g(v ) } , \nabla_{g(w){\smash{\cdot}}df } \big ] } - 2 \sigma_2 { \big [ \nabla^2_{g(w ) } , \nabla_{g(v){\smash{\cdot}}df } \big ] } \\ & \qquad = 4 \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\nabla ( g(w){\smash{\cdot}}df ) ) - 4 \operatorname{\mathcal{s}}(g(w ) { \smash{\cdot}}\nabla ( g(v){\smash{\cdot}}df ) ) \\ & \qquad \qquad - 2 df { \smash{\cdot}}g(w ) { \smash{\cdot}}\nabla g(v ) + 2 df { \smash{\cdot}}g(v ) { \smash{\cdot}}\nabla g(w ) \\ & \qquad = 6 \operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}\nabla g(w ) ) { \smash{\cdot}}df - 6 \operatorname{\mathcal{s}}(g(w ) { \smash{\cdot}}\nabla g(v ) ) { \smash{\cdot}}df \\ & \qquad = 6 \gamma_3(v , w ) { \smash{\cdot}}df - 6 \gamma_3(w , v ) { \smash{\cdot}}df \\ & \qquad = 0 .
\end{aligned}\ ] ] the first - order symbol is the first non - vanishing part .
we split its calculation in two by first calculating @xmath330 } - \sigma_1{\big [ \delta_{g(w ) } , \nabla_{g(v){\smash{\cdot}}df } \big ] } \\ &
\qquad = \sigma_1{\big [ \nabla^2_{g(v ) } , \nabla_{g(w){\smash{\cdot}}df } \big ] } - \sigma_1{\big [ \nabla^2_{g(w ) } , \nabla_{g(v){\smash{\cdot}}df } \big ] } \\ & \qquad \qquad + \sigma_1{\big [ \nabla_{\delta g(v ) } , \nabla_{g(w){\smash{\cdot}}df } \big ] } - \sigma_1{\big [ \nabla_{\delta g(w ) } , \nabla_{g(v){\smash{\cdot}}df } \big ] } \\ & \qquad = \nabla^2_{g(v ) } ( g(w){\smash{\cdot}}df ) - \nabla^2_{g(w ) } ( g(v){\smash{\cdot}}df ) + [ \delta g(v ) , g(w ) { \smash{\cdot}}df ] - [ \delta g(w ) , g(v ) { \smash{\cdot}}df ] \\ & \qquad = \delta_{g(v ) } ( g(w){\smash{\cdot}}df ) - \delta_{g(w ) } ( g(v){\smash{\cdot}}df ) - df { \smash{\cdot}}g(w ) { \smash{\cdot}}\nabla \delta g(v ) + df { \smash{\cdot}}g(v ) { \smash{\cdot}}\nabla \delta g(w ) \\ & \qquad = df { \smash{\cdot}}g(v ) { \smash{\cdot}}\nabla \delta g(w ) - df { \smash{\cdot}}g(w ) { \smash{\cdot}}\nabla \delta g(v ) \\ & \qquad \qquad + \delta_{g(v ) } ( g(w ) ) { \smash{\cdot}}df + 2 g(v)^{xy } \nabla_y g(w)^{au } \nabla^2_{xu } f \\ & \qquad \qquad
\qquad + g(w ) { \smash{\cdot}}d \delta_{g(v ) } f - g(w)^{au } \nabla_u ( g(v)^{xy } )
\nabla^2_{xy}f - g(w ) { \smash{\cdot}}\nabla \delta g(v ) { \smash{\cdot}}df \\ & \qquad \qquad - \delta_{g(w ) } ( g(v ) ) { \smash{\cdot}}df - 2 g(w)^{uv } \nabla_v g(v)^{ax } \nabla^2_{xu } f \\ & \qquad \qquad \qquad - g(v ) { \smash{\cdot}}d \delta_{g(w ) } f + g(v)^{ax } \nabla_x ( g(w)^{uv } ) \nabla^2_{uv}f + g(v ) { \smash{\cdot}}\nabla \delta g(w ) { \smash{\cdot}}df \\ & \qquad = 3 \gamma_3(v , w)^{axy } \nabla^2_{xy } f - 3 \gamma_3(w , v)^{axy } \nabla^2_{xy } f + \gamma_2(v , w ) { \smash{\cdot}}df - \gamma_2(w , v ) { \smash{\cdot}}df \\ & \qquad \qquad + g(w ) { \smash{\cdot}}d \delta_{g(v ) } f - g(v ) { \smash{\cdot}}d \delta_{g(w ) } f \\ & \qquad = g(w ) { \smash{\cdot}}d \delta_{g(v ) } f - g(v ) { \smash{\cdot}}d \delta_{g(w ) } f. \end{aligned}\ ] ] here , once again , played a central role .
similarly , we find that @xmath331 } & = 2{\big [ g(v){\smash{\cdot}}df , g(w){\smash{\cdot}}df \big ] } \\ & = 3 df { \smash{\cdot}}\gamma_3(v , w ) { \smash{\cdot}}df - 3 df { \smash{\cdot}}\gamma_3(w , v ) { \smash{\cdot}}df \\ & \qquad + g(w ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(v ) { \smash{\cdot}}df ) - g(v ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(w ) { \smash{\cdot}}df ) \\ &
= g(w ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(v ) { \smash{\cdot}}df ) - g(v ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(w ) { \smash{\cdot}}df ) . \end{aligned}\ ] ] combining the last two computations , we see that @xmath332 } & = \delta_{g(v ) } \delta g(w ) - \delta_{g(v ) } \delta g(w ) \\ & \qquad + 2g(w ) { \smash{\cdot}}d \delta_{g(v ) } f - 2 g(v ) { \smash{\cdot}}d \delta_{g(w ) } f \\ & \qquad + 2 g(w ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(v ) { \smash{\cdot}}df ) - 2 g(v ) { \smash{\cdot}}d ( df { \smash{\cdot}}g(w ) { \smash{\cdot}}df ) \\ & \qquad + 4{\lambda}g(w ) { \smash{\cdot}}d v'[f ] - 4{\lambda}g(v ) { \smash{\cdot}}d w'[f ] \\ & = \delta_{g(v ) } \delta g(w ) - \delta_{g(v ) } \delta g(w ) \\ & \qquad + 2 g(v ) { \smash{\cdot}}d { c}(w ) - 2 g(w ) { \smash{\cdot}}d { c}(v ) , \end{aligned}\ ] ] where the last equation follows by recalling . finally ,
for the zeroth - order symbol , we rely on to get that @xmath333 } & = 2{\lambda}b(w ) v'[f ] - 2{\lambda}b(v ) w'[f ] = -2{\lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } { c}(v , w ) .
\end{aligned}\ ] ] this finishes the proof of the proposition . finally , to calculate the curvature of the hitchin connection , we need the extorior derivative of the one - form @xmath287 over @xmath3 .
this is calculated in the following proposition .
[ prop:11 ] the two - form @xmath334 is given by @xmath335 - g(w ) { \smash{\cdot}}dv[f ] } + 2 { \lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v , w ) \end{aligned}\ ] ] on sections of @xmath27 , and by @xmath336 } - 2 \nabla_{g(v ) { \smash{\cdot}}w'[f ] } - 2ik \theta(v , w ) + 2{\lambda}{\partial}_\mathcal{t } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v , w ) \end{aligned}\ ] ] when restricted to sections of @xmath284 .
for any section @xmath74 of @xmath27 over @xmath3 , we have the identity @xmath337s & = w[\nabla_x g(v)^{xy } \nabla_y]s \\ & = w[{{\nabla^{g}}}]^z_{zx } g(v)^{xy } \nabla_y + \nabla_x w[g(v)]^{xy } \nabla_y s \\ & = \delta_{w[g(v ) ] } s , \end{aligned } \end{aligned}\ ] ] where the last equality follows from .
in particular , for commuting @xmath7 and @xmath149 , we get @xmath338 - w[\delta_{g(v ) } ] = \delta_{v[g(w ) ] } - \delta_{w[g(v ) ] } = \delta_{d_{\scriptscriptstyle \mathcal{t } } g(v , w ) } = -i \delta_{\theta(v , w ) } , \end{aligned}\ ] ] where was used for the last equation .
similarly , we calculate @xmath339 s & = \nabla_{g(v){\smash{\cdot}}w[f ] } s + \nabla_{w[g(v ) ] { \smash{\cdot}}df } s , \end{aligned}\ ] ] so that finally @xmath340 - w[b(v ) ] \notag \\ & \qquad = -i \delta_{\theta(v , w ) } - 2i\nabla_{\theta(v , w ) { \smash{\cdot}}df } - 2\nabla_{g(v){\smash{\cdot}}dw[f ] - g(w ) { \smash{\cdot}}dv[f ] } + 2{\lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v , w ) .
\end{aligned}\ ] ] this proves the first statement of the proposition . for the second statement , observe that the stated formulas on sections of @xmath27 and @xmath284 agree for @xmath341 . for the case of @xmath342 , suppose that @xmath74 is a section of @xmath284 and observe that @xmath343 using the conjugated version of , we also get that @xmath344 } s. \end{aligned}\ ] ] finally , plugging these two expressions into , we get @xmath345 } s - 2ik \theta(v',w '' ) s + 2{\lambda}{\partial}_\mathcal{t } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v ' , w '' ) s , \end{aligned}\ ] ] which finishes the proof of the proposition . with the above propositions at hand , calculating the curvature of the hitchin connection
is a straightforward matter .
[ thm:4 ] the curvature of the hitchin connection acts by @xmath346 on sections of the bundle @xmath284 . for commuting vector fields @xmath7 and @xmath149 , we get @xmath347 } \\ & \qquad = \frac{{\big [ b(v ) , b(w ) \big ] } } { ( 4k+2{\lambda})^2 } + \frac{d_{\scriptscriptstyle \mathcal{t}}b(v , w ) + { \big [ b(v ) , w'[f ] \big ] } - { \big [ b(w ) , v'[f ] \big ] } } { 4k+2{\lambda } } - { \partial}_{\scriptscriptstyle \mathcal{t}}\bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v , w ) . \end{aligned}\ ] ]
now can be applied to see that @xmath348 \big ] } - { \big [ b(w ) , v'[f ] \big ] } = 2\nabla_{g(v ) { \smash{\cdot}}w'[f ] } - 2\nabla_{g(w ) { \smash{\cdot}}v'[f ] } + { \partial}_{\scriptscriptstyle \mathcal{t}}{c}(v , w ) , \end{aligned}\ ] ] and combining the above with and yields the following expression for the ( 2,0)-part of the curvature @xmath349}}{(4k+2{\lambda})^2 } + \frac{{\partial}_{\scriptscriptstyle \mathcal{t}}{c}(v , w)}{4k+2{\lambda } } = \frac{k}{(2k+{\lambda})^2 } p_k({\partial}_{\scriptscriptstyle \mathcal{t}}{c}(v , w ) ) , \end{aligned}\ ] ] where @xmath350 denotes the prequantum operator defined in .
for the ( 1,1)-part of the curvature , yields @xmath351 finally , the ( 0,2)-part of the curvature clearly vanishes , and the theorem is proved . the fact that the ( 1,1)-part of the curvature is an zeroth - order operator , combined with the fact that the hitchin connection preserves the subbundle of quantum spaces @xmath284 implies that the ( 1,1)-part must take values in holomorphic and hence constant functions on @xmath6 .
this was , however , already known from for the particular expression we found in .
also , being a first - order operator , it is no surprise that the ( 2,0)-part of the curvature acts by prequantum operator .
after all , prequantum operators of the form are the only first - order operators which preserve the holomorphic sections . in the previous section
, we saw how the higher - order symbols of the curvature of the hitchin connection vanished and left a first - order operator with a relatively simple expression .
for the ( 2,0)-part , the higher order symbols vanished for general reasons , related to rigidity of the family of khler structures , but for the ( 1,1)-part , the curvature reduced to a zero - order operator only when restricted the subbundle @xmath284 of quantum spaces .
it turns out that , with a slight modification of the formula , we can achieve a cancellation of higher - order terms for the ( 1,1)-part of the connection defined on the whole bundle @xmath27 of prequantum spaces over @xmath3 , but still maintain the vanishing of higher - order symbols for the ( 2,0 ) and ( 0,2)-parts . in this section
, we let @xmath29 be any complex number with integer real part , @xmath352 , and we consider the connection on @xmath27 given by @xmath353,\end{aligned}\ ] ] where @xmath354 has the conjugated symbols of @xmath325 so that @xmath355 \quad \text{and } \quad \bar b(v ) = \delta_{\bar g(v ) } + 2 \nabla _ { \bar g(v ) { \smash{\cdot}}df } - 2{\lambda}v''[f].\end{aligned}\ ] ] we will refer to @xmath356 as the hitchin - witten connection .
it is a generalization of the connection for quantum chern - simons theory with complex gauge group @xmath357 discussed by witten in @xcite , where he arrives at exactly the formula .
this relation will be further explored in the final section of the paper .
we will prove that the curvature of @xmath358 , acting on sections of @xmath27 , has essentially the same expression as the curvature of @xmath279 , acting on sections of @xmath284 .
we start with the following .
[ prop:12 ] the commutator of the operators @xmath325 and @xmath359 , acting on sections of @xmath27 , is a second - order operator with symbols given by @xmath360 } & = - 4ik \;\ ! \theta(v',w '' ) \\
\sigma_1{\big [ b(v ) , \bar b(w ) \big ] } & = - 4ik \;\ !
\theta(v',w '' ) - 8ik \;\ !
\theta(v',w '' ) { \smash{\cdot}}df \\ \sigma_0{\big [ b(v ) , \bar b(w ) \big ] } & = - \frac{ik}{4}\;\ ! \delta \delta \theta ( v',w '' ) - \frac{ik}{4}\;\ ! r(\theta(v',w '' ) ) - 2ik{\lambda}\;\ ! \theta(v',w '' ) \\ & \qquad - ik \;\
! df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df - ik \;\ !
\delta \theta ( v',w '' ) { \smash{\cdot}}df , \end{aligned}\ ] ] and furthermore @xmath361 } = 2\bar t \ , b(v)w''[f ] + 2 t \ , \bar b(w ) v'[f ] - 4ik{\lambda}\ , \theta(v',w '' ) , \end{aligned}\ ] ] for any vector fields @xmath7 and @xmath149 on @xmath3 .
the third - order symbol vanishes , as readily gives @xmath362 } = \sigma_3{\big [ \delta_{g(v ) } , \delta_{\bar g(w ) } \big ] } = 0 .
\end{aligned}\ ] ] the calculation of the second - order symbol is also straightforward , @xmath363 } \\ & \qquad = \sigma_2 { \big [ \delta_{g(v ) } , \delta_{\bar g(w ) } \big ] } + 2 \sigma_2 { \big [ \nabla^2_{g(v ) } , \nabla_{\bar g(w){\smash{\cdot}}df } \big ] } - 2 \sigma_2 { \big [ \nabla^2_{\bar g(w ) } , \delta_{g(v){\smash{\cdot}}df } \big ] } \\ & \qquad = -4ik \theta(v',w '' ) + 4\operatorname{\mathcal{s}}(g(v ) { \smash{\cdot}}{\partial}\bar { \partial}f { \smash{\cdot}}\bar g(w ) ) - 4\operatorname{\mathcal{s}}(\bar g(w ) { \smash{\cdot}}\bar { \partial}{\partial}f { \smash{\cdot}}g(v ) ) \\
& \qquad = -4ik \theta(v',w '' ) .
\end{aligned}\ ] ] for the first - order symbol , we first calculate @xmath364}$ ] .
we get @xmath365 } \\ & \qquad = 2\nabla^2_{g(v ) } ( \bar g(w){\smash{\cdot}}df ) - 4ikg(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df + 2g(v)^{xy}r^a_{uxy } \bar g(w)^{uv } df_v \end{aligned}\ ] ] and @xmath366 } \\ & \qquad = 2\nabla_{\delta g(v ) } ( \bar g(w ) { \smash{\cdot}}df ) - 2 df { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}\nabla \delta g(v ) \\ & \qquad = 2\nabla_{\delta g(v ) } ( \bar g(w ) { \smash{\cdot}}df ) + 2idf { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}\rho { \smash{\cdot}}g(v ) - 2df_u \bar g(w)^{uv } r^a_{vxy } g(v)^{xy } , \end{aligned}\ ] ] and when combining these , we get @xmath367 } \\ & \qquad = 2\delta_{g(v ) } ( \bar g(w ) { \smash{\cdot}}df ) - 4ikg(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df - 2i g(v ) { \smash{\cdot}}\rho { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df .
\end{aligned}\ ] ] the first term of this expression can be rewritten as @xmath368 for the first - order symbol , we must also understand @xmath369 \big ] } } $ ] , which yields @xmath370 \big ] } = 4 g(v ) { \smash{\cdot}}dw''[f ] = - i g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) - 2i g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df .
\end{aligned}\ ] ] altogether , the previous identities yield @xmath367 } - 2{\lambda}\ , \sigma_1{\big [ \nabla^2_{g(v ) } , w''[f ] \big ] } \\ & \qquad = 2\delta_{g(v ) } ( \bar g(w ) { \smash{\cdot}}df ) - 4ikg(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df - 2i g(v ) { \smash{\cdot}}\rho { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df \\ & \qquad \qquad - i{\lambda}\delta ( \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) ) + 2i{\lambda}g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df \\ & \qquad = -i \delta ( g(v ) { \smash{\cdot}}\rho { \smash{\cdot}}\bar g(w ) ) - 4ikg(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df + 4 g(v ) { \smash{\cdot}}{\partial}\bar { \partial}f { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df \\ & \qquad \qquad + 2i{\lambda}\ , \delta\theta(v',w '' ) \\ & \qquad = \delta ( g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\bar g(w ) - 4ikg(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df + 4 df { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}\nabla ( g(v ) { \smash{\cdot}}df ) \\ & \qquad \qquad + 2i{\lambda}\ , \delta \theta(v',w '' ) . \end{aligned}\ ] ] a completely analogous computation shows @xmath371 } - 2{\lambda}\ , \sigma_1 { \big [ \nabla^2_{\bar g(w ) } , v'[f ] \big ] } \\ & \qquad = \delta ( \bar g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}g(v ) - 4ik \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df + 4 df { \smash{\cdot}}g(v ) { \smash{\cdot}}\nabla ( \bar g(w ) { \smash{\cdot}}df ) \\ & \qquad \qquad + 2i{\lambda}\ , \delta \theta(v',w '' ) .
\end{aligned}\ ] ] finally , combining all of the above with , we get @xmath372 } & = \sigma_1 { \big [ \delta_{g(v ) } , \delta_{\bar g(w ) } \big ] } + 4{\big [ g(v ) { \smash{\cdot}}df , \bar g(w ) { \smash{\cdot}}df \big ] } \\ & \qquad + 2\sigma_1{\big [ \delta_{g(v ) } , \nabla_{\bar g(w ) { \smash{\cdot}}df } \big ] } - 2{\lambda}\ , \sigma_1{\big [ \nabla^2_{g(v ) } , w''[f ] \big ] } \\ & \qquad - 2\sigma_1 { \big [ \delta_{\bar g(w ) } , \nabla_{g(v ) { \smash{\cdot}}df } \big ] } + 2{\lambda}\ , \sigma_1 { \big [ \nabla^2_{\bar g(w ) } , v'[f ] \big ] } \\ & = \sigma_1 { \big [ \delta_{g(v ) } , \delta_{\bar g(w ) } \big ] } - 8ik \ , \theta(v',w '' ) { \smash{\cdot}}df \\ & \qquad + \delta ( g(v ) { \smash{\cdot}}r ) { \smash{\cdot}}\bar g(w ) - \delta ( \bar g(w ) { \smash{\cdot}}r ) { \smash{\cdot}}g(v ) \\ & = - 4ik \ , \delta \theta(v',w '' ) - 8ik \ , \theta(v',w '' ) { \smash{\cdot}}df \end{aligned}\ ] ] it only remains to calculate the zero - order symbol .
first we compute @xmath373 } & = -2ik\,\omega ( g(v ) { \smash{\cdot}}\nabla ( df { \smash{\cdot}}\bar g(w ) ) ) \\ & = -2ik\ , \omega ( g(v ) { \smash{\cdot}}{\partial}\bar { \partial}f { \smash{\cdot}}\bar g(w ) ) \\ & = -2ik\ , { \partial}\bar { \partial}f ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) \\ & = - k\:\!\rho ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) + k{\lambda}\ , \omega ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) \\ & = - ik\ , r ( \theta(v',w '' ) ) - 4ik{\lambda}\ , \theta(v',w '' ) , \end{aligned}\ ] ] and similarly , we have @xmath374 } & = ik\ , r ( \theta(v',w '' ) ) + 4ik{\lambda}\ , \theta(v',w '' ) . \end{aligned}\ ] ] furthermore , we calculate @xmath375 } - 2\sigma_0 { \big [ \nabla_{\delta \bar g(w ) } , \nabla_{g(v ) { \smash{\cdot}}df } \big ] } \\ & \qquad = -2ik \delta g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df + 2ik \delta \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df \\ & \qquad = -4ik \ , \delta \theta ( v',w '' ) { \smash{\cdot}}df , \end{aligned}\ ] ] and finally @xmath376 } = - 4ik \ , df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df . \end{aligned}\ ] ] all of the above contribute to the zero - order symbol , but we also need to calculate @xmath377 $ ] and @xmath378 $ ] .
first we compute @xmath379 \\ & \qquad = 4 \delta ( g(v ) { \smash{\cdot}}dw''[f ] ) \\ & \qquad = - i \delta ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) ) - 2i \delta ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df ) \\ & \qquad = i \delta \delta ( \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) ) - 2i { \partial}\bar { \partial}f ( g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) ) - 2i \delta
g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df \\ & \qquad = - i \delta \delta(\theta(v',w '' ) ) - i\:\ ! r ( \theta(v',w '' ) ) - 4i{\lambda}\ , \theta(v',w '' ) - 2i\:\ ! \delta g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}df , \end{aligned}\ ] ] where the last equality used the fact that the double divergence of a bivector field only depends on its symmetric part .
similarly , we find @xmath380 = i \delta \delta\theta(w'',v ' ) + i\:\!r ( \theta(w'',v ' ) ) + 4i{\lambda}\ , \theta(w'',v ' ) + 2i \delta \bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}g(v ) { \smash{\cdot}}df .
\end{aligned}\ ] ] we also have @xmath381 & = - i df { \smash{\cdot}}g(v ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta \bar g(w ) - 2i df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df\\ 4\nabla_{\bar g(w ) { \smash{\cdot}}df } v'[f ] & = + i df { \smash{\cdot}}\bar g(w ) { \smash{\cdot}}\omega { \smash{\cdot}}\delta g(v ) + 2i df { \smash{\cdot}}\theta(w'',v ' ) { \smash{\cdot}}df , \end{aligned}\ ] ] so finally , we conclude that @xmath382 & = - i \delta \delta(\theta(v',w '' ) ) - i\ , r ( \theta(v',w '' ) ) - 4i{\lambda}\ , \theta(v',w '' ) \\ & \qquad - 4i \delta \theta(v',w '' ) { \smash{\cdot}}df - 2i df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df - 8{\lambda}v'[f]w''[f ] . \end{aligned}\ ] ] using this identity , conjugation yields @xmath383 = \overline{b(w)v''[f ] } = b(v)w''[f ] , \end{aligned}\ ] ] so the contributions of @xmath384 $ ] and @xmath377 $ ] to the zero - order symbol of the commutator cancel .
this finally gives @xmath385 } & = \sigma_0 { \big [ \delta_{g(v ) } , \delta_{\bar g(w ) } \big ] } + 2\sigma_0 { \big [ \delta_{g(v ) } , \nabla_{\bar g(w ) { \smash{\cdot}}df } \big ] } - 2\sigma_0 { \big [ \delta_{\bar g(w ) } , \nabla_{g(v ) { \smash{\cdot}}df } \big ] } \\
& \qquad + 4\sigma_0 { \big [ \nabla_{g(v ) { \smash{\cdot}}df } , \nabla_{\bar g(w ) { \smash{\cdot}}df } \big ] } - 2{\lambda}\,b(v)w''[f ] + 2{\lambda}\ , \bar b(w ) v'[f ] \\ & = - ik\ , \delta \delta \theta ( v',w '' ) - ik\ , r(\theta(v',w '' ) ) - 8ik{\lambda}\ , \theta(v',w '' ) \\ & \qquad - 4ik \ , df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df - 4ki \ , \delta \theta ( v',w '' ) { \smash{\cdot}}df , \end{aligned}\ ] ] as claimed in the proposition . to prove the final statement of the proposition ,
we note from the above that @xmath386 + 2t\ , \bar b(w ) v'[f ] \\ & \qquad = - i k \delta \delta(\theta(v',w '' ) ) - ik\ , r ( \theta(v',w '' ) ) - 4ik{\lambda}\ , \theta(v',w '' ) \\ & \qquad \qquad - 4i k\ , df { \smash{\cdot}}\theta(v',w '' ) { \smash{\cdot}}df - 4i k \ , \delta \theta ( v',w '' ) { \smash{\cdot}}df - 4{\lambda}kv'[f]w''[f ] , \end{aligned}\ ] ] which gives @xmath385 } & = 2 \bar t \ , b(v)w''[f ] + 2t\ , \bar b(w ) v'[f ] - 4ik{\lambda}\ , \theta(v',w '' ) + 4{\lambda}kv'[f]w''[f ] \\ & = 2\bar t \sigma_0 { \big [ b(v ) , w''[f ] \big ] } + 2t\ , \sigma_0 { \big [ \bar b(w ) , v'[f ] \big ] } - 4ik{\lambda}\ , \theta(v',w '' ) . \end{aligned}\ ] ] this proves the proposition . with the bulk of computations encoded in
, we can calculate the curvature of the hitchin - witten connection .
this is the content of the following theorem .
[ thm:1 ] the curvature of the hitchin - witten connection @xmath356 acts as a first - order operator with symbols @xmath387 \sigma_0 { f_{\raisebox{0pt}{\mbox{\boldmath$\hspace{-0.5pt } { \scriptstyle \tilde \nabla}$}}}}^{2,0 } & = \frac{t - { \lambda}}{2t^2 } { \partial}_{\scriptscriptstyle \mathcal{t } } { c } & \sigma_0 { f_{\raisebox{0pt}{\mbox{\boldmath$\hspace{-0.5pt } { \scriptstyle \tilde \nabla}$}}}}^{1,1 } & = \frac{ik{\lambda}}{t \bar t } \big ( \theta - 2i { \partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f \big ) & \sigma_0 { f_{\raisebox{0pt}{\mbox{\boldmath$\hspace{-0.5pt } { \scriptstyle \tilde \nabla}$}}}}^{0,2 } & = - \frac{\bar t + { \lambda}}{2\bar t^2 } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } \bar { c}\end{aligned}\ ] ] on sections of the bundle @xmath27 over @xmath3 .
the calculation of the ( 2,0)-part proceeds as in the proof of and yields a first - order operator with symbols given by @xmath388 as claimed in the theorem .
the calculation of the ( 0,2)-part is completely analogous .
we have the conjugate statements to , @xmath389 } & = - 4i x''_{\bar { \partial}_{{{\scriptscriptstyle \mathcal{t } } } } \bar { c}(v , w ) } \qquad \text{and } \qquad \sigma_0 { \big [ \bar b(v ) , \bar b(w ) \big ] } = -2{\lambda}\bar { \partial}_{\scriptscriptstyle \mathcal{t } } \bar c(v , w ) , \end{aligned}\ ] ] and , @xmath390 - \bar g(w ) { \smash{\cdot}}dv[f ] } - 2 { \lambda}{\partial}_{\scriptscriptstyle \mathcal{t } } \bar { \partial}_{\scriptscriptstyle \mathcal{t } } f(v , w ) , \end{aligned}\ ] ] and these can be combined with the conjugate of to compute @xmath391 it only remains to compute the ( 1,1)-part , and we get @xmath392 } + 2\bar t\ , d_{\scriptscriptstyle \mathcal{t } } b(v',w '' ) - 2t\ , d_{\scriptscriptstyle \mathcal{t}}\bar b(v',w '' ) \\ & \qquad + 2\bar t { \big [ b(v ) , w''[f ] \big ] } + 2 t { \big [ \bar b(w ) , v'[f ] \big ] } . \end{aligned}\ ] ] by and the above expression for @xmath393 , the second - order symbol vanishes , @xmath394 } + 2 \bar t \,\sigma_2 \big ( d_{\scriptscriptstyle \mathcal{t } } b(v',w '' ) \big ) - 2 t \ ,
\sigma_2 \big ( d_{\scriptscriptstyle \mathcal{t } } \bar b(v',w '' ) \big ) \\ & = 4ik \ , \theta(v',w '' ) - 2i ( t + \bar t ) \;\ !
\theta(v',w '' ) \\ & = 0 .
\end{aligned}\ ] ] the same is true for the first - order symbol @xmath395 } + 2i \bar t \;\ !
\sigma_1 \big ( d_{\scriptscriptstyle \mathcal{t } } b(v',w '' )
\big ) - 2i t \;\ ! \sigma_1 \big ( d_{\scriptscriptstyle \mathcal{t } } \bar b(v',w '' ) \big ) \\ & \qquad + 2\bar t \ , \sigma_1 { \big [ b(v ) , w''[f ] \big ] } + 2t\ , \sigma_1 { \big [ \bar b(w ) , v'[f ] \big ] } \\ & = 4ik\;\ ! \delta \theta(v',w '' ) + 8ik\;\ ! \theta(v',w '' ) { \smash{\cdot}}df \\ & \qquad - 2i(t + \bar t ) \delta \theta(v',w '' ) - 4i(t + \bar t ) \theta(v',w '' ) { \smash{\cdot}}df \\ & \qquad - 4\bar t g(v ) { \smash{\cdot}}w''[f ] - 4 t \bar g(w ) { \smash{\cdot}}dv'[f ] \\ & \qquad + 4\bar t g(v ) { \smash{\cdot}}w''[f ] + 4 t \bar g(w ) { \smash{\cdot}}dv'[f ] \\ & = 0 .
\end{aligned}\ ] ] finally , the last statement of gives the zeroth - order part @xmath396 this completes the proof of the theorem . in case
the hitchin connection @xmath279 in the bundle @xmath284 over @xmath3 is projectively flat , the parallel translation maps along homotopic curves are equal up to scale .
thus , if the parameter space @xmath3 is simply connected , the hitchin connection gives a canonical identification of the projectivized quantum spaces associated with different complex structures . in this sense , the quantization is independent of the complex structure .
the following theorem is an immediate consequence of the explicit curvature calculations of and .
[ thm:2 ] the connections @xmath279 and @xmath356 , acting on @xmath284 and @xmath27 , respectively , are both projectively flat if and only if the holomorphic vector field on @xmath6 given by @xmath319 vanishes , for all vector fields @xmath7 on @xmath3 .
projective flatness amounts to the curvature being a two - form on @xmath3 with values in constant functions on @xmath6 .
in particular , it takes values in zeroth - order differential operators .
being proportional to the first - order symbol of the ( 2,0)-part of the curvature of both @xmath279 and @xmath356 , the vector field @xmath397 will of course vanish if either of these are projectively flat .
conversely , suppose that @xmath398 vanishes , which means that the curvature of @xmath279 is a zeroth - order operator , and since it preserves the subbundle @xmath284 of holomorphic sections of @xmath247 it takes values in holomorphic functions on @xmath6 . by assumption ,
such functions are constants , which proves that @xmath279 is projectively flat .
in particular , the form @xmath399 takes values in constant functions on @xmath6 , which is also the case for @xmath400 , as already noted in . by
, this proves that the hitchin - witten connection @xmath401 is projectively flat .
we observe that our main follows as an immediate corollary of the theorem above .
this ends the general discussion of the hitchin and the hitchin - witten connection . in the final section
, we will apply the results to the setting originally motivating the study of these connections .
is this section , we apply the results of previous sections to chern - simons theory with either compact or complex gauge group . to be specific ,
let @xmath161 denote the real lie group @xmath32 , sitting as the maximal compact subgroup of its complexification @xmath402 , which can be identified with @xmath33 .
indeed , if @xmath403 denotes the lie algebra of @xmath161 , consisting of skew - hermitian traceless matrices , then its complexification @xmath404 can be identified with the lie algebra @xmath405 of traceless matrices through the unique splitting of any complex matrix into hermitian and skew - hermitian parts .
furthermore , let @xmath406 be an invariant inner product on @xmath404 , such as @xmath407 , which is normalized so that @xmath408 \rangle$ ] represents an integral generator of @xmath409 , where @xmath410 is the maurer - cartan form on @xmath161 .
let @xmath411 be a closed surface of genus @xmath412 , and let @xmath413 be the surface obtained by puncturing @xmath414 at a point @xmath415 .
fix an element @xmath416 and a small loop @xmath417 around the puncture , and consider the moduli spaces @xmath418 of representations of the fundamental group mapping @xmath417 to the element @xmath419 in the common center of @xmath161 and @xmath402 . for @xmath420 ,
we restrict to reductive representations , which is indicated by a plus in the notation .
the subspaces @xmath421 and @xmath422 corresponding to irreducible connections are of particular relevance to us . for these spaces
, we have en embedding @xmath423 , since any irreducible unitary representation can only be conjugate to another such by a unitary transformation . the spaces @xmath6 and @xmath420 also have a gauge theoretic realization as moduli spaces of flat connections .
let @xmath424 denote the trivial principal bundle over @xmath413 with structure group @xmath161 .
fix an element @xmath425 such that @xmath426 , and denote by @xmath427 the space of flat connections on @xmath424 which are all equal to @xmath428 for some fixed polar coordinate system , with angular coordinate @xmath180 , in a disc around the puncture .
these connections are said to be in temporal gauge near the puncture and clearly have holonomy @xmath429 around @xmath417 .
if @xmath430 denotes the space of gauge transformations equal to the identity in a neighbourhood of the puncture , then the holonomy representations provide an identification @xmath431 similarly , let @xmath432 be the space of flat reductive connections on the trivial @xmath433-bundle @xmath434 over @xmath413 , and let @xmath435 be the space of gauge transformations , both spaces with restrictions similar to the above around the puncture .
the holonomy representation then gives an identification @xmath436 by a more careful selection of connections and gauge transformations , using exponential decay in weighted sobolev norms around the puncture @xcite , the subsets @xmath421 and @xmath422 corresponding to irreducible connections can be endowed with the structure of smooth manifolds .
the tangent space @xmath437}m^s_{{{\scriptscriptstyle { { \mathbbm{c}}}}}}$ ] at a connection @xmath438 , is given by the compactly supported first cohomology @xmath439 , with values in the adjoint bundle and exterior derivative @xmath440 induced by @xmath438 .
we can then define a complex symplectic form @xmath441 by @xmath442 , [ \beta ] ) = - 4\pi \int_{\sigma } \langle \alpha \wedge \beta \rangle , \ ] ] in complete analogy , the tangent space @xmath437}m^s$ ] is given by the cohomology @xmath443 .
the formula defines in this case a real symplectic form on @xmath423 , which is of course just the restriction of @xmath441 . if @xmath29 is a complex number with integer real part , @xmath444 , we wish to quantize the space @xmath422 with respect to the real symplectic form @xmath445 as well as the space @xmath421 equipped with the restriction of @xmath446 , which is given by @xmath447 .
suppose that @xmath448 is a connection , and that @xmath449 is a gauge transformation in @xmath435 .
since @xmath402 is simply connected and hence 2-connected , we can choose a homotopy @xmath450 \to g_{{\scriptscriptstyle { { \mathbbm{c}}}}}$ ] from the trivial gauge transformation to @xmath56 , keeping identity values fixed .
if @xmath451 \to \sigma_p$ ] denotes projection onto the first factor , we can pull back @xmath438 to a connection @xmath452 on @xmath453 $ ] and consider the chern - simons form @xmath454 \rangle , \ ] ] which will of course be complex - valued in general . using the complex number @xmath29 from before , we shall consider the real form @xmath455 and the chern - simons cocycle given by @xmath456}$ } } } \;\ ; \alpha_t ( \tilde a^{\tilde g } ) \big ) \:\ : \in \ : \operatorname{u}(1),\end{aligned}\ ] ] where @xmath457 denotes the gauge - transformed connection .
since @xmath438 is flat and in temporal gauge , the chern - simons form vanishes in a neighbourhood of @xmath458 $ ] and the integral converges .
furthermore , since the real part of @xmath31 is an integer , the expression is independent of the choice of homotopy @xmath57 and defines a map @xmath459 , which can be shown to satisfy the cocycle relation @xmath460 this cocycle can be used to lift the action by the gauge group @xmath435 on @xmath432 to the trivial line bundle @xmath461 by @xmath462 and the quotient defines a hermitian line bundle , @xmath463 over the smooth part of the moduli space .
furthermore , this bundle comes with a unitary connection @xmath224 , given on the trivial bundle @xmath461 by the one - form @xmath464 for a tangent vector @xmath465 at @xmath448 .
it is easily verified that the curvature of this connection is given by @xmath466 so that @xmath467 defines a prequantum line bundle over @xmath422 , with symplectic form @xmath446 , and clearly this restricts to a prequantum line bundle @xmath247 over @xmath468 , equipped with @xmath447 as its symplectic structure . to perform geometric quantization
, we must introduce a polarization .
this will come from a choice of riemann surface structure on @xmath411 , which amounts to a hodge star - operator @xmath469 , satisfying @xmath470 for dimensional reasons .
using hodge theory to identify @xmath471 with the space of harmonic forms , we define an almost complex structure @xmath10 on @xmath422 by the expression @xmath472 on harmonic representatives , the space of which is preserved by @xmath473 .
it is easily checked that @xmath10 is compatible with @xmath446 .
there is another obvious almost complex structure @xmath169 on @xmath422 , which is simply given by @xmath474 , and clearly this anti - commutes with @xmath10 .
both of these almost complex structure are in fact integrable , as seen through the correspondence with higgs bundles discussed below , giving @xmath422 the structure of a hyperkhler manifold .
the khler structure @xmath10 only depends on the riemann surface structure up to isotopy , so in fact we get a family @xmath475 of khler structures parametrized by the teichmller space @xmath476 of the surface . at a unitary connection @xmath438 , the family @xmath10 clearly preserves the real subspace @xmath477 tangent to @xmath421
, so @xmath10 also defines a family of khler structures on @xmath421 .
this is indeed the polarization we shall use for quantizing @xmath421 , but for @xmath422 we shall instead rely on a certain real polarization , considered by witten in @xcite , which also depends on the riemann surface structure of @xmath411 .
let us first recall how the quantization of the moduli space @xmath421 for compact gauge group proceeds .
the moduli space @xmath421 is simply connected , so in particular @xmath478 vanishes , and furthermore @xmath479 ( see @xcite ) . for a given point @xmath480 in teichmller space , the properties of @xmath481 as a complex manifold can be understood through the classical work of narasimhan and seshadri @xcite , which identifies @xmath6 with the moduli space of s - equivalence classes of semi - stable holomorpic vector bundles , over the riemann surface @xmath411 , of rank @xmath78 , degree @xmath93 and fixed determinant .
this space has the structure of an normal projective algebraic variety which is typically singular but contains the moduli space of stable bundles as an open smooth subvariety , corresponding exactly to the manifold @xmath421 of irreducible connections . in this picture , the holomorphic tangent space @xmath482}m^s$ ] at a stable holomorphic bundle @xmath76 is given by the cohomology @xmath483 , with values in the traceless endomorphisms . as we saw above
, the moduli space @xmath421 admits a prequantum line bundle @xmath13 , which is just the determinant line bundle in this picture .
it generates the picard group , so in particular @xmath484 $ ] is a generator of @xmath485 , and @xmath486 $ ] with @xmath487 as proved in @xcite . if the rank @xmath78 and the degree @xmath93 are coprime , there are no strictly semi - stable bundles , so @xmath488 is a compact khler manifold . in this case , clearly @xmath489 , but as observed by hitchin @xcite , this holds even for non - coprime @xmath78 and @xmath93 by the hartogs theorem , since the complement of @xmath421 in @xmath6 has codimension at least 2 .
hitchin also notes that the family @xmath10 of khler structures on @xmath421 parametrized by teichmller space , which is itself in a canonical way a contractible complex manifold , is holomorphic and rigid in the sense of and . finally , the khler metric on the moduli space of stable bundles was studied by zograf and takhtajan in @xcite , where they give a ricci potential in terms of the determinant of the laplacian on the endomorphism bundle of a stable bundle .
altogether , the discussion above demonstrates that the moduli space @xmath421 satisfies all the conditions of to ensure the existence of a hitchin connection .
furthermore , in the case of coprime @xmath78 and @xmath93 , narasimhan and ramanan @xcite have shown that @xmath421 does not admit any holomorphic vector fields , so implies that the hitchin connection must be projectively flat . projective flatness also holds in the non - coprime case , which similarly does not admit holomorphic vector fields .
hitchin @xcite proves this by regarding a holomorphic vector field on @xmath421 as a holomorphic function on the cotangent bundle @xmath490 , which sits inside the moduli space of semi - stable higgs bundles @xmath491 , as discussed below . once again appealing to the hartogs theorem ,
the function , which is homogeneous of degree 1 in the action of @xmath492 and constant along fibers of the hitchin fibration , can be extended to @xmath491 , which is quite easily seen not to support such functions .
the proof does not apply to the special situation when the genus @xmath56 and the rank @xmath78 are both equal to 2 , and the holonomy around @xmath417 is trivial , because the moduli space @xmath420 is isomorphic to @xmath493 and clearly has holomorphic vector fields .
in summary , we have proved projective flatness of the hitchin connection for the moduli spaces as claimed in .
our proof of projective flatness follows the original proof by hitchin @xcite from the point where the curvature is known to be at most order one , but vanishing of the higher - order terms was established by hitchin through particular properties of the hitchin integrable system @xcite , whereas we derive it from rigidity of the family of khler structures .
we now turn to the problem of quantizing the moduli space @xmath422 of irreducible @xmath402-connections .
the theory of higgs bundles naturally enters in this discussion as well .
suppose again that @xmath411 is endowed with a riemann surface structure and recall that a higgs bundle is a pair @xmath494 , where @xmath495 is a holomorphic vector bundle and the higgs field @xmath496 is a holomorphic one - form with values in the traceless endomorphisms of @xmath76 .
stability for higgs bundles is defined by imposing the usual stability condition on the slope of subbundles @xmath497 , but only the invariant ones satisfying @xmath498 .
this leads to the moduli spaces @xmath491 and @xmath499 , of semi - stable , respectively stable , higgs bundles of rank @xmath78 , degree @xmath93 and fixed determinant .
nitsure @xcite proves that @xmath491 is a quasi - projective algebraic variety , which contains @xmath499 as an open smooth subvariety . through the work of hitchin , simpson , donaldson and corlette @xcite ,
the moduli space @xmath491 can be identified , via the hitchin equations and non - abelian hodge theory , with the moduli space @xmath420 of flat reductive @xmath402-connections , with stable bundles corresponding to irreducible connections .
if @xmath76 is itself already a stable bundle , then obviously stability as a higgs bundle is implied for any higgs field @xmath500 . in particular , by simply taking the higgs field to be zero
, the moduli space @xmath421 sits canonically inside @xmath501 , as we discussed already at the level of representations .
but in fact , the entire cotangent bundle of @xmath421 can be embedded in @xmath422 , although the embedding depends crucially on the riemann surface structure of @xmath411 .
indeed , we get the following identification by serre duality , @xmath502 where we recognize the right - hand side as the holomorphic cotangent space of the moduli space @xmath421 . in other words ,
the higgs fields on stable bundles can be viewed as cotangent vectors to @xmath421 through serre duality .
the higgs bundle model @xmath499 of @xmath503 illuminates the hyperkhler structure but it also carries a natural action of @xmath492 by scaling the higgs field . to quantize the modulu space @xmath422 , we follow witten @xcite and define a family of real polarizations , parametrized by the teichmller space of @xmath411 , in the following way . for any riemann surface structure on @xmath411 , in the form of a star - operator @xmath473 as usual , we can use hodge theory to split the space @xmath504 into types , @xmath505 although this uses the complex structure on @xmath504 , it defines a splitting of the underlying real space , which is a model for the real tangent space @xmath437}m^s_{{\scriptscriptstyle { { \mathbbm{c}}}}}$ ] . since the symplectic form @xmath441 is invariant under the hodge - star , the same of course applies to @xmath446 , so that each of the summands are lagrangian
. we shall take the subspaces @xmath506 as a polarization , and define the quantum space to be the polarized sections of @xmath507 , which are covariantly constant along its leaves .
notice that the polarization at a unitary connection @xmath438 is transverse to the tangent space @xmath437}m^s = h^1_a(\sigma , { \mathfrak{g}})$ ] , simply because the only real form of type ( 1,0 ) is the zero form .
this means that a polarized section is determined by its values on the space @xmath421 , or in other words , that the quantum space is identified with the prequantum space @xmath508 of smooth sections over @xmath421 , although the identification depends on the complex structure on @xmath411 .
surely this identification requires covariantly constant sections to exist on the leaves of the foliation , which would be guaranteed for instance if they were simply connected , but we shall not deal with this question here . instead , the discussion can be taken as motivation for using @xmath508 as a model for the quantum space of @xmath422 .
to understand how the identification between the quantum space of @xmath422 and the prequantum space @xmath508 depends on the riemann surface structure , we consider the latter as the fiber of a trivial bundle over teichmller space , @xmath509 then the expression defines a connection @xmath510 on this bundle , which is projectively flat by , once again due to the fact that the moduli space @xmath421 does not admit any holomorphic vector fields for any of the khler structures in the family @xmath10 parametrized by teichmller space .
this establishes the second statement of on the hitchin - witten connection .
we also expand the statement in the following theorem .
[ thm:3 ] for any @xmath29 with @xmath352 , the trivial bundle @xmath511 over teichmller space , with fiber given by the smooth sections of the chern - simons line bundle over the moduli space of flat @xmath32 connections , has a projectively flat connection given by @xmath512 ) - \frac{1}{2\bar t } ( \delta_{\bar g(v ) } \ ! + \!2 \nabla_{\ !
\bar g(v ) { \smash{\cdot}}df } \!- \ ! 2{\lambda}v''[f ] ) \!+ \ ! v[f ] , \end{aligned}\ ] ] for any vector field @xmath7 on teichmller space .
j. e. andersen and j. l. blaavand .
asymptotics of toeplitz operators and applications in tqft . in _ geometry and quantization _ , volume 19 of _ trav .
_ , pages 167201 .
luxembourg , 2011 .
j. e. andersen and n. l. gammelgaard .
hitchin s projectively flat connection , toeplitz operators and the asymptotic expansion of tqft curve operators . in _
grassmannians , moduli spaces and vector bundles _ , volume 14 of _ clay math .
_ , pages 124 .
soc . , providence , ri , 2011 .
g. d. daskalopoulos and r. a. wentworth .
geometric quantization for the moduli space of vector bundles with parabolic structure . in _ geometry ,
topology and physics ( campinas , 1996 ) _ , pages 119155 .
de gruyter , berlin , 1997 .
k. kodaira . , volume 283 of _ grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences]_. springer - verlag , new york , 1986 . translated from the japanese by kazuo akao , with an appendix by daisuke fujiwara .
g. b. segal . the definition of conformal field theory . in _
differential geometrical methods in theoretical physics ( como , 1987 ) _ , volume 250 of _ nato adv .
_ , pages 165171 .
kluwer acad .
, dordrecht , 1988 .
a. tsuchiya , k. ueno , and y. yamada .
conformal field theory on universal family of stable curves with gauge symmetries . in _
integrable systems in quantum field theory and statistical mechanics _ ,
volume 19 of _ adv .
pure math .
_ , pages 459566 . academic press , boston , ma , 1989 . | we give a direct calculation of the curvature of the hitchin connection , in geometric quantization on a symplectic manifold , using only differential geometric techniques .
in particular , we establish that the curvature acts as a first - order operator on the quantum spaces .
projective flatness follows if the khler structures do not admit holomorphic vector fields .
following witten , we define a complex variant of the hitchin connection on the bundle of prequantum spaces .
the curvature is essentially unchanged , so projective flatness holds in the same cases .
finally , the results are applied to quantum chern - simons theory , both for compact and complex gauge groups . |
we consider the unitary random matrix model @xmath0 defined on hermitian @xmath1 matrices @xmath2 in a critical regime where the limiting mean density of eigenvalues vanishes at an interior point .
it is a basic fact of random matrix theory @xcite that the eigenvalues of the random matrix ensemble ( [ 1-randommatrixmodel ] ) follow a determinantal point process with correlation kernel @xmath3 where @xmath4 denotes the @xmath5th degree orthonormal polynomial with respect to the weight @xmath6 on @xmath7 .
we assume in this paper that the confining potential @xmath8 in ( [ 1-randommatrixmodel ] ) is real analytic and that it satisfies the growth condition @xmath9 these assumptions ensure that the mean eigenvalue density @xmath10 has a limit as @xmath11 , @xmath12 , see e.g. @xcite , which we denote by @xmath13 .
it is known that @xmath14 is the density of the measure @xmath15 which minimizes the weighted energy @xmath16 among all probability measure on @xmath7 .
the measure @xmath15 is called the equilibrium measure in the external field @xmath17 .
the fact that @xmath17 is real analytic ensures that the support @xmath18 consists of a finite union of intervals @xcite .
it is a remarkable fact that local scaling limits of the kernel ( [ 1-kern ] ) depend only on the nature of the density @xmath14 .
this has been proved rigorously in the bulk of the spectrum for a quartic @xmath17 in @xcite and for general real analytic @xmath17 in @xcite . indeed , if @xmath19 , then @xmath20 exists , and @xmath21 the scaling limits are different at special points of the spectrum .
at edge points of the spectrum the density @xmath14 typically vanishes like a square root , and then it is known that for some constant @xmath22 , @xmath23 where @xmath24 and @xmath25 is the airy function .
the airy kernel is related to the tracy - widom distribution @xcite . in ( [ edge - scaling ] )
we have assumed that @xmath26 is a right edge point . for a left edge point
we change @xmath27 , @xmath28 in the left - hand side of ( [ edge - scaling ] ) .
other special points in the spectrum include * edge points where the density vanishes to a higher order .
the possible edge point behaviors ( at a right end point @xmath26 ) are @xmath29 where @xmath22 and @xmath5 is a non - negative integer .
* interior points where the density vanishes .
then @xmath30 where @xmath22 and @xmath5 is a positive integer . in these critical cases
it is believed that the local scaling limit at @xmath26 of the kernel only depends on the order of vanishing of the density at @xmath26 @xcite .
the case where @xmath14 vanishes quadratically at an interior point of @xmath31 , that is , the case @xmath32 in ( [ interior - vanishing ] ) , was considered by bleher and its @xcite for the case of a critical quartic potential @xmath33 , with @xmath34 and @xmath35
. then @xmath36,\ ] ] so that @xmath14 vanishes quadratically at the origin .
bleher and its consider the double scaling limit where @xmath37 changes with @xmath38 and tends to @xmath39 as @xmath40 in such a way that @xmath41 remains constant .
for the quartic potential this is equivalent to considering ( [ 1-randommatrixmodel ] ) where @xmath11 , @xmath12 , such that @xmath42 exists .
bleher and its gave a one - paremeter family @xmath43 of limiting kernels , depending on @xmath44 , so that for some @xmath22 , @xmath45 where @xmath46 is proportional to the value of the limit ( [ nn - to - infty ] ) .
the critical kernels are expressed in terms of so - called @xmath47-functions associated with the hastings - mcleod solution of the painlev ii equation @xcite .
consider as in @xcite the linear differential equations for a @xmath48-vector ( or @xmath49 matrix ) @xmath50 , @xmath51 where @xmath52 and @xmath53 the compatibility condition for ( [ psi - diffeqns ] ) is that @xmath54 satisfies the painlev ii equation @xmath55 and that @xmath56 .
we assume that @xmath57 is the hastings - mcleod solution of painlev ii , which is characterized by the asymptotic condition @xmath58 the critical kernels op @xcite are given by @xmath59 where @xmath60 is the special solution to ( [ psi - diffeqns ] ) which is real for real @xmath61 , satisfies @xmath62 and has asymptotics on the real line @xmath63 if we put @xmath64 then @xmath65 and @xmath66 is a special solution of the differential equations @xmath67 and @xmath68 the equations ( [ 1-diiffvglpsi1])-([1-diiffvglpsi2 ] ) for the @xmath47-functions correspond to the ones used by flaschka and newell @xcite and we will also use those in what follows .
the vector @xmath66 is the unique solution of ( [ 1-diiffvglpsi1 ] ) with asymptotics @xmath69 as @xmath70 uniformly in @xmath71 for any @xmath72 . before discussing our results
, we like to point out an integral formula for the kernel @xmath43 .
if we take a derivative of ( [ critk1 ] ) with respect to @xmath46 and use ( [ psi - diffeqns ] ) and ( [ psi - diffb ] ) , we get after some calculations @xmath73.\ ] ] using the deift / zhou steepest - descent method for @xmath74 as done in @xcite , one can show that @xmath75 as @xmath76 , so that we get @xmath77 d\sigma.\ ] ] since @xmath78 and @xmath79 are real for real @xmath61 , formula ( [ critk3 ] ) clearly shows that @xmath80 , as it should be .
it is the aim of this paper to show that the kernel @xmath43 is a universal limit . whenever the limiting mean eigenvalue density @xmath14 vanishes quadratically at an interior point , the correlation kernel @xmath81 has a double scaling limit given by ( [ 1-phikernel ] ) . in our theorem [ theorem-1-maintheorem ] below , we use the equilibrium measure @xmath82 of a compact set @xmath83 .
this is the unique probability measure on @xmath84 that minimizes the logarithmic energy @xmath85 among all borel probability measures @xmath86 on @xmath84 .
if @xmath84 is a single interval @xmath87 $ ] , then @xmath88 has a density @xmath89 given by @xmath90 if @xmath84 is a finite union of disjoint intervals , say @xmath91 $ ] with @xmath92
. then @xmath88 has density @xmath93 where @xmath94 is a monic polynomial of degree @xmath95 with exactly one zero in each of the gaps @xmath96 , @xmath97 , see e.g. ( * ? ? ?
* lemma 4.4.1 ) .
note that ( [ wsdensity ] ) has an extension to an analytic function in @xmath98 , where @xmath99 , which is a fact that we will use in what follows .
the following is our main result .
[ theorem-1-maintheorem ] let @xmath17 be real analytic on @xmath7 such that @xmath100 .
let @xmath14 be the density of the equilibrium measure in the external field , and let @xmath26 be an interior point of @xmath101 which is such that @xmath102 let @xmath11 such that the limit @xmath103 exists with @xmath104 .
let @xmath81 be the correlation kernel ( [ 1-kern ] ) for the eigenvalues of the random matrix model ( [ 1-randommatrixmodel ] ) .
then there exist constants @xmath22 and @xmath44 such that @xmath105 uniformly for @xmath106 in compact subsets of @xmath7 .
explicit formulas for the constants @xmath107 and @xmath46 are @xmath108 and @xmath109 where @xmath110 is the density of the equilibrium measure of @xmath31 . as noted before , bleher and its @xcite proved ( [ main1 ] ) for the case of a critical quartic @xmath17 .
see @xcite for a rigorous expansion of the free energy in this critical case .
the random matrix model ( [ 1-randommatrixmodel ] ) may be generalized to include a spectral singularity at the origin @xmath111 if @xmath112 and @xmath113 , then the scaled limit of the correlation kernels is a bessel kernel which involves bessel functions of order @xmath114 , see @xcite . in the multicritical case where @xmath14 vanishes quadratically at @xmath115 ,
an analogue of theorem 2.1 is valid . in work in progress @xcite
, we are considering the double scaling limit of ( [ spec - sing ] ) and we show that the limiting kernels are expressed in terms of the @xmath47-functions associated with a special solution of the general painlev ii equation @xmath116 the main ingredients in the proof of theorem [ theorem-1-maintheorem ] are equilibrium measures and riemann - hilbert problems .
we give some comments on both .
[ [ equilibrium - measures . ] ] equilibrium measures .
+ + + + + + + + + + + + + + + + + + + + + recall that the equilibrium measure in external field @xmath17 minimizes ( [ 1-minenergie ] ) .
we need to know how the equilibrium measure @xmath15 in the external field changes as a result of a change in @xmath17 .
the particular modification we consider here is @xmath117 so that @xmath118 .
let us put @xmath119 then it is known that @xmath120 and @xmath121 are increasing as a function of @xmath37 , see e.g. @xcite .
we also have the buyarov - rakhmanov formula @xcite @xmath122 which expresses the equilibrium measure in the external field as an average of equilibrium measures of sets .
a consequence of ( [ 1-buraformule ] ) is that @xmath123 which partly explains why the equilibrium measure @xmath124 plays a role in the formula ( [ main3 ] ) for @xmath46 . for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively.,title="fig : " ] for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively.,title="fig : " ] for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively.,title="fig : " ] for the case of interest in this paper we have that @xmath14 vanishes at @xmath129 .
then for @xmath130 , there is a positive density at @xmath26 , while for @xmath131 , @xmath26 is out of the support of @xmath132 . for @xmath37
slightly less than @xmath127 , there is a gap in @xmath133 , see figure [ figure - tweedeschaling ] . an asymptotic analysis based on the equilibrium measure
@xmath132 would require a discussion of the two different situations @xmath130 and @xmath131 , as is done in @xcite .
therefore we found it convenient to introduce a modification of the equilibrium problem in external field , which will enable us to treat both cases simultaneously .
the modification we make is that we do not require the measure to be non - negative in a neighborhood of the point @xmath26 . for a sufficiently small @xmath134
, we consider the problem to minimize @xmath135 among all signed measures @xmath136 on @xmath7 , where @xmath137 are nonnegative measures , such that @xmath138.\ ] ] we denote the minimizer by @xmath139 and we let @xmath140 . for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively ( compare with figure [ figure - tweedeschaling]).,title="fig : " ] for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively ( compare with figure [ figure - tweedeschaling]).,title="fig : " ] for @xmath125 with @xmath37 equal to @xmath126 , @xmath127 and @xmath128 , respectively ( compare with figure [ figure - tweedeschaling]).,title="fig : " ] then @xmath141 for @xmath142 , but for @xmath131 there is a clear distinction between @xmath139 and @xmath132 , see figure 2
. however we still have the analogue of ( [ 1-bura2 ] ) ( as we prove ) @xmath143 what s more , we also have ( [ 1-bura3 ] ) at the level of densities , that is , if @xmath144 denotes the density of @xmath139 , and if @xmath145 is an interior point of @xmath31 ( in particular if @xmath129 ) , then @xmath146 where @xmath110 is the density of the equilibrium measure of @xmath31 , which is what we need for the proof of theorem [ theorem-1-maintheorem ] .
note that we do not have ( [ 1-bura4 ] ) for @xmath129 if @xmath144 is the density of @xmath132 .
[ [ riemann - hilbert - problem . ] ] riemann - hilbert problem . + + + + + + + + + + + + + + + + + + + + + + + + the second main tool for the proof of theorem [ theorem-1-maintheorem ] is the characterization of orthonormal polynomials by means of a riemann - hilbert problem , due to fokas , its , and kitaev @xcite , and the subsequent asymptotic analysis of the riemann - hilbert problem by means of the deift / zhou steepest descent analysis of riemann - hilbert problems , introduced in @xcite , and further developed in @xcite , and other more recent papers .
the deift / zhou steepest descent analysis of the riemann - hilbert problem consists of a sequence of explicit transformations , which result in a riemann - hilbert problem that is explicitly solvable in terms of neumann series .
( in fact , in this paper , we only need the first term of this series . )
the critical point @xmath26 needs special attention .
of particular interest for us is the paper @xcite by baik , deift , and johansson on the length of the longest increasing subsequence of a random permutation of @xmath147 .
these authors show that the fluctuations of this random variable are distributed according to the tracy - widom distribution @xcite in the limit as @xmath40 .
one of the technical tools in this important paper is the asymptotic analysis of a riemann - hilbert problem on the unit circle which is related to an equilibrium measure ( also on the unit circle ) whose density vanishes at the point @xmath148 .
this situation is comparable to ours .
the authors of @xcite , see also subsequent papers @xcite , construct a local parametrix near @xmath148 with the aid of the @xmath47-functions associated with the hastings - mcleod solution of painlev ii .
these @xmath47-functions satisfy a model riemann - hilbert problem and the local parametrix is constructed by appropriately mapping the model riemann - hilbert problem onto a neighborhood of @xmath148 so that it satisfies certain desired jump properties approximately .
we follow the approach of @xcite but we introduce a modification in the construction of the local parametrix so that it has the desired jump properties _ exactly _ , in contrast to @xcite where the desired jump properties only hold _
the fact that we have the exact jump properties simplifies the arguments considerably and we feel that this is also a main contribution of the present paper .
[ [ outline - of - the - rest - of - the - paper . ] ] outline of the rest of the paper .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in section 3 we collect the necessary facts about equilibrium measures .
in particular we study the modified equilibrium problem with external field in some detail . in section 4
we discuss the riemann - hilbert problem satisfied by the @xmath47-functions . here
we follow @xcite .
then in section 5 we state the riemann - hilbert problem for orthogonal polynomials , discuss the relation with the correlation kernel @xmath81 , and perform the transformations in the steepest descent analysis .
finally in section 6 we give the proof of theorem [ theorem-1-maintheorem ] .
as explained in the previous section , we consider a modification of the equilibrium problem where we drop the non - negativity condition in a small neighborhood of @xmath26 .
we take @xmath134 sufficiently small so that @xmath149 \setminus \{x^*\}.\ ] ] and we use @xmath139 to denote the signed measure that minimizes ( [ mod - energy1 ] ) under the conditions ( [ mod - energy2 ] ) .
we define @xmath150 the existence and uniqueness of @xmath139 follows as in @xcite .
let @xmath151 be the logarithmic potential of @xmath152
. then standard arguments of potential theory @xcite show that @xmath139 is the unique signed measure satisfying ( [ mod - energy1 ] ) with the property that @xmath153 , \\
\label{variational - nut2 } 2u^{\nu_t}(x ) + \frac{1}{t } v(x ) & \geq \ell_t , \qquad x \in \mathbb r.\end{aligned}\ ] ] for some constant @xmath154 . as before we use @xmath132 to denote the equilibrium measure in the external field @xmath155 .
this is a probability measure that satisfies for some constant @xmath156 , @xmath157 it is known that @xmath120 and @xmath133 are increasing with @xmath158 , see @xcite . for @xmath142
, we have @xmath159 which implies that @xmath141 in view of ( [ delta0 ] ) and the variational conditions ( [ variational - nut1])([variational - mut2 ] ) . for @xmath131
, we have that @xmath26 is outside the support of @xmath132 , and in fact the strict inequality in ( [ variational - mut2 ] ) holds for @xmath160 if @xmath131 . in general
there is the following inequality between @xmath132 and @xmath139 . for every @xmath161 we have @xmath162 let @xmath163 . from ( [ variational - nut1 ] ) , ( [ variational - nut2 ] ) , ( [ variational - mut1 ] ) , and ( [ variational - mut2 ] ) it follows that @xmath164 the potential @xmath165 is subharmonic on @xmath166 and since @xmath167 , it is subharmonic at infinity as well . by the maximum principle for subharmonic functions @xcite ,
the maximum of @xmath165 is attained in @xmath168 only . since @xmath169
we then have by ( [ ulambda6 ] ) that equality in ( [ ulambda7 ] ) holds for every @xmath170 , and so @xmath171 .
this implies ( [ mut - leq - nut ] ) .
it follows from ( [ mut - leq - nut ] ) that @xmath172 is empty .
since for @xmath37 slightly less than @xmath127 a gap opens in @xmath133 , which depends continuously on @xmath37 , see @xcite , it follows that for any given @xmath173 , there is @xmath174 such that @xmath175 , \qquad \mbox { for } t > t_0.\ ] ] this shows that for @xmath131 sufficiently close to @xmath127 , the definition of @xmath139 is independent of the choice of @xmath176 .
a very useful fact is that for real analytic @xmath17 , say @xmath17 is analytic in a neighborhood @xmath177 of the real line , the measures @xmath132 have densities @xmath178 which can be expressed in terms of the negative part of an analytic function in @xmath177 . indeed ,
if @xmath179 then it was shown in @xcite that @xmath180 where @xmath181 denotes the negative part of @xmath182 .
a consequence of ( [ tilde - psit ] ) is that @xmath133 is the closure of the set where @xmath182 is negative .
the arguments of @xcite can be readily extended to the signed measures @xmath139 , provided that @xmath183 $ ] for some @xmath184 .
we define @xmath185 and then the following holds .
[ prop2 - 1 ] there exists @xmath186 such that for every @xmath187 , the signed measure @xmath139 has a density @xmath144 and @xmath188 ^ 2 , \qquad \mbox { for } x \in s_t.\ ] ] in addition we have @xmath189 ^ 2 \qquad \mbox { for } x \in \mathbb r \setminus s_t.\ ] ] this follows as in ( * ? ? ?
* proposition 2.51 ) .
obviously , for @xmath142 we have @xmath190 . since @xmath191 has a double zero at @xmath26 , we see from ( [ qt1 ] ) that @xmath192 has a zero at @xmath26 of order four .
for @xmath37 slightly bigger than @xmath127 , this fourth order zero splits into two double zeros in the complex plane away from the real axis . for @xmath37 slightly less than @xmath127 , there is a difference in the behavior of the zeros of @xmath193 and @xmath182 .
indeed , @xmath182 has two simple real zeros near @xmath26 which are endpoints of the support of @xmath132 , and in addition there is a double real zero in between them .
on the other hand , we have that @xmath193 has two double real zeros near @xmath26 , which are endpoints of the support of @xmath194 .
the fact that @xmath195 has only double zeros near @xmath26 allows us to take an analytic square root , and in view of ( [ qt1 ] ) we choose it so that for @xmath196 , @xmath197 where the sign of the square root at @xmath129 is taken negative if @xmath131 and positive if @xmath130 .
the right - hand side of ( [ qt3 ] ) has an analytic extension to a neighborhood of @xmath26 , which is independent of the value of @xmath187 .
thus @xmath144 has an analytic extension to a fixed neighborhood of @xmath26 , which will also be denoted by @xmath144 .
recall from the discussion before the statement of theorem 2.1 , that the density @xmath110 of the equilibrium measure of @xmath31 also has an analytic extension to a neighborhood of @xmath26 , which we also denote by @xmath110 .
the remaining part of this section is devoted to the proof of the following proposition .
[ prop2 - 2 ] we have @xmath198 uniformly for @xmath199 in a neighborhood of @xmath26 .
we start with a lemma which contains a weaker form of ( [ diff - tpsi ] ) .
we have @xmath200 where @xmath124 is the equilibrium measure of @xmath31 .
buyarov and rakhmanov @xcite proved ( for a very general class of @xmath17 ) that @xmath201 and @xmath202 where @xmath203 is the equilibrium measure of the set @xmath204 see ( [ variational - mut1 ] ) with @xmath205 . for real analytic @xmath17 ,
the sets @xmath31 and @xmath206 differ by an at most finite number of points , so that @xmath207 .
thus by ( [ bura1 ] ) and ( [ bura2 ] ) we have @xmath208 now write @xmath209 in view of ( [ bura3 ] ) it suffices to prove that @xmath210 where the norm denotes the total variation of a signed measure .
we may assume that @xmath131 .
because of ( [ mut - leq - nut ] ) we have @xmath211 for @xmath131 , the support of @xmath194 is contained in an interval @xmath212 $ ] , with @xmath213 see ( * ? ? ?
* lemma 8.1(iii ) ) . from ( [ qt3 ] ) and
the fact that @xmath193 has two double zeros in @xmath212 $ ] which are the end - points of the support of @xmath194 , we then easily get that @xmath214 combining ( [ deltat1 ] ) and ( [ deltat2 ] ) , we find @xmath215 as @xmath216 , which by ( [ nut - mut2 ] ) implies ( [ nut - mut ] ) .
this completes the proof of the lemma .
we now give a characterization of @xmath14 , which will be of use in the proof of proposition [ prop2 - 2 ] .
[ lemmaclaim ] for @xmath217 we have @xmath218 where @xmath110 is the density of @xmath124 . by the sokhotski - plemelj formulas , see e.g. @xcite , we have that @xmath219 satisfies @xmath220 let @xmath221 $ ] and set @xmath222 where the square root is positive for @xmath223 . using ( [ sp1 ] ) and the fact that @xmath224 as @xmath225 , we see that @xmath226 for any polynomial @xmath227 .
suppose @xmath227 has degree at most @xmath95 . then ( [ sprongpfr ] ) and the fact that @xmath228 as @xmath229 imply that @xmath230 which we can rewrite for @xmath231 as @xmath232 by contour integration .
now we use ( [ eq : fz - lemma ] ) and ( [ sp2 ] ) to obtain @xmath233 and this holds for any polynomial @xmath227 of degree at most @xmath95 .
since we know by ( [ wsdensity ] ) that @xmath234 for some monic polynomial @xmath227 of degree @xmath95 , we get ( [ eq : psiv - lemma ] ) .
now we can give the proof of proposition [ prop2 - 2 ] .
* of proposition [ prop2 - 2 ] .
* by ( [ qt ] ) we have for @xmath231 , @xmath235 so that in view of ( [ diff - tnu ] ) we find , @xmath236 thus by ( [ qt3 ] ) for @xmath237 in a neighborhood of @xmath26 , @xmath238 by ( [ eq : psiv - lemma ] ) , the right - hand side of ( [ diff - tpsi2 ] ) is @xmath239 in case @xmath237 is real , and by analytic continuation it continues to be @xmath239 in a punctured neighborhood of @xmath26 .
thus ( [ diff - tpsi ] ) holds for @xmath237 in a neighborhood of @xmath26 , and then it easily follows for @xmath240 as well .
in this section we recall the rh problem associated with the second painlev equation , see @xcite .
we transform this rh problem into a rh problem that will be used in the next section to construct a parametrix around @xmath26 .
consider the differential equation of ( [ 1-diiffvglpsi1 ] ) , @xmath241 where @xmath242 is a @xmath243 complex matrix - valued function and @xmath46 , @xmath244 and @xmath245 are considered as parameters .
all solutions of ( [ 4-diffvglpsi ] ) are entire functions of @xmath61 . for @xmath246 ,
let @xmath247 be the sector @xmath248 there exists a unique solution @xmath249 of equation ( [ 4-diffvglpsi ] ) so that @xmath250 as @xmath251 in the sector @xmath247 .
here we use @xmath252 to denote the third pauli matrix .
there exist complex values @xmath253 , @xmath246 ( called stokes multipliers ) so that @xmath254 with @xmath255 and @xmath256 furthermore , we have @xmath257 now define the rays @xmath258 oriented away from the origin , and the matrix - valued function @xmath259 then @xmath242 satisfies the following rh problem * @xmath242 is analytic on @xmath260 .
* @xmath261 on @xmath262 , * @xmath263 as @xmath251 , * @xmath242 is bounded near @xmath115 .
[ fig : jumpspsialgemeen ] ( 100,100)(0,0 ) ( 77,57)@xmath264 ( 51,80)@xmath265 ( 14,72)@xmath266 ( 7,41)@xmath267 ( 33,18)@xmath268 ( 68,26)@xmath269 ( 50,50)(0,1)23 ( 50,50)(0,-1)23 ( 50,73)(0,1)27 ( 50,27)(0,-1)27 ( 50,50)(2,1)23 ( 50,50)(2,-1)23 ( 73,61.5)(2,1)27 ( 73,38.5)(2,-1)27 ( 50,50)(-2,1)23 ( 50,50)(-2,-1)23 ( 27,61.5)(-2,1)27 ( 27,38.5)(-2,-1)27 the stokes multipliers @xmath253 depend on @xmath46 , @xmath244 and @xmath245 . an isomonodromy deformation is a variation of these parameters such that the stokes multipliers remain constant .
flaschka and newell @xcite showed that the isomonodromy deformations are given by the painlev ii equation @xmath270 and @xmath271 .
any solution of the painlev ii equation is a meromorphic function with an infinite number of poles .
we write @xmath272 for the @xmath242 function ( [ def - psi ] ) with parameters @xmath46 , @xmath54 , @xmath273 , where @xmath244 is the hastings - mcleod solution and @xmath274 .
let @xmath275 be the set of poles of @xmath244
. then @xmath272 is defined and analytic for @xmath276 and for @xmath277 .
it is known that there are no poles on the real line @xcite .
the stokes multipliers corresponding to the hastings - mcleod solution are @xmath278 thus @xmath272 is analytic across the imaginary axis .
we reverse the orientation of @xmath279 and @xmath280 , and we define @xmath281 then @xmath272 solves the following rh problem , see also figure [ fig : jumpspsi ] . *
@xmath242 is analytic on @xmath282 , * @xmath283 on @xmath284 , + @xmath285 * @xmath286 on @xmath287 , * @xmath288 as @xmath251 , * @xmath242 is bounded near @xmath115 .
the rh problem has a solution if and only if @xmath277 . the properties ( c ) and ( d ) are valid uniformly for @xmath46 in compact subsets of @xmath289 .
( 100,50)(0,0 ) ( 17,2)@xmath290 ( 17,46)@xmath291 ( 64,2)@xmath290 ( 67,46)@xmath291 ( 0,0)(2,1)33.1 ( 33.33,16.665)(2,1)36 ( 69.5,34.72)(2,1)31 ( 0,50)(2,-1)33.1 ( 33.33,33.33)(2,-1)36 ( 69.5,15.275)(2,-1)31 we modify the rh problem for @xmath242 so that it resembles the rh problem that we will need locally near @xmath26 .
we introduce an additional parameter @xmath292 and define @xmath293 for any given @xmath277 and @xmath294 , we then have that @xmath2 is defined for @xmath295 , see figure [ fig : jumpsm ] , and satisfies the following rh problem * @xmath296 is analytic for @xmath297 and @xmath277 . *
@xmath298 for @xmath299 , * @xmath300 for @xmath301 , * @xmath302 for @xmath303 . * @xmath304 as @xmath305 in the upper half plane , + @xmath306 * @xmath307 as @xmath70 in the lower half plane , + @xmath308 * @xmath309 is bounded for @xmath61 near @xmath115 . the properties ( c1 ) ,
( c2 ) , and ( d ) hold uniformly for @xmath46 in compact subsets of @xmath310 and for @xmath292 .
( 140,50)(-20,0 ) ( 30,6)@xmath311 ( 31.6,42)@xmath312 ( -19,23.6)@xmath313 ( 101,23.6)@xmath313 ( 0,0)(2,1)33.1 ( 30,8)(-2,1)6 ( 70.5,8)(2,1)6 ( 29.8,43)(-2,-1)7(70.7,43)(2,-1)7 ( 0,0)(2,1)33.1 ( 33.33,16.665)(2,1)36 ( 69.5,34.72)(2,1)31 ( 0,50)(2,-1)33.1 ( 33.33,33.33)(2,-1)36 ( 69.5,15.275)(2,-1)31 ( 0,25)(1,0)33.1 ( 33.33,25)(1,0)36 ( 69.5,25)(1,0)31
the proof of theorem [ theorem-1-maintheorem ] is based on the steepest descent analysis of the riemann - hilbert problem for orthogonal polynomials . since the main point of
the present discussion is the treatment of the critical point @xmath26 , we will restrict ourselves to the one - interval case .
we also assume that there are no other singular points besides @xmath26 .
thus @xmath314 $ ] and @xmath14 vanishes like a square root at @xmath315 and @xmath316 , and quadratically at @xmath26 but at no other points of @xmath31 .
in addition , we assume that the inequality in the variational condition ( [ variational - mut2 ] ) is strict for @xmath317 $ ] .
we then have that @xmath318\ ] ] for certain @xmath319 for every @xmath37 close to @xmath320 .
note that @xmath321 is an increasing and @xmath322 a decreasing function of @xmath37 .
we will comment below on the modifications that have to be made in the multi - interval case , see remark [ remark-5-vereenvoudiging ] . for each @xmath38 and @xmath323
, we consider a riemann - hilbert problem , introduced by fokas , its and kitaev @xcite .
we will look for a @xmath243 matrix - valued function @xmath324 ( we drop the subscripts for simplicity ) that satisfies the following conditions : * @xmath325 is analytic in @xmath326 , * @xmath327 for @xmath328 , + @xmath329 * @xmath330 as @xmath229 .
here @xmath331 ( resp .
@xmath332 ) denotes the limit as we approach @xmath333 from the upper ( resp .
lower ) half - plane .
the rh problem possesses a unique solution which is given by @xmath334 -2\pi i \kappa_{n-1,n } p_{n-1,n}(z ) & { \displaystyle}-\kappa_{n-1,n } \int_{\mathbb r}\frac{p_{n-1,n}(s ) e^{- n v(s)}}{s - z}ds \end{pmatrix } \label{3-fik}\ ] ] where @xmath335 denotes the orthonormal polynomial as before , see also @xcite .
the correlation kernel ( [ 1-kern ] ) can be expressed directly in terms of the solution of the rh problem . indeed , by the christoffel - darboux formula for orthogonal polynomials @xmath336 which involves the orthogonal polynomials of degrees @xmath38 and @xmath95 only . using ( [ 3-fik ] ) and the fact that @xmath337 for every @xmath338 , we then get @xmath339 applying the deift - zhou steepest - descent method to the rh problem will allow us to find the asymptotics of @xmath81 as given in theorem [ theorem-1-maintheorem ] .
we note that an expression like ( [ 3-knny ] ) was first given in the context of random matrices with external source @xcite where the eigenvalue correlation kernel is given in terms of the solution of a @xmath340 matrix valued rh problem , and the expression similar to ( [ 3-knny ] ) was found to be very convenient for asymptotic analysis .
also in the present @xmath243-case we find it helpful to work with ( [ 3-knny ] ) . in the first transformation we normalize the rh problem at infinity .
a standard approach would be to use the equilibrium measure @xmath341 in external field @xmath155 where @xmath342 .
if we would do this for the case where @xmath343 , we would have an equilibrium measure with a gap in the support around @xmath26 , and an annoying consequence is that the equality in the variational conditions is not valid near @xmath26 .
for this reason we have introduced the signed measures @xmath139 in section 2 and we will use these measures now to normalize the rh problem .
we let @xmath342 and assume @xmath37 is sufficiently close to @xmath127 so that ( [ eq : st - interval ] ) holds . as before we use @xmath344 to denote the density of @xmath139 .
then we define the @xmath345-function @xmath346 where we take the branch cut of the logarithm along the negative real axis .
then the following properties of @xmath347 are easy to check .
* @xmath348 is analytic in @xmath349 $ ] , * @xmath350 as @xmath351 .
* @xmath352 for @xmath353 , * @xmath354 for @xmath355 $ ] , * @xmath356 for @xmath357 $ ] . in terms of the analytic function @xmath193 , see ( [ qt ] ) and proposition [ prop2 - 1 ] , we have @xmath358,\ ] ] and @xmath359 @xmath360 where @xmath361 is analytic for @xmath362 $ ] and that square root is taken which is positive for large real @xmath46 .
we define @xmath363 and @xmath364 which are defined and analytic in the neighborhood @xmath177 of the real line where @xmath17 is analytic with cuts along @xmath365 and @xmath366 , respectively .
note that @xmath193 has simple zeros in @xmath321 and @xmath322 and only double zeros in @xmath367 $ ] , so that @xmath368 is indeed analytic there .
it is possible that @xmath193 has double real zeros , and that @xmath368 has sign changes in @xmath369 or @xmath370 .
however , by ( [ gt - prop2 ] ) and ( [ gt - prop3 ] ) we have that @xmath371 for @xmath372 and @xmath373 for @xmath374 , since we are in a situation with strict inequality in ( [ variational - nut2 ] ) for @xmath372 and @xmath374 .
we will now perform the first transformation of the rh problem : we take @xmath342 and define @xmath375 using the jump condition ( [ rhpyb ] ) of @xmath325 and ( [ 3-ty ] ) , we easily check that @xmath376 for @xmath377 , where @xmath378 because of the properties ( [ gt - prop1])([gt - prop3 ] ) and the definitions ( [ def - phi])([def - varphi ] ) , we see that the jump matrix @xmath379 has the following forms on the respective intervals @xmath380 , @xmath370 , and @xmath381 , @xmath382 thus @xmath383 is the unique solution of the rh problem * @xmath383 is analytic in @xmath384 , * @xmath385for @xmath386 , * @xmath387 . the jump matrix @xmath379 on the interval @xmath388 has the factorization @xmath389 now we open lenses around the intervals @xmath390 and @xmath391 as shown in figure [ figure - lens ] . let @xmath392 be the upper lips of the lenses and @xmath393 the lower lips , with orientation as in figure [ figure - lens ] .
we open the lenses in such a way that they are fully contained in @xmath177 , the region of analyticity of @xmath17 and @xmath193 .
in addition , we can take @xmath392 and @xmath393 in such a way that @xmath394 on @xmath392 and @xmath393 , with the exception of a neighborhood of @xmath26 if @xmath131 .
this effect is due to the fact that @xmath139 has a negative density near @xmath26 if @xmath131 .
however , if @xmath131 increases to one , the exceptional neighborhood shrinks to a point .
it follows that for any given @xmath395 , there is a constant @xmath396 such that for @xmath37 sufficiently close to @xmath127 we have @xmath397 for all @xmath398 with @xmath399 .
we define @xmath400 the rh problem for @xmath383 and the factorization ( [ 3-factorisatie1 ] ) imply that @xmath84 solves the following rh problem : * @xmath84 is analytic in @xmath401 * @xmath402 for @xmath403 , + @xmath404 * @xmath405 as @xmath229 , * @xmath84 remains bounded near @xmath321 , @xmath322 , and @xmath26 , where @xmath406 is given by @xmath407 \begin{pmatrix } 1 & e^{-2n \tilde{\varphi}_t } \\ 0 & 1 \end{pmatrix } & \textrm { on } ( -\infty , a_t ) , \\[3ex ] \begin{pmatrix } 0 & 1 \\ -1 & 0 \\ \end{pmatrix } & \textrm { on } ( a_t , b_t),\\[3ex ] \begin{pmatrix } 1 & 0 \\ e^{2n \varphi_t(z ) } & 1
\\ \end{pmatrix } & \textrm { on } c_1 \cup c_2 , \end{array } \end{cases}\label{3-sprongvs}\ ] ] the jump matrices on @xmath408 and on @xmath409 tend to the identity matrix as @xmath40 and @xmath410 . ignoring these jumps
, we find the parametrix for the outside region .
uniform convergence breaks down in neighborhoods of @xmath321 , @xmath322 , and @xmath26 , so that we will also need to construct local parametrices near those points .
the outside parametrix @xmath411 solves the following rh problem * @xmath412 is analytic in @xmath413 $ ] , * @xmath414 on @xmath380 , * @xmath415 as @xmath229 .
as in @xcite it has the solution @xmath416,\ ] ] where @xmath417.\ ] ] note that @xmath418 depends on @xmath37 . at edge points @xmath321 and @xmath322 the density @xmath144 vanishes like a square root .
this allows the construction of local parametrices near @xmath321 and @xmath322 with the use of airy functions
. we will not give the details , see @xcite . in a neighborhood @xmath419 of @xmath26 , we want to construct a parametrix @xmath420 with the following properties * @xmath420 is analytic in @xmath421 , * @xmath422 on @xmath423 , * @xmath424 as @xmath425 , @xmath410 , uniformly for @xmath426 , * @xmath427 remains bounded for @xmath199 near @xmath26 .
we seek a parametrix @xmath420 near @xmath26 in the form @xmath428 where @xmath429 is analytic in @xmath430 , @xmath431 is a conformal map from @xmath430 to a neighborhood of @xmath115 , @xmath432 is analytic in @xmath430 , and @xmath433 is a real constant .
recall that @xmath2 is given by ( [ def - m ] ) . in view of the jump properties of @xmath2 , we seek @xmath431 , @xmath432 , and @xmath433 so that for @xmath434 , @xmath435 since @xmath436 , the right - hand side of ( [ mappings1 ] ) does indeed define an analytic function in @xmath430 .
the conformal map @xmath431 will not depend on @xmath37 . recall that @xmath437 has a zero of order four at @xmath26 , and that it is negative on @xmath438 .
thus @xmath439 is analytic in @xmath430 and has a zero of order three at @xmath26 .
we may assume there are no other zeros in @xmath430 .
then we can take a third root and define @xmath440^{1/3 } , \qquad z \in \delta.\ ] ] then @xmath431 is analytic in @xmath430 with @xmath441 where @xmath107 is given by ( [ main2 ] ) .
taking smaller @xmath430 if necessary , we then have that @xmath431 is indeed a conformal map on @xmath430 .
note that @xmath442 and that @xmath431 is real and positive on @xmath443 and real and negative on @xmath444 .
we still have some freedom in opening the lenses .
we take @xmath392 and @xmath393 in @xmath430 so that @xmath431 maps them to the rays where @xmath2 has its jumps .
that is , @xmath445 is mapped into @xmath446 and @xmath447 , and @xmath448 is mapped into @xmath449 and @xmath450 .
having @xmath431 with @xmath442 , we take @xmath240 in ( [ mappings1 ] ) and we see that we should take @xmath451 then @xmath433 is real and it is also given by @xmath452 and so @xmath453 having @xmath431 and @xmath433 we finally take @xmath454 so that ( [ mappings1 ] ) holds , that is , @xmath455 where for the last line we used ( [ def - fz ] ) and ( [ def - theta2 ] ) . since the right - hand side of ( [ def - st ] ) is analytic in @xmath430 and vanishes for @xmath240 we can divide by @xmath456 ( which has a simple zero at @xmath457 ) and obtain an analytic function @xmath432 in @xmath430 . then @xmath454 is real for real @xmath199 , and @xmath458 .
so the above construction yields analytic functions @xmath431 , @xmath432 and a constant @xmath433 so that ( [ mappings1 ] ) holds .
then we can define the parametrix @xmath420 by ( [ 5-constructiep ] ) , provided that @xmath459 for @xmath434 stays away from the poles of the hastings - mcleod solution of painlev ii , since for @xmath460 we have that @xmath296 is not defined . for given values of @xmath38 and @xmath37 , we can take @xmath430 small enough so that this is indeed the case . however
, we want to let @xmath40 , @xmath410 so that @xmath461 , and work with a neighborhood @xmath430 that is independent of @xmath38 and @xmath37 , although it may depend on @xmath462 . note that by ( [ def - st ] ) and ( [ qt3 ] ) @xmath463 because of ( [ diff - tpsi ] ) we have @xmath464 as @xmath410 , uniformly for @xmath465 in a neighborhood of @xmath26 . using this in ( [ def - st2 ] ) we get that @xmath466 where @xmath467 is uniformly in @xmath199 as @xmath410 , and @xmath468 is uniformly in @xmath37 as @xmath469 . by ( [ f - near - xstar ] ) and ( [ def - st3 ] ) we then also have @xmath470 where again @xmath467 is uniformly in @xmath199 as @xmath410 , and @xmath468 is uniformly in @xmath37 as @xmath469 . then
if @xmath471 we get @xmath472 where we used the definition ( [ main3 ] ) of the constant @xmath46 .
since there are no poles on the real line , we can find a neighborhood @xmath430 of @xmath26 such that @xmath473 for all @xmath434 if @xmath38 is large enough and @xmath461 .
note that @xmath430 depends on @xmath462 , but not on @xmath38 and @xmath37 .
then @xmath474 is well - defined for @xmath199 in a fixed neighborhood @xmath430 of @xmath26 .
finally , we define @xmath429 in such a way that the matching condition at @xmath475 is satisfied .
we do this by defining @xmath476 so that @xmath477 is analytic near @xmath26 by the jump property ( [ rhpsinftyb ] ) of @xmath478 . because of the asymptotic behavior of @xmath296 in ( [ rhpmc1 ] ) and ( [ rhpmc2 ] ) ( which is valid uniformly for @xmath46 away from the poles and for @xmath292 ) we have the matching condition @xmath479 uniformly for @xmath480 .
this completes the construction of the parametrix @xmath420 in the neighborhood of @xmath26 .
we set @xmath481 since @xmath84 and @xmath420 have the same jumps inside each of the disks , and @xmath84 and @xmath478 have the same jumps on @xmath482 $ ] , @xmath483 has jumps on a contour @xmath484 as shown in figure [ figure - contourr ] . after the third and final transformation . ]
@xmath483 solves the following rh problem : * @xmath483 is analytic in @xmath485 * @xmath486 as @xmath487 , * @xmath488 as @xmath229 , * @xmath483 is bounded , for certain jump matrices @xmath489 , which as @xmath40 , @xmath410 so that @xmath461 , satisfy @xmath490 as in @xcite it now follows that @xmath491 uniformly for @xmath492 .
now we are ready for the proof of theorem [ theorem-1-maintheorem ] .
we start with the expression ( [ 3-knny ] ) for the kernel @xmath81 . after the transformation ( [ 3-ty ] )
, we find that for @xmath493 , @xmath494 using formula ( [ 3-st ] ) for @xmath84 in the upper parts of the lenses , we get for @xmath495 , @xmath496 assume that @xmath145 and @xmath465 are inside the disk @xmath430 around @xmath26 .
then by ( [ 5-rsp ] ) , ( [ 5-constructiep ] ) , ( [ def - m ] ) , and ( [ mappings1 ] ) we get @xmath497 and so @xmath498 similarly @xmath499 now we fix @xmath500 and @xmath501 and take @xmath502 so that for @xmath38 large enough , @xmath145 and @xmath465 are inside the disk around @xmath26 , so that ( [ 6-kerns ] ) , ( [ sformula1 ] ) , and ( [ sformula2 ] ) hold
. then it follows from ( [ f - near - xstar ] ) and ( [ xy - in - uv ] ) that @xmath503 from ( [ def - st5 ] ) and ( [ xy - in - uv ] ) we get @xmath504 as @xmath40 , @xmath410 such that @xmath461 .
furthermore , from ( [ r - asymp ] ) , ( [ xy - in - uv ] ) , and the fact that @xmath483 is analytic near @xmath26 , we get @xmath505 from ( [ def - sinfty ] ) , ( [ defet ] ) , and ( [ xy - in - uv ] ) we easily get @xmath506 as @xmath40 .
the constants implied by the @xmath507-symbols in ( [ r - limits ] ) and ( [ et - limits ] ) are independent of @xmath500 and @xmath501 , when @xmath500 and @xmath501 are restricted to a compact subset of @xmath7 . combining ( [ r - limits ] ) and ( [ et - limits ] ) we get that @xmath508 since @xmath433 is real . then multiplying ( [ sformula2 ] ) and ( [ sformula1 ] ) and letting @xmath40 , @xmath410 such that @xmath471 , we get by using ( [ 6-kerns ] ) , ( [ xy - in - uv ] ) , ( [ f - limits ] ) , ( [ st - limits ] ) , and ( [ randet - limits ] ) that @xmath509 uniformly for @xmath500 and @xmath501 in compact subsets of @xmath7 . by ( [ 1-asymptotiekphi ] ) , ( [ psij ] ) and ( [ def - psi ] ) we have that @xmath510 in view of the jump ( [ rhppsib1 ] ) satisfied by @xmath272 on @xmath284 we have that @xmath511 since @xmath512 , we also get ( after simple calculation ) @xmath513 note that @xmath514 includes the full real line , so that we can take @xmath515 in ( [ phi - in - psi1 ] ) and @xmath516 in ( [ phi - in - psi2 ] ) which we use in ( [ sformula3 ] ) to obtain ( [ main1 ] ) .
this completes the proof of theorem [ theorem-1-maintheorem ] .
[ remark-5-vereenvoudiging ] the above proof of theorem [ theorem-1-maintheorem ] was given under the assumption that @xmath31 consists of one interval , and that there are no other singular points except for @xmath26 . here
singular point refers to the classification of @xcite according to which there are three types of non - regular behavior for a real - analytic external field @xmath17 , see also @xcite .
the singular points of type i are points in @xmath517 where equality in the variational inequality ( [ variational - mut2 ] ) with @xmath205 holds .
singular points of type ii are interior points of @xmath31 where the density @xmath14 vanishes , and singular points of type iii are edge points of @xmath31 where @xmath14 vanishes to higher order than a square root . here
we indicate briefly the modifications that have to be made if these assumptions are not satisfied .
if @xmath31 consists of more than one interval , then the main complication is that the construction of the outside parametrix @xmath418 is more complicated , since it uses @xmath518-functions as in ( * ? ? ?
* lemma 4.3 ) .
it should be noted that @xmath418 will also depend on @xmath38 . as a result
it will follow that @xmath429 as defined in ( [ defet ] ) also depends on @xmath38 .
however this will not effect the asymptotic behavior ( [ et - limits ] ) as @xmath40 , so that the above proof goes through . in case
there are other singular points , we have to construct special local parametrices around each of them . for our purposes , we only need to know the existence of those parametrices , which is established in ( * ? ? ?
* section 5 ) .
we thank maarten vanlessen for useful remarks .
the authors are supported by fwo research projects g.0176.02 and g.0455.04 .
the second author is also supported by k.u.leuven research grant ot/04/24 , by intas research network necca 03 - 51 - 6637 , by nato collaborative linkage grant pst.clg.979738 , by grant bfm2001 - 3878-c02 - 02 of the ministry of science and technology of spain and by the european science foundation program methods of integrable systems , geometry , applied mathematics ( misgam ) and the european network in geometry , mathematical physics and applications ( enigma ) . 50 a.i .
aptekarev , p.m. bleher , and a.b.j .
kuijlaars , large @xmath38 limit of gaussian random matrices with external source , part ii , preprint arxiv : math - ph/0408041 .
g. akemann , p.h .
damgaard , u. magnea , and s.m .
nishigaki , multicritical microscopic spectral correlators of hermitian and complex matrices , nuclear phys .
b 519 ( 1998 ) , 682714 .
j. baik , p. deift , and k. johansson , on the distribution of the length of the longest increasing subsequence of random permutations , j. amer .
soc . 12 ( 1999 ) , 11191178 .
j. baik , p. deift , and k. johansson , on the distribution of the length of the second row of a young diagram under plancherel measure , geom .
10 ( 2000 ) , 702731 .
j. baik , p. deift , and e. rains , a fredholm determinant identity and the convergence of moments for random young tableaux , comm . math .
( 2001 ) , 627672 .
j. baik and e. rains , limiting distributions for a polynuclear growth model with external sources , j. stat .
100 ( 2000 ) , 523541 .
j. baik and e. rains , the asymptotics of monotone subsequences of involutions , duke math .
j. 109 ( 2001 ) , 205281 .
p. bleher and b. eynard , double scaling limit in random matrix models and a nonlinear hierarchy of differential equations , j. phys .
a 36 ( 2003 ) , 30853105 .
p. bleher and a. its , semiclassical asymptotics of orthogonal polynomials , riemann - hilbert problem , and universality in the matrix model , annals math . 150 ( 1999 ) , 185266 .
p. bleher and a. its , double scaling limit in the random matrix model : the riemann - hilbert approach , comm .
pure appl . math . 56 ( 2003 ) , 433516 .
p. bleher and a. its , asymptotics of the partition function of a random matrix model , preprint arxiv : math - ph/0409082 .
p.m. bleher and a.b.j .
kuijlaars , random matrices with external source and multiple orthogonal polynomials .
internat . math .
notices 2004 ( 2004 ) , 109129 .
p. bleher and a.b.j .
kuijlaars , large @xmath38 limit of gaussian random matrices with external source , part i , comm .
252 ( 2004 ) , 43 - 76 .
buyarov and e.a .
rakhmanov , families of equilibrium measures in an external field on the real axis , sb .
( 1999 ) , no .
6 , 791802 .
t. claeys , a.b.j .
kuijlaars , and m. vanlessen , in preparation .
damelin and a.b.j .
kuijlaars , the support of the equilibrium measure in the presence of a monomial external field on [ -1,1 ] , trans .
( 1999 ) , 45614584 .
p. deift , orthogonal polynomials and random matrices : a riemann - hilbert approach , courant lecture notes in mathematics 3 , amer . math .
soc . , providence , ri , 1999 .
p. deift , t. kriecherbauer , and k.t - r mclaughlin , new results on the equilibrium measure for logarithmic potentials in the presence of an external field , j. approx .
theory 95 ( 1998 ) , 388475 .
p. deift , t. kriechterbauer , k.t - r mclaughlin , s. venakides , and x. zhou , strong asymptotics of orthogonal polynomials with respect to exponential weights , comm .
pure appl . math .
52 ( 1999 ) , 14911552 .
p. deift , t. kriechterbauer , k.t - r mclaughlin , s. venakides , and x. zhou , uniform asymptotics for polynomials orthogonal with respect to varying exponential weights and applications to universality questions in random matrix theory , comm .
pure appl .
52 ( 1999 ) , 13351425 .
p. deift , s. venakides , and x. zhou , new results in small dispersion kdv by an extension of the steepest descent method for riemann - hilbert problems , internat .
notices 1997 ( 1997 ) , 286299 .
p. deift and x. zhou , a steepest descent method for oscillatory riemann - hilbert problems .
asymptotics for the mkdv equation , annals math . 137 ( 1993 ) , 295368 .
p. deift and x. zhou , asymptotics for the painlev ii equation , comm .
pure appl .
48 ( 1995 ) , 277337 .
h. flaschka and a.c .
newell , monodromy and spectrum - preserving deformations i , comm .
( 1980 ) , 65116 .
fokas , a.r .
its , and a.v .
kitaev , the isomonodromy approach to matrix models in @xmath48d quantum gravity , comm .
147 ( 1992 ) , 395430 .
fokas , a.r .
its , and a.v .
kitaev , discrete painlev equations and their appearance in quantum gravity , comm .
( 1991 ) , 313344 .
gakhov , boundary value problems , pergamon press , oxford , 1966 .
hastings , and j.b .
mcleod , a boundary value problem associated with the second painlev transcendent and the korteweg - de vries equation , arch . rational mech
73 ( 1980 ) , 3151 .
a.r . its and a.a .
kapaev , quasi - linear stokes phenomenon for the second painlev transcendent , nonlinearity 16 , ( 2003 ) , 363 - 386 .
its , and v.yu .
novokshenov , the isomonodromic deformation method in the theory of painlev equations , lecture notes in math .
springer - verlag , berlin - new york ( 1986 ) .
a.b.j . kuijlaars , riemann hilbert analysis for orthogonal polynomials , lecture notes in mathematics 1817 ( 2003 ) , 167210 .
a.b.j . kuijlaars and k.t - r mclaughlin , generic behavior of the density of states in random matrix theory and equilibrium problems in the presence of real analytic external fields , comm .
pure appl .
math . 53 ( 2000 ) , 736785 .
a.b.j . kuijlaars and m. vanlessen , universality for eigenvalue correlations at the origin of the spectrum , comm .
243 ( 2003 ) , 163191 .
mehta , random matrices , 2nd edition , academic press , boston , 1991 .
t. ransford , potential theory in the complex plane , cambridge university press , cambridge , 1995 .
saff and v. totik , logarithmic potentials with external fields , springer - verlag , berlin , 1997 .
h. stahl and v. totik , general orthogonal polynomials , cambridge university press , cambridge , 1992 . v. totik , weighted approximation with varying weight , lecture notes in mathematics 1569 , springer - verlag , berlin , 1994 . c. tracy and h. widom , level - spacing distributions and the airy kernel , comm .
( 1994 ) , 151174 . | we study unitary random matrix ensembles in the critical case where the limiting mean eigenvalue density vanishes quadratically at an interior point of the support .
we establish universality of the limits of the eigenvalue correlation kernel at such a critical point in a double scaling limit .
the limiting kernels are constructed out of functions associated with the second painlev equation .
this extends a result of bleher and its for the special case of a critical quartic potential .
the two main tools we use are equilibrium measures and riemann - hilbert problems . in our treatment of equilibrium measures
we allow a negative density near the critical point , which enables us to treat all cases simultaneously .
the asymptotic analysis of the riemann - hilbert problem is done with the deift / zhou steepest descent analysis . for the construction of a local parametrix at the critical point
we introduce a modification of the approach of baik , deift , and johansson so that we are able to satisfy the required jump properties exactly . |
ising and dimer models were among the first models in two - dimensional statistical mechanics for where the partition function and for some of the correlation functions were computed explicitly in terms of pfaffians of certain matrices .
for this reason both of these models can be regarded as theories of gaussian discrete two dimensional fermi field .
the ising model was solved by onsager and for dimer models the pfaffian solution was found by kasteleyn . the @xmath0-vertex model , a particular case of which is the ice model is interesting for a number of reasons . physically , it is a model of ferro- and antiferro- electricity .
it has many equivalent reformulations , one of them ( which we will use ) describe the 6-vertex configurations on a planar connected simply connected region in terms of stepped surfaces .
one of the combinatorial reformulations of the 6-vertex model for specific value of parameters @xcite is related to alternating sign matrices @xcite .
the 6-vertex model generalizes dimer models and can be regraded as the theory of gaussian discrete fermions with four fermionic interaction .
the partition function of the 6-vertex model with periodic boundary conditions was computed in @xcite using the bethe ansatz method .
the computation of correlation functions in the @xmath0-vertex model is highly non - trivial and to the large degree is still a challenge .
there are two known approaches to this problem based on the internal symmetry of the model , i.e. on the representation theory of quantum affine algebras .
first approach is based on determinantal formulae for certain matrix elements @xcite .
for some recent results based on this method see @xcite and @xcite .
the second approach is based on form - factor formulae derived in @xcite . for an overview of this approach see @xcite and for the latest results see @xcite . the 6-vertex model on a planar
simply regions can be reformulated as the theory of random stepped surfaces .
a configuration of arrows in a 6-vertex model can be interpreted as a configuration of paths which can be viewed as level corves of a _ height function _ defining a stepped surface .
gibbs measure on 6-vertex configurations define a gibbs measure on stepped surfaces . for certain class of such gibbs measures , random surfaces in the thermodynamical limit
develop the limit shape phenomenon @xcite also known as the acric circle phenomenon @xcite .
it means that on macroscopical scale the random surface becomes deterministic .
the fluctuations remain at smaller scale and the structure of fluctuations may change depending on how singular the limit shape is at this point . the limit shape phenomenon is studied in details in dimer models @xcite .
the numerical results from @xcite@xcite show how limit shapes develop in a 6-vertex model with domain wall boundary conditions . in this paper
we will focus on the limit shape phenomenon for the 6-vertex model on a planar connected simply connected regions with fixed boundary conditions .
computing the limit shape involves two steps .
step one is the derivation of the formula for the free energy of the 6-vertex model as a function of magnetic fields . unlike the partition function of dimer models
, the free energy can not be computed explicitly .
however , it can be written in terms of the solution to a linear integral equation .
many analytical properties of the partition function are known but scattered in the literature , see for example @xcite@xcite@xcite@xcite@xcite @xcite@xcite@xcite .
we collected most of them in the section [ f - energy ] together with some new results .
step two is to derive and solve the variational principle which determine the limit shape for given boundary conditions .
the free energy of the model as a function of magnetic fields determine the functional in the variational problem .
such variational problem was first introduced for dimer models in @xcite .
the idea of using this variational problem in the 6-vertex first appeared in @xcite where some interesting partial results were obtained for correlation functions in the bulk of the limit shape .
the structure of fluctuations near the limit shape is determined by the asymptotical behavior of correlation functions at smaller scales .
we will discuss this problem for the 6-vertex model in the last section .
finally let us mention the special case of domain wall boundary conditions .
these boundary conditions first appeared in the computation of norms of bethe vectors @xcite .
the remarkable fact about them is that the partition function of the 6-vertex model with these boundary conditions can be written as a determinant @xcite .
another remarkable fact is that exactly these boundary conditions relate the 6-vertex model with alternating sign matrices @xcite .
the large volume asymptotic of the partition function of the 6-vertex mode with these boundary conditions was computed in @xcite@xcite .
here is the outline of the paper . in the first two sections we recall some basic facts about the @xmath0-vertex model , about its reformulation it in terms of height functions , and about the thermodynamical limit in the model .
the third section contains the description of the free energy per site for as the function of electric fields in thermodynamical limit for periodic boundary conditions .
some asymptotical behaviors of the free energy are computed in section 4 .
this section is a combination of an overview and original results . in section 5
we study the asymptotical behavior of the limit shapes near the `` freezing point '' . in the last section
we discuss fluctuations .
we thank c. evans , r. kenyon , a. okounkov , and s. sheffield for interesting discussions .
the work of n.r . was supported by the nsf grant dms 0307599 , by the niels bohr initiative at aarhus university , by the humboldt foundation and by the crdf grant rumi1 - 2622 the work of k.p .
was supported by the nsf rtg grant and dms 0307599 .
first , let us fix the notation .
a square @xmath1 grid @xmath2 is a graph with @xmath3- and @xmath4-valent vertices embedded into @xmath5 such that @xmath3-valent vertices are located at points @xmath6 ( see fig .
[ dwbc_conf ] ) and @xmath4-valent vertices are located at @xmath7 and at @xmath8 .
an edge connecting two @xmath3-valent vertices is called an _ inner _ edge and an edge connecting a @xmath3-valent vertex with a @xmath4-valent vertex is called an _ outer _ edge .
states of the @xmath0-vertex model on @xmath2 are configurations of arrows assigned to each edge ( i.e. orientations of @xmath2 ) .
they satisfy the ice rule : at any vertex the number of incoming arrows should be equal to the number of outgoing arrows .
six possible configurations at a vertex are shown on fig .
[ vertices ] .
configurations of arrows on boundary edges are called boundary conditions .
each configuration of arrows on the lattice can be equivalently described as the configuration of `` thin '' and `` thick '' edges or empty and occupied edges shown on fig .
[ vertices ] .
there should be an even number of thick edges at each vertex as a consequence of the ice rule .
the thick edges form paths .
we assume that paths do not intersect ( two paths may meet at an @xmath9-vertex ) .
so , equivalently , configurations of the @xmath0-vertex model can be regarded as configurations of paths satisfying the rules from fig .
[ vertices ] . to each configuration of arrows on edges adjacent to a vertex we assign a boltzmann weight , which we denote by the same letters .
the physical meaning of a boltzmann weight is @xmath10 , where @xmath11 is the energy of a state and @xmath12 is the temperature ( in the appropriate units ) .
thus , all numbers @xmath9 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , and @xmath17 should be positive .
choosing the scale such that @xmath18 , it is natural to write boltzmann weights in the exponential form .
@xmath19 where @xmath20 , @xmath21 , and @xmath22 are dimensionless interaction energies of arrows at different types of vertices , and @xmath23 and @xmath24 are dimensionless horizontal and vertical components of the magnetic field , respectively . in this interpretation arrows are spins interacting with the magnetic field .
we set @xmath25 because for the types of boundary conditions we will consider the difference between the number of @xmath16-vertices and @xmath17 vertices is the same for all states and , therefore , the probability does not depend on the ratio @xmath26 .
we also use the standard notation @xmath27 these are the weights of the model when there is no magnetic field . the weight of a state is the product of weights of vertices in the state .
the weight of a state on @xmath28 ( up to a constant factor ) can be written in terms of energies and magnetic fields as @xmath29 where @xmath30 is the total number of @xmath31-vertices , @xmath32 is the total number of @xmath33-vertices , @xmath34 is the total number of @xmath35-vertices , @xmath36 is the total number of horizontal edges occupied by paths , and @xmath37 is the total number of vertical edges occupied by paths .
the partition function is the sum of weights of all states of the model @xmath38 where @xmath39 is one of the weights from fig .
[ vertices ] .
weights define the probabilistic measure on the set of states of the @xmath0-vertex model .
the probability of a state is given by the ratio of the weight of the state to the partition function of the model @xmath40 this is the gibbs measure of the @xmath0-vertex model .
types of vertices and the corresponding thin and thick edges configurations ] let us define the characteristic function of an edge @xmath41 as @xmath42 a local correlation function is the expectation value of the product of such characteristic functions : @xmath43 [ [ section ] ] let us fix arrows ( or , equivalently , think edges ) on outer edges of @xmath2 .
states in the 6-vertex with the same configurations of arrows on the boundary are called states with _ fixed
_ boundary conditions .
the difference between two such states can occur only at inner edges . the space of states with fixed boundary conditions is empty unless the boundary values satisfy the ice rule : the total number of incoming arrows on the boundary edges should be equal to the total number of outgoing arrows . in the path formulation
this means that the number of paths through north and west boundaries should be equal to the number paths through the south and east boundaries .
an example of such boundary conditions is the domain wall ( dw ) boundary conditions . for the dw boundary conditions the arrows on the boundary of the lattice are going into the lattice at the top and bottom of the lattice and are going out of the lattice at the right and left of it .
a configuration of paths on a @xmath44 lattice with dw boundary conditions is presented on fig .
[ dwbc_conf ] .
notice that the differences @xmath45 are the same for all configurations with given fixed boundary conditions .
here @xmath46 is the total number of vertices of type @xmath47 in the configuration .
in particular , the partition function for fixed boundary conditions trivially depends on magnetic fields : @xmath48 square grid for the dw boundary conditions ] in this paper we will focus on the @xmath0-vertex model with fixed boundary conditions .
[ [ section-1 ] ] another important type of boundary conditions are the _ periodic boundary conditions_. in this case the edges at opposite sides of @xmath2 are identified so that the configuration of arrows on the left and right boundary is the same as well as the configuration of arrows on the top and bottom boundary .
the @xmath0-vertex model with periodic boundary conditions is an example of an `` integrable '' ( solvable ) model in statistical mechanics and has been studied extensively , see @xcite,@xcite and references therein . in particular , it means that the row - to - row transfer - matrix of the model can be diagonalized by the bethe ansatz . by _ outer faces _ we mean unit squares centered at @xmath49 , @xmath50 , with @xmath51 and @xmath52 , @xmath53 , with @xmath54 .
each corner outer face has two edges in their boundary , other outer faces have three edges in their boundary .
a height function @xmath55 is an integer - valued function on the faces @xmath56 of the grid @xmath28 ( including the outer faces ) , which is * zero at the southwest corner of the grid , * non - decreasing when going up or right , * if @xmath57 and @xmath58 are neighboring faces , then @xmath59 .
the _ boundary value _ of the height function is its restriction to the `` outer faces '' .
denote the set of outer faces by @xmath60 . given
a function @xmath61 on @xmath60 denote @xmath62 the space of all height functions with the boundary value @xmath61 .
if we enumerate faces by the coordinates of their centers the height function can be regarded as @xmath63 matrix with non - negative entries .
it is clear that there is a bijection between states of the @xmath0-vertex model with fixed boundary conditions and height functions with fixed boundary values .
indeed , given a height function consider its `` level curves '' , i.e. paths on the grid @xmath28 , where the height function changes its value by @xmath4 , see fig .
[ hf_example ] .
clearly , this defines a state for the @xmath0-vertex model on @xmath28 with boundary conditions determined by the boundary values of the height function . on the other hand , given a state in the @xmath0-vertex model , consider the corresponding configuration of paths .
it is clear that there is a unique height function whose level curves are these paths and which satisfies the condition @xmath64 at the southwest corner .
it is clear that this correspondence is a bijection .
there is a natural partial order on the set of height functions with given boundary values .
one function is bigger then the other if it is entirely above the other .
there exist the minimum @xmath65 and the maximum @xmath66 height functions such that @xmath67 for all height functions @xmath55 .
thus , we can consider the @xmath0-vertex model as a theory of fluctuating discrete surfaces constrained between @xmath66 and @xmath65 .
each surface occurs with probability given by the boltzmann weights of the @xmath0-vertex model . in the inhomogeneous @xmath0-vertex model the boltzmann weights depend on the edge .
thus , we have @xmath68 parameters @xmath69 , @xmath70 , and @xmath71 .
let us assume that the inhomogeneity is only in magnetic fields , i.e. weights @xmath31,@xmath33 , and @xmath35 do not change from edge to edge , but the magnetic fields @xmath72 , @xmath73 do .
let @xmath74 be the collection of paths corresponding to a state in the @xmath0-vertex model and @xmath75 be the corresponding height function .
let us assign weights @xmath76 to the edges of the lattice and @xmath4 to the outer edges , then @xmath77 where @xmath78 for inner faces , @xmath79 for the outer face @xmath80 adjacent to the edge @xmath41 , @xmath81 for outer faces not adjacent to any edge of the lattice ( corner faces ) with @xmath82 for edges at the upper and right sides of the boundary , @xmath83 for edges at the lower and left sides of the boundary .
the proof is an elementary exercise .
let @xmath84 for horizontal edges and @xmath85 for vertical edges .
then the probability of the state with the height function @xmath55 in such a model is @xmath86 if @xmath87 and @xmath88 , the weights @xmath89 are the same for all faces @xmath90 inside the lattice and the probability is given by @xmath91 where @xmath92 are the boltzmann weights with constant magnetic fields and @xmath93 is the volume `` under '' the height function @xmath55 .
[ [ section-2 ] ] let @xmath94 .
we place the grid @xmath2 inside of the rectangle @xmath95 so that the vertices of the grid are the points with coordinates @xmath96 where @xmath97 .
we recall that a height function is a monotonic integer - valued function on the faces of the grid , which satisfies lipshitz condition ( it changes at most by @xmath4 on any two adjacent faces ) .
the height function can be regarded as a function on the centers of the faces of the grid , i.e on @xmath98 , where @xmath99 .
the points @xmath100 , @xmath101 , @xmath102 , and @xmath103 , where @xmath104 correspond to the `` outer '' faces of @xmath2 .
we introduce the normalized height function as a piecewise linear function on the unit square with the value @xmath105 for @xmath106 and @xmath107 . here
@xmath108 is a height function on @xmath2 .
normalized height functions are nondecreasing in @xmath109 and @xmath110 directions and they satisfy : @xmath111
if @xmath112 and @xmath113 . the boundary value of the normalized height function defines a piece - wise constant monotonic function on each side of the region which changes by @xmath114 or do not change between two neighboring boundary sites .
denote the space of such normalized height functions with the boundary value @xmath115 by @xmath116 .
there is a natural partial ordering on the set of all normalized height functions with given boundary values descending from the partial order on height functions : @xmath117 if @xmath118 for all @xmath119 .
we define the operations @xmath120 it is clear that @xmath121 it is also clear that in this partial order there is unique minimal and unique maximal height functions , which we denote by @xmath122 and @xmath123 , respectively . a _ sequence of stabilizing fixed boundary conditions _ is a sequence of functions @xmath124 which are boundary values of a normalized height function , and which converges as @xmath125 to a function @xmath126 on the boundary of @xmath127 which is non - decreasing along @xmath109 and @xmath110 direction , @xmath128 , and satisfies the condition : @xmath129 we will call the function @xmath126 the boundary condition for the domain @xmath127 ( or simply the boundary condition ) .
any function on @xmath127 which satisfies ( [ lip - cont ] ) , which is non - decreasing in @xmath109 and @xmath110 directions , and coinside with @xmath126 at the boundary is called a
_ _ h__eight function on @xmath127 with boundary condition @xmath126 . among all possible boundary conditions @xmath126
we will distinguish piecewise linear boundary conditions with the slope @xmath130 or @xmath4 along the coordinate axes .
we will call these boundary values * critical*. it is clear that any boundary conditions can be approximated critical boundary conditions .
if @xmath131 is fixed and is not equal to @xmath4 in the large volume limit , the system will be in a neighborhood of the minimal height function for @xmath132 and in the neighborhood of the maximal height function for @xmath133 .
one should expect that the partition function and local correlation functions will have finite limit .
when @xmath134 for some @xmath135 , one should expect the existence of the limit shape .
we will study this limit in section [ section_var ] .
the gibbs measure of the @xmath0-vertex model on an infinite lattice has the _ slope _
@xmath136 if @xmath137 and @xmath138 it is clear from the definition of the height function that the slope should satisfy conditions @xmath139 .
the slope is simply the average number of horizontal and vertical edges occupied by paths per length .
the important corollary of the result of @xcite is that , when a gradient gibbs measure satisfies certain convexity conditions , there exists a unique translationally invariant measure .
this implies that for the @xmath0-vertex model one should expect the uniqueness of such a measure for the generic slope .
translationally invariant measures can be obtained by taking the thermodynamic limit of the 6-vertex model with magnetic fields on a torus .
then the slope is the legendre conjugate to magnetic fields .
@xmath140 for a horizontal @xmath41 and @xmath141 for a vertical @xmath41 .
the free energy per site in the thermodynamic limit is @xmath142 where @xmath143 is the partition function with the periodic boundary conditions on the rectangular grid @xmath2 .
it is a function of the boltzmann weights and magnetic fields .
normally , it is expected that the free energy is not identically zero .
physically , this means that the `` excitations '' have the characteristic length which is much smaller than the characteristic length of the system . in some cases the free energy is identically zero
, then one expects that a typical excitation will be comparable with the size of the system . for generic @xmath23 and @xmath24 the @xmath0-vertex model in the thermodynamic limit
has the translationally invariant gibbs measure with the slope @xmath136 : @xmath144 the parameter @xmath145 defines many characteristics of the @xmath0-vertex model in the thermodynamic limit .
the weights @xmath31 , @xmath33 , and @xmath35 in this region satisfy one of the two inequalities , either @xmath147 or @xmath148 . if @xmath147 , the boltzmann weights @xmath31 , @xmath33 , and @xmath35 can be parameterized as @xmath149 with @xmath150 .
if @xmath151 , the boltzmann weights can be parameterized as @xmath152 with @xmath153 . for both of these parametrization of weights @xmath154 .
the phase diagram of the model for @xmath147 ( and , therefore , @xmath155 ) is shown on fig .
[ ferro_diagram_1 ] and for @xmath148 ( and , therefore , @xmath156 ) on fig .
[ ferro_diagram_2 ] .
when magnetic fields @xmath157 are in one of the regions @xmath158 of the phase diagram , the system in the thermodynamic limit has the translationally invariant gibbs measure supported on the corresponding frozen configuration .
there are four frozen configurations @xmath159 , @xmath160 , @xmath161 , and @xmath162 , shown on fig .
[ ferroelectric_phase ] . for a finite but large grid
the probability of any other state is at most of order @xmath163 for some positive @xmath164 .
local correlation functions are given by the value of the corresponding observable on the frozen state @xmath165 where @xmath166 is the one of the ferromagnetic states @xmath167 .
the frozen regions in the @xmath157-plane are described by the set of inequalities .
the boundaries of these regions can be derived by analyzing the next to the largest eigenvalue of the row - to - row transfer matrix .
the description is separated into two cases : @xmath147 and @xmath148 .
notice that @xmath168 since @xmath146 .
@xmath169 @xmath147 , see fig .
[ ferro_diagram_1 ] , @xmath170 -plane for @xmath171 , @xmath172 , and @xmath173,width=302,height=302 ] @xmath169 @xmath148 , see fig .
[ ferro_diagram_2 ] , @xmath174 -plane for @xmath175 , @xmath176 , and @xmath173,width=302,height=302 ] the free energy is a linear function in @xmath23 and @xmath24 in the four frozen regions : @xmath177 if @xmath157 is a region @xmath178 or @xmath179 the corresponding translationally invariant gibbs measure has the slope @xmath136 given by ( [ slope ] ) . in this phase
the system is disordered , which means that local correlation functions decay as a power of the distance @xmath180 between @xmath181 and @xmath182 when @xmath183 . in the regions @xmath178 and @xmath179 the free energy is given by @xcite : @xmath184 where @xmath185 can be found from the integral equation @xmath186 in which @xmath187 @xmath185 satisfies the following normalization condition : @xmath188 the contour of integration @xmath189 ( in the complex @xmath190-plane ) is symmetric with respect to the conjugation @xmath191 , is dependent on @xmath23 ( see appendix [ a_free_energy_hom ] ) and is defined by the condition that the form @xmath192 has purely imaginary values on the vectors tangent to @xmath189 : @xmath193 the formula ( [ free_energy ] ) for the free energy follows from the bethe ansatz diagonalization of the row - to - row transfer - matrix .
its derivation is outlined in appendix [ a_free_energy_hom ] .
it relies on a number of conjectures that are supported by numerical and analytical evidence and in physics are taken for granted . however , there is no rigorous proof .
there are two points where three phases coexist ( two frozen and one disordered phase ) .
these points are called _
tricritical_. the angle @xmath194 between the boundaries of @xmath178 ( or @xmath179 ) at a tricritical point is given by @xmath195 the existence of such points makes the @xmath0-vertex model ( and its degeneration known as the @xmath196-vertex model @xcite ) remarkably different from dimer models @xcite where generic singularities in the phase diagram are cusps .
physically , the existence of singular points where two curves meet at the finite angle manifests the presence of interaction in the @xmath0-vertex model .
notice that when @xmath197 the phase diagram of the model has a cusp at the point @xmath198 .
this is the transitional point between the region @xmath146 and the region @xmath199 which is described below . in this case , the boltzmann weights have a convenient parameterization by trigonometric functions . when @xmath200 @xmath201 where @xmath202 , @xmath203 , and @xmath204 . when @xmath205 @xmath206 where @xmath202 , @xmath207 , and @xmath208 . the phase diagram of the @xmath0-vertex model with @xmath199 is shown on fig .
[ disordered_diagram ] .
the phases @xmath158 are frozen and identical to the frozen phases for @xmath146 .
the phase @xmath127 is disordered . for magnetic fields @xmath157
the gibbs measure is translationally invariant with the slope @xmath209 .
-plane for @xmath175 , @xmath176 , and @xmath210,width=377,height=377 ] the frozen phases can be described by the following inequalities : @xmath211 the free energy function in the frozen regions is still given by the formulae ( [ linear_energy ] ) .
the first derivatives of the free energy are continuous at the boundary of frozen phases , the second derivative is continuous in the tangent direction at the boundary of frozen phases and is singular in the normal direction .
it is smooth in the disordered region where it is given by ( [ free_energy ] ) which , as in case @xmath146 involves a solution to the integral equation ( [ int - eqn ] ) .
the contour of integration in ( [ int - eqn ] ) is closed for zero magnetic fields and , therefore , the equation ( [ int - eqn ] ) can be solved explicitly by the fourier transformation @xcite .
the @xmath0-vertex gibbs measure with zero magnetic fields converges in the thermodynamic limit to the superposition of translationally invariant gibbs measures with the slope @xmath212 .
there are two such measures .
they correspond to the double degeneracy of the largest eigenvalue of the row - to - row transfer - matrix @xcite .
there is a very interesting relationship between the @xmath0-vertex model in zero magnetic fields and the highest weight representation theory of the corresponding quantum affine algebra .
the double degeneracy of the gibbs measure with the slope @xmath212 corresponds to the fact that there are two integrable irreducible representations of @xmath213 at level one .
correlation functions in this case can be computed using @xmath131-vertex operators @xcite .
for latest developments see @xcite .
the boltzmann weights for these values of @xmath215 can be conveniently parameterized as @xmath216 where @xmath217 and @xmath218 .
the gibbs measure in thermodynamic limit depends on the value of magnetic fields .
the phase diagram in this case is shown on fig .
[ antiferroelectric_diagram ] for @xmath219 . in the parameterization ( [ hyp - param ] )
this correspond to @xmath220 .
when @xmath221 the @xmath3-tentacled `` amoeba '' is tilted in the opposite direction as on fig .
[ ferro_diagram_1 ] .
when @xmath157 is in one of the @xmath167 regions in the phase diagram the gibbs measure is supported on the corresponding frozen configuration , see fig .
[ ferroelectric_phase ] .
-plane for @xmath175 , @xmath176 , and @xmath222,width=377,height=377 ] the @xmath167 regions on the phase diagram are defined by inequalities ( [ boundaries_phases ] ) .
the free energy in these regions is linear and is given by ( [ linear_energy ] ) .
if @xmath157 is in the region @xmath127 , the gibbs measure is the translationally invariant measure with the slope @xmath136 determined by ( [ slope ] ) .
the free energy in this case is determined by solutions to the linear integral equation ( [ int - eqn ] ) and is given by the formula ( [ free_energy ] ) .
if @xmath157 is in the region @xmath223 , the gibbs measure is the superposition of two gibbs measures with the slope @xmath212 . in the limit
@xmath224 these two measures degenerate to two measures supported on configurations @xmath225 , respectively , shown on fig .
[ antiferr ] . for a finite @xmath215
the support of these measures consists of configurations which differ from @xmath226 and @xmath227 in finitely many places on the lattice .
we notice that any two configurations lying in the support of each of these gibbs measures can be obtained from @xmath226 or @xmath227 via flipping the path at a vertex `` up '' or `` down '' as it is shown on fig .
[ fluct ] finitely many times .
it is also clear that it takes infinitely many flips to go from @xmath226 to @xmath227 .
the @xmath0-vertex model in the phase @xmath223 is disordered and is also noncritical . here
non - criticality means that the local correlation function @xmath228 decays as @xmath229 with some positive @xmath164 as the distance @xmath180 between @xmath181 and @xmath182 increases to infinity . the free energy in the @xmath223-region can be explicitly computed by solving the equation ( [ int - eqn ] ) via the fourier transform @xcite .
this boundary of the antiferromagnetic region @xmath223 can be derived similarly to the boundaries of the ferromagnetic regions @xmath230 and @xmath231 by analyzing next to the largest eigenvalue of the row - to - row transfer - matrix .
the difference is that for the region @xmath223 the largest eigenvalue will correspond to @xmath232 and to compute it we should use the solution to the integral equation ( [ int - eqn ] ) in the case when the contour of integration is closed .
this computation was done in @xcite , @xcite .
the result is a simple closed curve , which can be described parameterically as @xmath233 where @xmath234 @xmath235 and @xmath236 the parameter @xmath237 is defined by the equation @xmath238 , where @xmath239 the curve is invariant with respect to the reflections @xmath240 and @xmath241 since the function @xmath242 satisfies the identities @xmath243 this function is also @xmath244-periodic : @xmath245 .
the boundary of the antiferromagnetic region is a real algebraic curve in @xmath246 and @xmath247 given by @xmath248 where @xmath249 is the positive value of @xmath24 on the curve when @xmath250 .
notice that @xmath237 depends on the boltzmann weights @xmath251 only through @xmath252 .
the parametric description of the boundary curve implies that @xmath253 the addition formula for the jacobi elliptic function @xmath254 @xcite @xmath255 can be used to express @xmath256 and @xmath257 for @xmath258 and @xmath259 in terms of @xmath260 and @xmath24 . using elementary identities for the elliptic functions , we obtain @xmath261 for @xmath258 and @xmath259 this identity turns into @xmath262
denote @xmath263 , then this identity becomes ( [ antiferroelectic_boundary ] ) .
here we study the asymptotics of the free energy of the @xmath0-vertex model as the point @xmath157 inside a disordered region approaches its boundary on the phase diagram of the model .
let us consider the interface between the disordered region @xmath127 ( @xmath179 for @xmath146 ) and the @xmath159-region , see fig .
[ disordered_diagram ] .
it is given by @xmath264 , where @xmath265 let @xmath266 be a regular point on the interface , i.e. the interface can be parameterized by real analytic functions in its neighborhood , then @xmath267 .
we denote the normal vector to the interface at @xmath268 by @xmath269 and the tangent vector by @xmath270 .
a point @xmath271 in the vicinity of @xmath272 can be represented as @xmath273 , where @xmath274 and @xmath275 are local coordinates in the normal and tangent directions , respectively , and @xmath276 is a scaling factor such that @xmath277 .
let us choose the normal vector @xmath269 so that it points in the direction of the disordered region @xmath127 , then the point @xmath278 belongs to @xmath127 if @xmath279 .
let @xmath278 be defined as above .
the asymptotics of the free energy of the @xmath0-vertex model in the limit @xmath277 is given by @xmath280 where @xmath281 and @xmath282 here the constants @xmath283 and @xmath194 depend on the boltzmann weights of the model and on @xmath284 and are given by @xmath285 and @xmath286 where @xmath287 is defined in ( [ inter_d_f ] ) .
moreover , @xmath288 and , therefore , @xmath289 .
we refer the reader to @xcite for the details .
assume that @xmath155 .
the theorem below describes the asymptotic of the free energy function when @xmath290 and @xmath291 these values of @xmath157 describe points inside the right `` tentacle '' on the fig .
[ ferro_diagram_1 ] .
let us parameterize these values of @xmath24 as @xmath292 where @xmath293 $ ] .
when @xmath294 and @xmath295 $ ] the asymptotic of the free energy is given by the following formula : @xmath296 from the integral equation for @xmath185 we can derive the large @xmath23 asymptotics of the density function : @xmath297 .
the contour is a small deformation of the segment of the circle of radius @xmath298 centered at the origin with endpoints having arguments @xmath299 . for large @xmath23
the free energy function is given by @xmath300 as @xmath294 the density is given by ( [ tent - dens ] ) and , taking into account the asymptotical description of the contour of integration we obtain @xmath301 where @xmath302
this integral is easy to compute : @xmath303 the minimum occur at @xmath304 or , at @xmath305 the formula ( [ tent - ass ] ) follows after the substitution of this into the expression for @xmath306 .
the @xmath196-vertex model can be obtained as the limit of the @xmath0-vertex model when @xmath307 .
magnetic fields in this limit behave as follows : * @xmath147 . in the parameterization ( [ dg1par-1 ] ) after changing variables @xmath308 , and @xmath309 take the limit @xmath310 keeping @xmath135 fixed .
the weights will converge ( up to a common factor ) to : @xmath311 * @xmath151 . in the parameterization ( [ dg1par-2 ] ) after changing variables @xmath308 , and @xmath312 take the limit @xmath310 keeping @xmath313 fixed .
the weights will converge ( up to a common factor ) to : @xmath314 the two limits are related by inverting horizontal arrows . from now on we will focus on the 5-vertex model obtained by the limit from the 6-vertex one when @xmath147 .
the phase diagram of the 5-vertex model is easier then the one for the 6-vertex model but still sufficiently interesting .
perhaps the most interesting feature is that the existence of the tricritical point in the phase diagram .
we will use the parameter @xmath315 notice that @xmath316 .
the frozen regions on the phase diagram of the @xmath196-vertex model , denoted on fig .
[ amoeba_5v ] as @xmath159 , @xmath160 , and @xmath161 , can be described by the following inequalities : @xmath317 -vertex model with @xmath318 ( @xmath319).,width=302,height=302 ] the disordered region @xmath127 near the tricritical point forms a corner @xmath320 the angle @xmath194 between the boundaries of the disordered region at this point is given by @xmath321 one can argue that the finiteness of the angle @xmath194 manifests the presence of interaction in the model . in comparison
, translationary invariant dimer models can only have cusps as such singularities .
as it follows from results @xcite the limit from the 6-vertex model to the 5-vertex model commutes with the thermodynamical limit and for the free energy of the @xmath196-vertex model we can use the formula @xmath322 where @xmath323 is the free energy of the @xmath0-vertex model .
let @xmath324 and @xmath325 the asymptotics of the free energy along this ray inside the corner near the tricritical point is given by @xmath326 where @xmath327 and @xmath328 the proof is computational .
the details can be found in @xcite .
the scaling along any ray inside the corner near the tricritical point in the 6-vertex model differ from this only by in coefficients .
the exponent @xmath329 is the same .
the details will be given in a separate publication .
if @xmath331 , the region @xmath223 consists of one point located at the origin .
when @xmath332 we have @xmath333 and @xmath334 .
moreover , @xmath335 , @xmath336 , and @xmath337 .
since @xmath338 , when @xmath339 , we have @xmath340 since @xmath341 is an odd function we obtain the following asymptotic of @xmath242 @xcite : @xmath342 in this limit the antiferromagnetic region degenerates into the origin @xmath250 and @xmath343 exponentially fast .
we note that the point @xmath198 is special for @xmath199 .
the following identity holds in the region
@xmath127 @xcite : @xmath344 here @xmath345 .
the constant @xmath346 does not vanish in the @xmath127-region including its boundary .
it is determined by the solution to the integral equation for the density @xmath185(see appendix c ) . directly from the definition of the free energy we have @xmath347 where @xmath348 and @xmath349 are the number of arrows pointing to the left and the number of arrows pointing to the right , respectively .
therefore , the matrix @xmath350 of second derivatives with respect to @xmath23 and @xmath24 is positive definite . as it follows from the asymptotical behavior of the free energy near the boundary of the @xmath127-phase , despite the fact that the hessian is nonzero and finite at the boundary of the interface , the second derivative of the free energy in the transversal direction at a generic point of the interface develops a singularity .
the legendre transform of the free energy @xmath351 as a function of @xmath352 is defined for @xmath353 . the variables @xmath109 and @xmath110 are known as polarizations and are related to the slope of the gibbs measure as @xmath354 and @xmath355 .
we will write the legendre transform of the free energy as a function of @xmath136 @xmath356 @xmath357 is defined on @xmath358 . for the periodic boundary conditions the surface tension function has the following symmetries : @xmath359 the last two equalities follow from the fact that if all arrows are reversed , @xmath360 is the same , but the signs of @xmath109 and @xmath110 are changed .
it follows that @xmath361 and @xmath362 .
the function @xmath306 is linear in the domains that correspond to conic and corner singularities of @xmath360 .
outside of these domains ( in the disordered domain @xmath127 ) we have @xmath363 here we gradient of a function as a mapping @xmath364 . when the 6-vertex model is formulated in terms of the height function , the legendre transform of the free energy
can be regarded as a surface tension .
the surface in this terminology is the graph of of the height function .
[ [ section-3 ] ] now let us describe some analytical properties of the function @xmath357 is obtained as the legendre transform of the free energy .
the legendre transform maps the regions where the free energy is linear with the slope @xmath365 to the corners of the unit square @xmath366 .
for example , the region @xmath159 is mapped to the corner @xmath367 and @xmath368 and the region @xmath161 is mapped to the corner @xmath367 and @xmath369 .
the legendre transform maps the tentacles of the disordered region to the regions adjacent to the boundary of the unit square .
for example , the tentacle between @xmath159 and @xmath161 frozen regions is mapped into a neighborhood of @xmath367 boundary of @xmath370 , i.e. @xmath371 and @xmath372 . applying the legendre transform to asymptotics of the free energy in the tentacle between @xmath159 and @xmath161 frozen regions we get @xmath373 and @xmath374 here @xmath375 and @xmath372 . from ( [ sur_ten_boundary ] )
we see that @xmath376 , i.e. @xmath360 is linear on the boundary @xmath367 of @xmath370 .
therefore , its asymptotics near the boundary @xmath367 is given by @xmath377 as @xmath375 and @xmath372 .
we note that this expansion is valid when @xmath378 .
similarly , considering other tentacles of the region @xmath127 , we conclude that the surface tension function is linear on the boundary of @xmath370 .
[ [ section-4 ] ] next let us find the asymptotics of @xmath360 at the corners of @xmath370 in the case when all points of the interfaces between frozen and disordered regions are regular , i.e. when @xmath379 .
we use the asymptotics of the free energy near the interface between @xmath159 and @xmath127 regions ( [ asympt_free_energy_interface ] ) .
first let us fix the point @xmath284 on the interface and the scaling factor @xmath276 in ( [ asympt_free_energy_interface ] ) .
then from the legendre transform we get @xmath380 and @xmath381 it follows that @xmath382 in the vicinity of the boundary @xmath383 and , hence , @xmath384 as @xmath385 .
thus , under the legendre transform , the slope of the line which approaches the corner @xmath386 depends on the boundary point on the interface between the frozen and disordered regions .
it follows that the first terms of the asymptotics of @xmath360 at the corner @xmath386 are given by @xmath387 where @xmath388 and @xmath389 can be found from ( [ slope_corner ] ) and @xmath267 . when @xmath390 the function @xmath360 is strictly convex and smooth for all @xmath391 .
it develops conical singularities near the boundary . when @xmath392 , in addition to the singularities on the boundary , @xmath360 has a conical singularity at the point @xmath212 .
it corresponds to the `` central flat part '' of the free energy @xmath80 , see fig .
[ antiferroelectric_diagram ] . when @xmath393 the function @xmath360 has corner singularities along the boundary as in the other cases .
in addition to this , it has a corner singularity along the diagonal @xmath394 if @xmath155 and @xmath395 if @xmath156 .
we refer the reader to @xcite for further details on singularities of @xmath360 in the case when @xmath146 .
[ [ section-5 ] ] let @xmath360 be the surface tension function of the @xmath0-vertex model with the periodic boundary conditions defined in ( [ surface_tension ] ) .
we consider the functional @xmath396=\int_d\sigma(\nabla \varphi)d^2x+\lambda\int_d \varphi d^2x,\ ] ] let @xmath397 be a minimizer of this functional on the space @xmath398 of functions nondecreasing in @xmath109 and @xmath110 directions and satisfying the condition @xmath399 and the boundary condition @xmath400 notice that height functions are lipschitz with @xmath401 .
the functional @xmath402 $ ] has unique minimizer .
indeed , since @xmath360 is convex , the minimizer is unique when it exists .
the existence of the minimizer follows from compactness of the space @xmath398 in the sup norm .
the arguments are completely parallel to those in @xcite .
[ [ section-6 ] ] if the vector @xmath403 is not a singular point of @xmath360 , the minimizer @xmath55 satisfies the euler - lagrange equation in a neighborhood of @xmath352 @xmath404 we can also rewrite this equation in the form @xmath405 where @xmath406 is an unknown function such that @xmath407 .
it is determined by the boundary conditions for @xmath55 . applying ( [ f - sig - comp ] )
, we see that @xmath408 from the definition of the slope , see ( [ slope ] ) we have @xmath409 and @xmath410 .
thus , if the minimizer @xmath55 is differentiable at @xmath352 , it satisfies the constrains @xmath411 and @xmath412 . the following statement is a minor variation of the theorem 4.3 from @xcite .
let @xmath125 , @xmath135 be finite and @xmath134 then the sequence of random normalized height functions @xmath413 converges in probability to the minimizer of ( [ var - pr ] ) .
the rate of convergence is exponential of @xmath414 .
the minimizer of the variational problem ( [ var - pr ] ) is called the _
limit shape _ of the height function .
this theorem is the manifestation of the general philosophy of the large deviations principle .
the probability of a having state with the height function @xmath55 has the has the following asymptotic @xmath125 : @xmath415 here @xmath360 is the surface tension function for the periodic boundary conditions .
clearly this probability has maximum at the limit shape .
states with the height function , which macroscopically differ from the limit shape , should be expected to be exponentially improbable .
the theorem states that this is exactly what is taking place .
the space of normalized height functions on @xmath2 has the partial order described in section 2 .
denote the minimal and maximal normalized height functiions with respect to this partial order @xmath417 and @xmath418 , respectively . for two functions @xmath419 , @xmath420 on @xmath127
define the distance @xmath421 similarly define the the distance between two functions on the boundary of @xmath127 .
let @xmath124 be a sequence of functions on the boundary of @xmath127 converging to @xmath422 , and such that @xmath124 is a boundary of a normalized height function of @xmath2 .
denote @xmath423 and @xmath424 the minimum and maximum height functions from @xmath425 .
the following is clear : let @xmath426 and @xmath427 be minimal and maximal height functions with the boundary condition @xmath422 .
then , @xmath428 and @xmath429 with respect to the distance ( [ distance ] ) . the functions @xmath426 and @xmath427 minimize the functionals @xmath430 in the sapec @xmath431 .
let us assume that the boundary conditions are critical , that is @xmath422 is piece - wise linear , non - decreasing in @xmath109 and @xmath110 direction with the slope @xmath130 or @xmath4 . in this case
the minimum and maximum height functions are piecewise linear functions , such that each linear part has the slope either @xmath130 or @xmath432 along coordinate axes .
we will say a point @xmath352 is _ regular _ if at this point the function @xmath426 is differentiable .
for critical boundary conditions regular points form regions with piece - wise linear boundary where the function @xmath426 has a constant slope . we will call them _ linear domains_. points where the gradient of @xmath426 is discontinuous will be called _ singular _ points . for critical boundary conditions
singular points are the points where several linear domains meet .
the _ _ v__alency of a singular point is the number of linear domains which meet at this point .
for generic critical boundary conditions the valency of each critical point is at most three .
the list of all possible phases at a tricritical point of the @xmath0-vertex model with generic critical boundary conditions is given on fig .
[ triple_pts ] . -vertex model.,width=302,height=377 ] here we study the asymptotic of minimizer of @xmath433 $ ] as @xmath434 .
it is more convenient to divide @xmath435 by @xmath436 , so we are looking for the asymptotics of the minimizer @xmath437 of @xmath438=\frac{1}{|\lambda|}\int_d \sigma(\nabla\varphi)d^2x \pm \int_d \varphi d^2x.\ ] ] let us focus on the limit @xmath439 .
the limit @xmath440 can be treated similarly .
when @xmath441 , the minimizer @xmath437 approaches the minimal height function @xmath426 described above .
let us look for the asymptotical formula for the minimizer in a small neighborhood of a point @xmath442 of the form @xmath443 where @xmath444 as @xmath441 .
let @xmath445 and @xmath446 . because @xmath426 is linear , its second derivatives vanish and we can rewrite the euler - lagrange equation ( [ e - l_eq ] ) as @xmath447 notice that for critical boundary conditions the function @xmath448 is piece - wise constant
. now assume that @xmath442 is a singular point , i.e. a point where two or more linear domains of @xmath426 meet .
recall that for generic critical boundary conditions only two or three linear domains can meet at a point . taking the limit @xmath449 in ( [ el - h ] ) we obtain .
let @xmath437 be the minimizer of @xmath450 and @xmath442 be a singular point , then for each @xmath451 there exits @xmath452 this function is the solution to @xmath453 with the boundary conditions @xmath454 as @xmath455 along a ray in the @xmath456-th linear domain adjacent to @xmath442 . here
@xmath457 in the @xmath456-th linear domain and since the boundary conditions are critical , @xmath458 .
the function @xmath23 is once differentiable .
there is a smooth curve separating @xmath5 into regions where @xmath23 is linear and regions where @xmath23 is smooth with positive definite matrix of second derivatives . for non - generic boundary conditions in the thermodynamical limit
more then three linear domains can met at one point . in this case
one should expect that the conjecture still holds with more then three linear domains meeting at a singular point .
[ [ section-7 ] ] a height function defines the surface @xmath459 in @xmath460 .
regions where @xmath55 is linear are planes . here
we will describe the solution to the equation ( [ h - eqn ] ) with the boundary conditions ( [ as - cond ] ) in the case when there are only two asymptotic planes meeting at@xmath442 .
let @xmath461 and @xmath462 be the directions of the steepest assent of these planes .
the numbers @xmath463 and @xmath464 are either @xmath130 or @xmath465 since we assume critical boundary conditions and therefore @xmath426 is piece - wise linear with slope @xmath466 or @xmath467 .
the function @xmath468 , defined in ( [ asympt_fun ] ) , has the asymptotic conditions ( [ as - cond])determined by these planes .
the function @xmath468 is also invariant with respect to translations in @xmath469-direction .
thus , we are looking for a function @xmath283 such that @xmath470 which satisfies the differential equation ( [ h - eqn ] ) with the asymptotic conditions ( [ as - cond ] ) .
let us introduce @xmath471 then the differential equation for @xmath23 becomes the first order ode for the function @xmath472 @xmath473 integrating it , we obtain the equation defining the function @xmath474 implicitly @xmath475 with some constant @xmath189 .
[ [ section-8 ] ] for the domain wall boundary conditions minimal and maximal height functions are shown on fig .
[ dw_min_max ] . in this case
the boundary between two linear domains is a line in the direction @xmath476 for the minimal height function and is the line in the direction @xmath467 for the maximal height function .
thus , for these boundary conditions the function @xmath23 describing the asymptotic of the minimizer is constant in the direction @xmath476 when @xmath441 and it is constant in the @xmath467-direction if @xmath477 .
let @xmath478 be the solution to ( [ eqn - for - kappa ] ) which is invariant with respect to translations in the @xmath476 direction .
the symmetries of the legendre transform @xmath360 of the free energy imply that @xmath361 . using the equation ( [ implic_eq_g ] ) and this symmetry of @xmath360 , we obtain @xmath479 taking into account ( [ f - sig - comp ] ) we obtain @xmath480 and , hence , @xmath481 in the case of two asymptotic planes we have an additional symmetry of @xmath468 with respect to the intersection line of these planes , i.e. @xmath482 . the free energy @xmath80 also has the symmetry @xmath483 .
therefore @xmath484 and we proved the following statement . for dw boundary conditions the minimizer @xmath437 when @xmath485 has the following asymptotic when @xmath486 , @xmath487 : @xmath488 if @xmath486 and @xmath489 and @xmath490 the asymptotic of the minimizer @xmath437 is given by @xmath491 [ [ section-9 ] ] when @xmath492 ( which is equivalent to @xmath493 ) the height function @xmath437 develops extra linear domains known as facets .
these are the regions of the antiferroelectric phase discussed in section [ af - region ] . in dimer models
this phenomenon is studied in @xcite .
the facets also appear in the function @xmath23 describing the asymptotic of the minimizer as @xmath494 .
for the function @xmath23 describing the asymptotic near a double singular point @xmath442 where two linear domains meet along the diagonal @xmath495 the facet is a strip @xmath496 in coordinates @xmath497 .
its width @xmath498 is given by the formula @xmath499 in the limit @xmath333 or @xmath332 the asymptotic of @xmath242 gives the following asymptotical value of @xmath498 : @xmath500 the free energy @xmath80 is linear in the antiferromagnetic region .
it is growing with exponent @xmath501 in the normal direction to the boundary outside of the antiferromagnetic region .
this agrees with the pokrovsky - talapov law which states that @xmath80 should be growing with exponent @xmath501 in the normal direction to the boundary of the facet @xcite .
in particular , @xmath55 grows with the exponent @xmath501 in the @xmath467-direction near the boundary of the @xmath35-droplet far enough from the boundary of the square .
for @xmath502 we have @xmath503 where @xmath115 is the value of the height function at the boundary of the facet and @xmath283 is a constant which can be computed explicitly .
as we have seen in the previous section at the macroscopical distances in the thermodynamical limit the height function is deterministic and is the minimizer for the variational problem ( [ var - pr ] ) .
but the height function at smaller distances remain random .
their fluctuations are described by the asymptotical behavior of correlation functions in the thermodynamical limit .
these asymtotics have been studied extensively in dimer models which describe the @xmath504 case of the 6-vertex model .
the thermodynamical limit of correlations functions in the bulk describes translationary invariant gibbs measure with given polarization .
these asymptotics of correlation functions have been studied a lot using various methods which are essentially based on representation theory of affine quantum groups .
the exact computation of the asymptotic of correlation functions for local observable separated by large distances on the lattice remain one of the main problems . on the other hand
this is one of the most important physically relevant information about correlation functions .
some information about this asymptotic of correlation functions can be obtained using the arguments of finite - size scaling and the assumption of conformal invariance of the leading terms of the asymptotic .
let us consider the 6-vertex configurations of paths which may end at some edges .
the weight of such configurations is given the same product of weights as before .
define local observables @xmath505 and @xmath506 as @xmath507 @xmath508 the value of a product of such observable when each of the factors correspond to a different edge is the product of values of observables .
the following formulae were obtained in @xcite for @xmath250 and when all edges are vertical on the same row using the finite - size scaling and the assumption of conformal invariance : @xmath509 @xmath510 here @xmath511 is the distance between @xmath512 and @xmath513 , @xmath514 .
the terms denoted by @xmath515 are with higher powers of @xmath516 and @xmath517 where @xmath518 is the distance between @xmath512 and @xmath513 .
these higher order terms may also be oscillating . because @xmath250 the integration contour @xmath189 in the integral equation fro @xmath185 is a segment of the real line in the additive parameterization .
the constant @xmath519 is the fermi - momentum and it is also can be expressed in term of @xmath185 .
it is also equal to the vertical electrical polarization .
the finite size computations were extended to the case @xmath520 in @xcite .
they argued that the complete spectrum of effective @xmath521 conformal field theory is given by @xmath522 where @xmath406 is defined by the hessian of the free energy and is related to @xmath164 as @xmath523 . combining these results we conjecture that for generic electric fields the asymptotical behavior of correlation functions is still given by formulae ( [ as-1 ] ) and ( [ as-2 ] ) with @xmath523
one can show easily that when @xmath504 we have @xmath524 and the asymptotics ( [ as-1 ] ) ( [ as-2 ] agrees with the results from @xcite on dimer models . here
we will list some open question about the 6-vertex model and other related models .
* find the classification of generic singularities of limit shapes . for @xmath146 one should expect the presence of corners as generic singularities of limit shapes . *
find the scale at which fluctuations near singularities of limits shapes are described by some random process and describe such processes .
for example in dimer models such fluctuations near the boundary of the limit shape are described by airy process , and near a generic cusp are described by the pearcey process .
+ one can argue that the same processes should describe correlation functions near similar singularities of the limit shape in the 6-vertex model but this is still a conjecture .
+ the scaling near the corner singularity seems particularly interesting problem since such singularities do not appear in dimer models .
* understand the role of integrability of the 6-vertex model in the formation of the limit shape . here by integrability we mean that the model can be solved by the bethe ansatz , and that the weights satisfy the yang - baxter equation ( and therefore transfer - matrices form a commuting family ) .
* the 6-vertex model is closely related to the representation theory of quantized universal enveloping algebra of @xmath525 .
it would be extremely interesting to see which aspect of the representation theory of this algebra naturally appear in the limit shape phenomenon and in the scaling of correlation functions near singularities . *
the 6-vertex model has natural generalizations related to other simple lie algebras .
configurations in these models can be described in terms of @xmath276 height functions where @xmath276 is the rank of the lie algebra .
the limit shape in this case is a surface in @xmath526 .
it would be extremely interesting to investigate such systems trying to answer all questions mentioned above .
99 d. allison and n. reshetikhin , numerical study of the @xmath0-vertex model with domain wall boundary conditions , ann .
fourier ( grenoble ) * 55 * , 18471869 , 2005 .
m. abramowitz and i.a .
stegun , _ handbook of mathematical functions with formulas , graphs , and mathematical tables _ , dover publications , 1972 .
baxter , _ exactly solved models in statistical mechanics _ , academic press , san diego , 1982 .
bogoliubov , a.g .
pronko , and m.b .
zvonarev , boundary correlation functions of the six - vertex model , j. phys .
a * 35 * , 55255541 , 2002 .
bogoliubov , a.g .
izergin , and n.y .
finite - size effects and infrared asymptotics of the correlation functions in two dimensions .
j. phys . a 20 ( 1987 ) , no . 15 , 53615369 .
bressoud , _ proofs and confirmations : the story of the alternating sign matrix conjecture _ , cambridge university press , 1999 .
bukman and j.d .
shore , the conical point in the ferroelectric six - vertex model , j. stat .
* 78 * , 12771309 , 1995 .
h. cohn , n. elkis , and j. propp , local statistics for random domino tilings of the aztec diamond , duke math . j.
* 85 * , 117166 , 1996 .
h. cohn , r. kenyon , and j. propp , a variational principle for domino tilings .
* 14 * , 297 - 346 , 2001 .
f. colomo , and a.g .
pronko , on two - point boundary correlations in the six - vertex model with domain wall boundary conditions . j. stat .
theory exp .
2005 , no . 5 , 05010 , 21 pp .
r. dobrushin , r. kotecky , and s. schlosman , _ wulff construction : a global shape from local interactions _ , ams translation series , providence ri , * 104 * , 1992 .
k. eloranta , diamond ice , j. statist . phys . * 96 * , 10911109 , 1999 . p.l .
ferrari , m. prhofer , and h. spohn , fluctuations of an atomic ledge bordering a crystalline facet , cond - mat/0303162 .
huang , f.y .
wu , h. kunz , d. kim , interacting dimers on the honeycomb lattice : an exact solution of the five - vertex model , physica a * 228 * , 132 , 1996 .
izergin , partition function of the @xmath0-vertex model in a finite volume , ( russian ) dokl .
nauk usrr , * 297 * , 331333 , 1987 .
m. jimbo and t. miwa , _ algebraic analysis of solvable lattice models _ , cbms regional conference series in math . *
85 * , 1993 .
h. boos , m. jimbo , t. miwa , f. smirnov , y. takeyama , hidden grassmann structure in the xxz model.hep-th/0606280 c. jayaprakash and w.f .
saam , phys .
b * 30 * , 3916 , 1984 .
p. kasteleyn , _ graph theory and crystal physics _ , graph theory and theoretical physics , 43 , 1967 .
m. kardar , g. parisi , and y.c .
zhang , phys .
lett . * 56 * , 889 , 1986 r. kenyon , height fluctuations in the honeycomb dimer model , math - ph/0405052 .
r. kenyon , and a. okounkov , planar dimers and harnack curves .
duke math .
j. 131 ( 2006 ) , no .
3 , 499524 .
r. kenyon , a. okounkov , and s. sheffield , dimers and amoebae .
ann . of math .
( 2 ) 163 ( 2006 ) , no .
3 , math - ph/0311005 .
r. kenyon , and d. wilson , critical resonance in the non - intersecting lattice path model .
probab . theory related fields 130 ( 2004 ) , no .
3 , 289318 . v.e .
korepin , n.m .
bogolyubov , and a.g .
izergin , _ quantum inverse scattering method and correlation functions _ , cambridge university press , 1993 .
korepin , calculation of norms of bethe wave functions , commun .
* 86 * , 391418 , 1982 .
v. korepin and p. zinn - justin , thermodynamic limit of the six - vertex model with domain wall boundary conditions , j. phys .
a , * 33 * , 70537066 , 2000 ; inhomogeneous six - vertex model with domain wall boundary conditions and bethe ansatz , j. math . phys . * 43 * , 32613267 , 2002 .
g. kuperberg , another proof of the alternating sign matrix conjecture , internat .
notices 3 , 139150 , 1996 .
lieb , phys . rev . * 162 * , 162 , 1967 ; e.h .
lieb , phys .
lett . * 18 * , 1046 , 1967 ; * 19 * , 108 , 1967 .
e.h . lieb and f.y .
wu , two dimensional ferroelectric models , in _ phase transitions and critical phenomena _ , vol .
1 , ed . by c.domb and m.s . green , 321 , academic press , london , 1972 .
noh and d. kim , finite - size scaling and the toroidal partition function of the critical asymmetric six - vertex model , cond - mat/9511001 .
i. m. nolden , the asymmetric six - vertex model , j. statist . phys . * 67 * , 155 , 1992 ; ph.d .
thesis , university of utrecht , 1990 .
k. palamarchuk , the @xmath0-vertex model in statistical mechanics , ph.d .
thesis , univesity of california at berkeley , 2006 .
pokrovsky and a.l .
talapov , phys .
lett . * 42 * , 65 , 1979 .
s. sheffield , phd thesis , stanford univ .
f. smirnov , form factors in completely integrable models of quantum field theory , world scientific , singapore 1992 .
b. sutherland , c.n .
yang , and c.p .
yang , exact solution of a model of two - dimensional ferroelectrics in an arbitrary external electric field , phys .
letters * 19 * , 588591 , 1967 .
b. sutherland , phys .
* 19 * , 103 , 1967 ; o.f . syljuasen and m.b .
zvonarev , directed - loop monte carlo simulations of vertex models , cond - mat/0401491 .
yang , phys .
* 19 * , 586 , 1967 c.n .
yang and c.p .
yang , phys . rev . * 150 * , 321 , 1966 ; * 150 * , 327 , 1966 p. zinn - justin , six - vertex model with domain wall boundary conditions and one - matrix model , phys . rev .
e * 62 * , 3411 - 3418 , 2000 .
p. zinn - justin , the influence of boundary conditions in the six - vertex model , cond - mat/0205192 .
zuber , on the counting of fully packed loop configurations .
some new conjectures , math - ph/0309057 .
here we compute the partition function of the inhomogeneous @xmath0-vertex model with the periodic boundary conditions applying the bethe ansatz method .
we briefly describe the results and refer the reader to @xcite for the details about the method .
bethe ansatz method works for the inhomogeneous 6-vertex model with the follwing inhomogeneities .
one should parameters @xmath527 , @xmath528 , .. , @xmath529 to each row of the grid @xmath2 and parameters @xmath530 , @xmath531 , .. , @xmath532 to each column of the grid as shown on fig .
[ spec_par ] .
thus , a pair of spectral parameters @xmath533 is associated to the vertex @xmath534 of the grid .
boltzmann weights assigned to this point are : @xmath535
@xmath536 @xmath537 where functions @xmath538 describe the parameterization of boltzmann weights of the homogeneous model .
-by-@xmath3 square grid with corresponding spectral parameters . ]
the partition function of the model with periodical boundary conditions can be written as @xmath539 where @xmath12 is the @xmath540 row - to - row transfer matrix of the 6-vertex model @xcite .
the raw - to - raw transfer - matrix is the trace of the `` quantum monodromy matrix '' : @xmath541 here @xmath542 is the product @xmath543 which acts in @xmath544 .
the matrix @xmath545 acts as the matrix @xmath546 in the basis @xmath547 of the product of the 0-th and j - th factors in the tensor product .
it acts trivially in other factors .
let us denote the number of arrows pointing down in a row of the grid by @xmath548 .
denote by @xmath549 the corresponding subspace in the space of all possible states on the raw of length @xmath260 of vertical edges .
the action of the transfer - matrix preserve these subspaces .
the bethe ansatz method gives the following result for the eigenvalues of @xmath12 : @xmath550 where @xmath551 and @xmath552 satisfy the bethe equations @xmath553 it follows that the partition function of the model is given by @xmath554 where @xmath555 are the eigenvalues of the row - to - row transfer matrix @xmath12 and the sum is taken over all eigenvalues with their multiplicities .
thus , in the homogeneous case when all @xmath556 are the same the asymptotic of the partition function as @xmath557 is @xmath558 where @xmath518 is the multiplicity of the largest eigenvalue @xmath559 of the row - to - row transfer - matrix on @xmath260 sites . therefore , if we can compute the asymptotic of the largest eigenvalue of the transfer - matrix as @xmath125 we can find the asymptotic of the partition function in the thermodynamical limit . strictly speaking , this logic gives the asymptotic of the partition function in the limit @xmath560 . under mild assumptions
one can argue that the leading term of the asymptotic of free energy is uniform .
however , in some cases a resonant phenomenon may occur which will make the leading term of the asymptotic of @xmath143 depend on the ratio @xmath561 @xcite .
in the homogeneous @xmath0-vertex model the boltzmann weights are the same for all vertices on the grid @xmath28 .
when @xmath562 the functions @xmath563 are : @xmath564 other values of @xmath215 can b obtained by analytical continuation .
the formula for eigenvalues of the row - to - row transfer matrix given in the previous section becomes @xmath565 where @xmath566 are solutions of the bethe ansatz equations @xmath567^n= e^{2nh}\mathop{\prod_{k=1}^n}_{k\neq j } \frac{\sin(u_k - u_j+\eta)}{\sin(u_k - u_j-\eta ) } , \quad j=1, .. ,n.\ ] ] introducing new variables @xmath568 we get @xmath569 where @xmath570 satisfy the bethe equations @xmath571 we want to find the asymptotic of the largest eigenvalue of the transfer - matrix when @xmath125 .
we shall do it in two steps .
first we will find the asymptotic of the largest eigenvalue among @xmath572 in the limit @xmath573 after this we will find for which @xmath574 this eigenvalue is the largest .
[ a_bethe_eqs ] let us denote @xmath575 then , taking the logarithm of both sides of each bethe equation ( [ bethe_eq_z ] ) and dividing by @xmath260 , we get @xmath576 where @xmath577 are integers .
all theoretical and numerical results on the @xmath0-vertex model suggest that the following is true .
[ main_conj ] in the limit @xmath578 with @xmath579 is fixed the roots of the bethe equations corresponding to the maximal eigenvalue of the transfer matrix in the subspace @xmath580 concentrate on a smooth curve @xmath189 in the complex @xmath190-plane .
the curve @xmath189 is symmetric with respect to the complex conjugation @xmath191 .
the maximal eigenvalue of the transfer - matrix in the subspace @xmath549 corresponds to @xmath581 .
these conjectures can be proven in the free - fermion case @xmath504 .
the conjecture ( [ main_conj ] ) implies that as @xmath125 one should expect finite limit density of @xmath582 on the contour @xmath189 . if @xmath583 the limit @xmath584 is the value of the limit density function of solutions to the bethe equations on @xmath189 .
let us denote the endpoints of the contour @xmath189 by @xmath585 and @xmath586 , @xmath587 and @xmath588 .
taking the limit in the equation ( [ b - eqn ] ) we obtain the integral equation for @xmath185 : @xmath589 here @xmath590 . differentiating this equation with respect to @xmath190
, we obtain another , more convenient form of this equation : @xmath591 where @xmath592 the kernel @xmath593 is singular on two curves @xmath594 and it can be written as @xmath595 notice that functions @xmath596 , i.e. @xmath597 and @xmath598 are inverse to each other . if @xmath504 , coutours @xmath226 and @xmath227 coincide and the kernel is zero .
if @xmath599 , there are two cases : * conours @xmath189,@xmath226 , @xmath227 do not intersect each other ; * contours @xmath189,@xmath226 , @xmath227 , _ all _ intersect at two points @xmath600 which are the solutions of @xmath601 and are conjugate to each other ( these two points may coincide in the degenerate case ) .
this is possible only if @xmath602 .
the integral equation on @xmath185 can be rewritten as @xmath603 notice that the contour @xmath189 can be deformed as long as it does not intersect the curves @xmath226 and @xmath227 . the contour @xmath189 is defined by the condition that the form @xmath192 has purely imaginary values on the vectors tangent to @xmath189 @xmath604 and by the normalization condition on @xmath185 @xmath605 the free energy per site is defined as @xmath606 the formula for eigenvaues @xmath572 has two terms .
each of them grow exponentially for large @xmath260 .
generically , one of the terms will dominate the other .
taking into account conjectures about roots of bethe equations in the thermodynamicallimit we obtain the follwoing integral representation for the free energy @xcite : @xmath607 or @xmath608 where @xmath185 is determined bythe integral equation ( [ eq - for - rhoz ] ) with conditions ( [ c - cond ] ) and ( [ rho - norm ] ) .
the free - fermion curve in the space of parameters of the 6-vertex model is @xmath504 . since in this case
@xmath609 the integral equation of @xmath610 immediately implies @xmath611 .
the contour @xmath189 in this case is an arc @xmath612 where @xmath613 $ ] . minimizing this expression in @xmath164 we get the equation @xmath615 this defines the critical value @xmath616 when the absolute value of the r.h.s is not greater then @xmath4 .
otherwise @xmath617 or @xmath602 .
taking into account the parameterization of boltzmann weights by electric fields @xmath23 and @xmath24 it is easy to see that ( [ s - int ] ) together with the equation for @xmath624 coincide with the formula for the free energy derived from the bethe ansatz . | we study the 6-vertex model with fixed boundary conditions . in the thermodynamical limit
there is a formation of the limit shape .
we collect most of the known results about the analytical properties of the free energy of the model as the function of electric fields and study the asymptotical behavior near singularities .
we also study the asymptotic of limit shapes and the structure of correlation functions in the bulk . |
measuring the distances to objects is one of the most fundamental tasks in astronomy , yet it is also one of the most difficult . among the many indirect methods astronomers have devised for dealing with this problem is the technique of `` standard candles : '' identifying a class of sources which have the same luminosity and are easy to recognize .
rr lyrae stars fit into this category : they vary in brightness by a considerable amount ( amplitudes of order half a magnitude ) in a short time ( periods of order half a day ) .
moreover , their light curves exhibit a characteristic shape : a rapid rise in brightness followed by a leisurely fall @xcite .
although rr lyr stars have become important tools for the investigation of galactic structure , they are not as well understood as one would wish for such fundamental calibrators . for example , we can not compare rigorously our models of stellar structure and their predictions for pulsation to observations , because we do not know precisely the mass of any rr lyr star .
the reason is simple : despite a few false alarms @xcite , and one case
tu uma of what may be a very wide binary containing an rr lyr @xcite , we have found no rr lyr stars in eclipsing binary systems which can be studied via photometric and spectroscopic methods .
the discovery of even a few rr lyr in eclipsing binary systems would provide a very valuable check to our understanding of these stars and improve our use of them as distance indicators . on the other hand ,
if comprehensive searches reveal that rr lyr stars occur in binary systems at rates far below that of other , similar , stars , such as the pulsating w vir variables which _ have _ been seen in eclipsing binary systems @xcite , we may deduce some features of the evolutionary sequence which leads to rr lyr stars .
we begin with the collection of data described in @xcite , which includes measurements of 3256 stars in the galactic bulge made during the course of the macho project @xcite .
the data are available freely from the macho collaboration , but we could not find the detailed description of their analysis promised by the reference `` cook et al .
( 2007 , in preparation ) . ''
we therefore do not know the particular procedures used to identify these stars as rr0 (= rrab ) lyr variables , nor to determine their fundamental periods .
all we have are the results of that analysis : photometry of several thousand rr0 lyr stars in the @xmath0 ( blue ) and @xmath1 ( red ) passbands of the macho project .
each star in this dataset is listed with its right ascension and declination and a macho identifier of the form @xmath2 , in which @xmath3 identifies the field , @xmath4 the tile , and @xmath5 the star within that tile .
we will use this macho identifier as a label for particular stars throughout this paper .
the data for each star consists of a fundamental period , a mean @xmath6band magnitude , and a series of measurements : the julian date , red magnitude and estimated uncertainty , blue magnitude and estimated uncertainty .
the mean @xmath6band magnitudes fall largely in the range @xmath7 , but the red and blue magnitudes listed for each measurement are on an instrumental system and lie between @xmath8 and @xmath9 .
figure [ fig : numepochs ] shows a histogram of the number of epochs of measurements of each star , which is typically several hundred .
for the benefit of readers who may decide to use this database for their own work , let us mention that some caution is required .
some measurements ought to be discarded : for example , those marked with magnitude values of -99 or magnitude uncertainties of 9.999 . we found that others are so noisy that they provide no significant information .
both the `` crowding '' and `` fwhm '' attributes associated with each measurement can be used to identify data of low significance . after examining the pattern of outliers in several test cases
, we decided to discard any measurements in which the `` fwhm '' values for both the red and blue images was larger than 6.5 pixels .
this catalog includes measurements made over a span of seven austral winters , starting in april , 1993 , and ending in october , 1999 , but most fields were not observed during all seven seasons .
let us choose a single star as an example , and follow it through our analysis .
star 101.21167.00060 is one of the brighter stars in the catalog , with a mean @xmath6band magnitude of 16.34 .
a graph of its photometry , figure [ fig : rawphotom ] , reveals that its field was not part of the regular observing sequence during the second season .
the goal of this work is to seek evidence of eclipses in the light curves of rr lyr stars
. it would be easier to find such evidence if the large variations in light due to the rr lyr pulsations are removed .
therefore , we created a model for the regular rr lyr light curve of each star and subtracted it from the measurements , leaving the residuals for further consideration .
our method to create the model for each star was simple : we phased the data with the period given in the macho catalog , divided the data into 20 bins of equal size in phase , and computed the median magnitude within each bin .
we assigned this median magnitude to the phase in the middle of its bin .
finally , we interpolated linearly between these median values to determine the magnitude at any phase .
figure [ fig : phasedrrlyr ] , shows the resulting models
one for the red magnitudes , one for the blue magnitudes for star 101.21167.00060 .
note that this simple approach has an obvious drawback : it does not match the actual light curve well in places where there is sharp change , such as phase 0.7 in figure [ fig : phasedrrlyr ] .
however , it does provide a reasonable model for stars with relatively few measurements , and it handles noisy data very well . after creating separate models for the red and blue measurements of each star , we subtracted the model from the data , leaving a set of residual magnitudes .
we show an example of these residuals , phased with the rr lyr period , in figure [ fig : phasedresiduals ] .
having subtracted model rr lyr light curves from the measurements of each star in the red and blue passbands , our next task was to search for patterns in the residuals which might indicate eclipses .
there are many approaches to this problem , in general , but given the nature of our data very inhomogeneous sampling with large gaps and often high noise levels and the uncertain nature of our expected signal which could range from sharp , narrow dips in light to smooth , continuous variation
we chose the `` string length '' method @xcite .
our implementation of this technique follows closely the description given in @xcite .
we generated string lengths for periods between 0.10 and 100 days , using steps equally spaced in frequency of size 0.0001 cycles per day .
for each possible period , we computed a string length separately for the red and blue measurements , then added the two lengths to form an overall figure of merit for that period .
we set thresholds for significance following the suggestions of @xcite and ignored periods which exceeded these thresholds .
we saved the periods which yielded the 10 shortest string lengths for further consideration .
in addition , we computed the string length for a period of 9999 days ; since this was much longer than the actual span of observations , it yielded a `` phased '' light curve which was simply in chronological order .
stars with very long periods of variation would show a short string length for this artificial period .
the next step was to examine the results for each star visually .
we created a graphical representation of the star s light phased with the best three periods , as shown in figure [ fig : phasedcandidates ] .
in addition to the measurements , the graph displayed the candidate periods , both in days and as a fraction of the star s rr lyr period .
a relatively quick view of this graph was sufficient to decide if any of the candidate periods yields any significant signal , and if the candidate periods are simply multiples or fractions of the rr lyr period .
the author examined graphs for all 3256 stars in the catalog and noted those which deserved further consideration .
stars which did show promising signs were subjected to additional tests .
first , light curves were generated for all 10 of the best candidate periods , in order to see which candidate period looked most significant to the eye .
second , we sometimes tried new periods , generating graphs and string lengths manually , in order to yield a phased light curve with 2 maxima .
for example , if the best period @xmath10 created a light curve with 5 maxima , we checked the period @xmath11 .
when the results looked good , we replaced the best automatically generated candidate period with the manually chosen period .
the final step was to improve the value of the best candidate period , and to derive an estimate for the uncertainty in that period .
we set a small range around the best candidate period , extending 0.05 days in both directions , and divided the range into 10,000 pieces .
after computing string lengths for each piece , we identified the range of periods which led to a local minimum in string length ( see figure [ fig : finetune ] ) . following the precepts of @xcite and @xcite , we fit a parabola to this local minimum and used the parameters of the fitted curve to estimate both the period and the uncertainty in the period .
based on a visual scan of the phased light curves of the residuals for the best three candidate periods , we selected 25 of the 3256 stars in the rr0 lyr catalog for further analysis .
after additional checks , we found that 11 stars show clear , periodic patterns in their residuals .
we list these 11 stars , which we shall call `` candidates '' , in table [ table : candidates ] .
values in the column labelled `` main period '' are taken from the catalog of @xcite , while those in the column labelled `` residual period '' were determined by this paper
. phased light curves of each candidate are shown in figures [ fig : cand1 ] to [ fig : cand11 ] .
l l l r l l l l macho i d & @xmath12 ra@xmath13 & @xmath12 dec@xmath13 & @xmath14 & main period@xmath15 & residual period@xmath15 & amp@xmath16 & notes@xmath17 + 101.21042.00862 & 18:05:18.60 & -27:16:14.5 & 17.55 & 0.413096 & @xmath18 & 0.13 & a + 102.22598.00556 & 18:08:54.96 & -27:31:49.4 & 17.36 & 0.444434 & @xmath19 & 0.13 & a + 115.22566.00425 & 18:09:09.00 & -29:40:30.4 & 16.83 & 0.530609 & @xmath20 & 0.08 & d + 121.21518.00593 & 18:06:25.92 & -30:12:03.2 & 17.44 & 0.452635 & @xmath21 & 0.07 & c + 124.22289.00461 & 18:08:15.72 & -30:50:30.5 & 17.72 & 0.498851 & @xmath22 & 0.12 & + 125.23719.00262 & 18:11:45.96 & -30:50:13.2 & 16.47 & 0.684945 & @xmath23 & 0.10 & b + 147.31018.00579 & 18:28:43.68 & -29:31:55.6 & 18.45 & 0.528649 & @xmath24 & 0.22 & + 159.25743.00689 & 18:16:18.12 & -25:52:22.4 & 17.56 & 0.459778 & @xmath25 & 0.15 & b + 163.27167.00172 & 18:19:39.36 & -26:15:58.0 & 16.98 & 0.428079 & @xmath26 & 0.16 & + 307.35371.00305 & 18:14:57.12 & -24:04:48.7 & 17.62 & 0.611346 & @xmath27 & 0.07 & a + 311.38064.00227 & 18:19:33.60 & -23:46:16.0 & 16.86 & 0.528800 & @xmath28 & 0.17 & c + + + + + +
none of these candidates shows the sharp , narrow dips of a detached system .
we suggest two reasons to explain the absence of such stars : first , stars with dips of large amplitude may have been excluded from the catalog of @xcite , if that catalog was constructed to contain only stars with light curve shapes of an isolated rrab variable .
correspondence with the first author of @xcite suggests that this was not the case , but we can not dismiss it as a possibility .
second , stars with dips of small amplitude may have escaped our analysis due to the relatively low signal - to - noise ratio of the individual measurements .
the dataset of @xcite includes an estimated uncertainty with each magnitude measurement .
we computed the mean value of the uncertainties for each star ; the median over 3565 stars of all those mean uncertainties is 0.023 mag in the red band and 0.028 mag in the blue band .
a system of shallow eclipses , with an amplitude of only two or three times the typical uncertainty and involving only a small fraction of the measurements , would not be found by our methods .
the smallest amplitude among our 11 candidates is 0.07 mag peak - to - peak , and that variation involves all the measurements in the light curve , not just a few .
all the candidates have gently undulating light curves , rather than the sharp dips of a detached eclipsing system .
what might cause this sort of periodic variation ? among the possibilities are artifacts from the subtraction of the model light curve ; aliases of the rr lyr frequency ; an additional frequency of oscillation in the rr lyr star ; blazhko variations in the rr lyr star ; blended light from another variable star(s ) ; or the orbital motion of the rr lyr star around a close companion .
let us examine these possibilities .
our method for subtracting the main rr lyr light curve from each star s measurements was based on a simple model made of linear segments .
if the model failed to reproduce some features properly , the subtraction would leave a signal with the same period as the rr lyr pulsation .
our 11 candidates include 3 stars for which the residual period is within 3 percent of the main rr lyr period ; we mark these in table [ table : candidates ] with a `` a '' in the `` notes '' column .
it is possible that these candidates may be due to a low - amplitude version of the blazhko effect , of which we say more below .
the majority of the measurements described in @xcite were collected on a nightly basis ; that is , each star was observed once per night , and often at roughly the same time .
we therefore expect to see aliases of the true frequency @xmath29 in the measurements with frequencies where @xmath31 and @xmath32 is some small integer .
we computed the alias frequencies for all candidates using @xmath33 and @xmath34 , and compared them to the frequencies of the residual variations .
we found two cases in which the residual frequencies were within 1 percent of the main rr lyr frequency ( marked `` b '' in table [ table : candidates ] ) , two more cases within 2 percent ( marked `` c '' ) and one more case within 3 percent ( marked `` d '' ) .
some rr lyr stars are known to oscillate at two frequencies ; these double - mode stars always have a ratio of periods @xmath35 @xcite .
none of our candidates have periods in their residuals which yield this ratio with the periods listed in @xcite .
the lack of such double - mode pulsators may not be unexpected , since they appear to be very rare in the central regions of our milky way ; @xcite found only 3 such stars among a sample of 1942 rrab and 771 rrc stars observed near the center of the milky way in the ogle - ii database @xcite .
their absence may also be a reflection of the selection criteria used by @xcite to create the catalog of rr lyr stars .
some rr lyr stars exhibit slow changes in the shape and amplitude of their light curves , with periods of tens to hundreds of days ; this is known as the blazhko effect .
@xcite finds roughly 25 percent of all rrab stars in a sample near the galactic center show the blazhko effect .
could it be responsible for any of our candidates ?
macho 124.22289.00461 , with a residual period of just over 17 days , is the only candidate for which this seems a possibility .
unfortunately , since the main rr lyr period is almost exactly half a day , the measurements made during each observing season cover only a small range in phase ; thus , we can not see if the shape of the light curve changes over this 17-day interval .
the three stars marked with an `` a '' in table [ table : candidates ] have residual periods which could be produced by blazhko - like periods of 15 to 50 days beating against the main rr lyr period ; however , we examined the phased light curves of these stars over each season and see no strong evidence of blazhko variations .
the macho study area in the galactic bulge was , by design , chosen to have a very high stellar density .
if the density is high enough , we may expect that blends of unrelated foreground or background variable stars may cause periodic signals in the residuals of rr lyr light curves .
let us perform a very quick quantitative check on this idea .
we examined one of the fields in this area , number 102 , counting all the stars in the macho database ( not just the variable ones ) as a function of apparent @xmath36-band magnitude .
since the typical fwhm in these measurements is @xmath37 , a rough estimate of the area of the seeing disk is about 10 square arcseconds .
we find that , on average , there are 1.8 stars with @xmath38 in each seeing disk , and about 0.5 stars with @xmath39 .
the rr lyr stars in the catalog of @xcite range roughly @xmath40 .
so indeed a significant fraction of all the rr lyr stars in the catalog must be contaminated by light from nearby stars at a level of 0.1 mag .
it may in fact be surprising that we find so few objects with periodic variations in their residuals ; however , since we do not know the details of the process by which rr lyr stars were selected from the macho database , we can not comment further .
could any of the candidates be due to the effects of a binary companion of the rr lyr star ?
if two stars orbit each other with a separation which is only slightly larger than their combined radii , their shapes may grow distorted enough that they produce a gently undulating light curve , even in the absence of eclipses .
an rr lyr star of mass @xmath41 with a companion of equal mass in a circular orbit of period 2 days would have a separation of about 7 @xmath42 .
the radius of a typical rr lyr star varies from about 4 @xmath42 to 6 @xmath42 @xcite , leaving little room for a companion .
if an rr lyr star did orbit a more compact companion with a period in this range , it would surely be greatly distorted , and so liable to vary in brightness as it moved in its orbit .
whether an rr lyr star would have stable pulsations in such a close orbit is beyond the scope of this paper .
note that in this situation , the period listed in table [ table : candidates ] would be _ half _ of the orbital period .
we conclude that our search through a sample of 3256 rrab lyr stars failed to find any detached eclipsing binary systems , and very likely failed to find eclipsing systems of any sort with amplitudes of @xmath43 mag .
the implied disjunction between stars pulsing in the fundamental rrab lyr mode and stars in binary systems may provide clues to the evolution of rrab lyr stars .
our simple technique for removing the ordinary variation of light in order to seek some signal in the residuals would be well suited to the more sinusoidal variations of rrc lyr stars , many of which can be found in the catalogs of @xcite and @xcite .
this paper utilizes public domain data obtained by the macho project , jointly funded by the us department of energy through the university of california , lawrence livermore national laboratory under contract no .
w-7405-eng-48 , by the national science foundation through the center for particle astrophysics of the university of california under cooperative agreement ast-8809616 , and by the mount stromlo and siding spring observatory , part of the australian national university .
the author thanks doug welch for reviewing an early version of this paper and providing many suggestions which improved it .
alcock , c. , et al . : 1997 , apj 486 , 697 alcock , c. , et al . : 1999 , pasp 111 , 1539 belserene , e. p. : 1983 , in r. m. genet , a. m. genet ( eds . ) , _ microcomputers in astronomy _ , fairborn observatory , p. 229 bhatti , w. a. : 2010 , apjs 186 , 233 dworetsky , m. m. : 1983 , mnras 203 , 917 fernie , j. d. : 1989 , pasp 101 , 225 hardie , r. : 1955 , apj 122 , 256 jameson , r. f. : 1986 , va 29 , 17 jurcsik , j. , et al .
: 2009 , mnras 400 , 1006 kunder , a. , et al . :
2008 , aj 135 , 631 mizerski , t. : 2003 , aca 53 , 307 nemec , j. m. : 1985 , aj 90 , 240 preston , g. w. : 1964 , ara&a 2 , 23 prsa , a. , et al . : 2008 , a&a 489 , 1209 sandage , a. , & tamman , g. a. : 2006 , ara&a 44 , 93 simon , n. r. : 1985 , apj 299 , 723 simon , n. r. , & teays , t. j. : 1982 , apj 261 , 586 smith , h. a. : 1995 , _ rr lyrae stars _ , cambridge university press sdor , .
: 2011 , arxiv 1101.5099 sdor , .
, jurcsik , j. , szeidl , b. : 2009 , mnras 394 , 261 soszyski , i. , et al .
: 2003 , aca 53 , 93 soszyski , i. , et al . :
2008 , aca 58 , 293 soszyski , i. , et al . : 2009 , aca 59 , 1 szczygiel , d. m. , & fabrycky , d. c. : 2007 , mnras 377 , 1263 udalski , a. , kubiak , m. , & szymaski , m. : 1997 , aca 47 , 319 wade , r. a. , et al . : 1999 , aj 118 , 2442 | the macho project collected photometry of many rr lyrae stars from its observations of the milky way s bulge .
we examined the lightcurves of 3256 stars identified as rrab lyr variables by @xcite , subtracting an empirical model of the pulsation lightcurve and searching for periodic variation in the residuals .
there are no systems which show the brief dips in light characteristic of detached eclipsing binary systems .
we discuss the results for objects which show the largest residual periodic modulation , most of which are probably due to aliases of the fundamental period . |
dynamical networks @xcite model physical and biological behavior in many applications ; examples range from networks of neurons @xcite , josephson junctions arrays @xcite to genetic networks @xcite , protein interaction nets @xcite and metabolic networks @xcite . synchronization in dynamical networks
is influenced by the topology of the network @xcite . a great need exists for the development of effective methods of inferring network structure from time series data ;
a method for detecting the topology of dynamical networks , based on chaotic synchronization , has been proposed in @xcite ; a recent approach deals with the case of low number of samples and proposed methods rooted on l1 minimization @xcite .
granger causality has become the method of choice to determine whether and how two time series exert causal influences on each other @xcite .
this approach is based on prediction : if the prediction error of the first time series is reduced by including measurements from the second one in the linear regression model , then the second time series is said to have a causal influence on the first one .
this frame has been used in many fields of science , including neural systems @xcite , reo - chaos @xcite and cardiovascular variability @xcite .
the estimation of linear granger causality from fourier and wavelet transforms of time series data has been recently addressed @xcite .
kernel algorithms work by embedding data into a hilbert space , and searching for linear relations in that space @xcite .
the embedding is performed implicitly , by specifying the inner product between pairs of points @xcite .
we have recently exploited the properties of kernels to provide nonlinear measures of bivariate granger causality @xcite .
we reformulated linear granger causality and introduced a new statistical procedure to handle over - fitting @xcite in the linear case .
our new formulation was then generalized to the nonlinear case by means of the kernel trick @xcite , thus obtaining a method with the following two main features : ( i ) the nonlinearity of the regression model can be controlled by choosing the kernel function ; ( ii ) the problem of false - causalities , which arises as the complexity of the model increases , is addressed by a selection strategy of the eigenvectors of a reduced gram matrix whose range represents the additional features due to the second time series . in this paper
we describe in detail the kernel granger approach and address use of granger causality to estimate , from multivariate time series data , the topology and the drive - response relationships of a dynamical network . to this aim ,
we generalize our method in @xcite to the case of multivariate data .
the paper is organized as follows . in the next section
we describe the kernel granger causality method in the bivariate case , adding some details to the presentation in @xcite . in section
iii we generalize the method to process multivariate time series data , and show that the proposed method can discern whether an interaction is direct or mediated by a third time series .
the case of dynamical networks is described in section iv : we analyze a system of interacting chaotic maps and a model of gene regulatory networks . in section
v we consider systems with sparse connectivity and limited data , and compare the bivariate granger approach with a multivariate method based on l1 minimization ; then we analyze a real data set , gene expressions of hela cells .
section v summarizes our conclusions .
in this section we review the kernel method for granger causality proposed in @xcite .
let us start with the linear case .
suppose to model the temporal dynamics of a stationary time series @xmath1 by an autoregressive model of order @xmath2 : @xmath3 and by a bivariate autoregressive model which takes into account also a simultaneously recorded time series @xmath4 : @xmath5 the coefficients of the models are calculated by standard least squares optimization ; @xmath2 is usually chosen according to akaike criterion @xcite or embedding techniques from the theory of nonlinear dynamical systems @xcite .
the concept of granger causality is @xcite : @xmath6 granger - causes @xmath7 if the variance of residuals @xmath8 is significantly smaller than the variance of residuals @xmath9 , as it happens when coefficients @xmath10 are jointly significantly different from zero .
this can be tested by performing an f - test or levene s test for equality of variances @xcite .
an index measuring the strength of the causal interaction is then defined as : @xmath11 where @xmath12 means averaging over @xmath13 ( note that @xmath14 ) .
exchanging the roles of the two time series , one may equally test causality in the opposite direction , i.e. to check whether @xmath7 granger - causes @xmath6 .
we use the shorthand notations : @xmath15 @xmath16 and @xmath17 , for @xmath18 .
we treat these quantities as @xmath19 realizations of the stochastic variables @xmath20 , @xmath21 and @xmath22 .
let us denote @xmath23 the @xmath24 matrix having vectors @xmath25 as columns , and @xmath26 the @xmath27 matrix having vectors @xmath28 as columns .
the values of @xmath22 are organized in a vector @xmath29=@xmath30 . in full generality
we assume that each component of @xmath20 and @xmath21 has zero mean , and that vector @xmath29 has zero mean and is normalized , i.e. @xmath31=@xmath32 . now , for each @xmath18 , we define : @xmath33 @xmath34 the vectors @xmath35=@xmath36 and @xmath37=@xmath38 are the estimated values by linear regression , in the two cases .
it is easy to show that @xmath35 and @xmath37 have the following geometrical interpretation .
let @xmath39 be the range of the @xmath40 matrix @xmath41 .
then @xmath35 is the projection of @xmath29 on @xmath42 . in other words , calling @xmath43 the ( orthonormal ) eigenvectors of @xmath44 with non - vanishing eigenvalue and @xmath45 the projector on the space @xmath42 , we have @xmath46 .
let us define @xmath47 .
analogously @xmath48 , @xmath49 being the projector on the @xmath50-dimensional space @xmath51 , equal to the range of the matrix @xmath52 . moreover , it is easy to show that @xmath53 now note that @xmath54 .
therefore we may decompose @xmath55 as follows : @xmath56 , where @xmath57 is the space of all vectors of @xmath55 orthogonal to all vectors of @xmath42 .
@xmath57 corresponds to the additional features due to the inclusion of @xmath58 variables .
calling @xmath59 the projector on @xmath57 , we can write : @xmath60 now we note that @xmath57 is the range of the matrix @xmath61 indeed , for any @xmath62 , we have @xmath63 , where @xmath64 , and @xmath65 .
it follows that @xmath57 is spanned by the set of the eigenvectors , with non vanishing eigenvalue , of @xmath66 . calling @xmath67 these eigenvectors
, we have : @xmath68 where @xmath69 is the pearson s correlation coefficient of @xmath70 and @xmath71 ( since the overall sign of @xmath71 is arbitrary , we can assume that @xmath69 is positive ) .
let @xmath72 be the probability that @xmath69 is due to chance , obtained by student s t test .
since we are dealing with multiple comparison , we use the bonferroni correction to select the eigenvectors @xmath73 , correlated with @xmath70 , with expected fraction of false positive q ( equal to @xmath74 ) .
therefore we calculate a new causality index by summing , in equation ( 4 ) , only over the @xmath75 such that @xmath76 , thus obtaining a _ filtered _ linear granger causality index : @xmath77 it is assumed that @xmath78 measures the causality @xmath79 , without further statistical test . in this subsection
we describe the generalization of linear granger causality to the nonlinear case , using methods from the theory of reproducing kernel hilbert spaces ( rkhs ) @xcite . given a kernel function @xmath80 , with spectral representation @xmath81 ( see mercer s theorem @xcite ) , we consider @xmath42 , the range of the @xmath40 gram matrix @xmath82 with elements @xmath83 . in order to make the mean of all variables @xmath84 zero , we replace @xmath85 , where @xmath86 is the projector onto the one - dimensional subspace spanned by the vector such that each component is equal to the unity @xcite ; in the following we assume that this operation has been performed on each gram matrix .
as in the linear case , we calculate @xmath35 , the projection of @xmath29 onto @xmath42 . due to the spectral representation of @xmath80 , @xmath35 coincides with the linear regression of @xmath29 in the feature space spanned by @xmath87 , the eigenfunctions of @xmath80 ; the regression is nonlinear in the original variables .
while using both @xmath20 and @xmath21 to predict @xmath22 , we evaluate the gram matrix @xmath88 with elements @xmath89 .
the regression values now form the vector @xmath37 equal to the projection of @xmath29 on @xmath55 , the range of @xmath88 . before evaluating the filtered causality index , as in the linear case , we note that not all kernels may be used to evaluate granger causality .
indeed if @xmath21 is statistically independent of @xmath20 and @xmath22 , then @xmath37 and @xmath35 should coincide in the limit @xmath90 .
this property , invariance of the risk minimizer when statistically independent variables are added to the set of input variables , is satisfied only by suitable kernels , as discussed in @xcite . in the following we consider two possible choices , which fulfill the invariance requirement .
_ inhomogeneous polynomial kernel_. the inhomogeneous polynomial ( ip ) kernel of integer order @xmath91 is : @xmath92 in this case the eigenfunctions are made of all the monomials , in the input variables , up to the @xmath93 degree .
the dimension of the space @xmath42 is @xmath94 , where @xmath95 is the beta function , and @xmath96 corresponds to the linear regression .
the dimension of space @xmath55 is @xmath97 . as in the linear case ,
we note that @xmath54 and decompose @xmath56 .
subsequently we calculate @xmath98 ; the dimension of the range of @xmath66 is @xmath99 . along the same lines as those described in the linear case
, we construct the kernel granger causality taking into account only the eigenvectors of @xmath66 which pass the bonferroni test : @xmath100 the sum being only over the eigenvectors of @xmath66 with probability @xmath101 . _
gaussian kernel_. the gaussian kernel reads : @xmath102 and depends on the width @xmath103 .
@xmath103 controls the complexity of the model : the dimension of the range of the gram matrix decreases as @xmath103 increases .
as in previous cases , we may consider @xmath42 , the range of the gram matrix @xmath82 , and @xmath55 , the range of @xmath88 , but in this case the condition @xmath54 would not necessarily hold ; therefore some differences in the approach must be undertaken .
we call @xmath104 the @xmath105-dimensional span of the eigenvectors of @xmath82 whose eigenvalue is not smaller than @xmath106 , where @xmath107 is the largest eigenvalue of @xmath82 and @xmath108 is a small number ( we use @xmath109 ) .
we evaluate @xmath110 , where @xmath111 is the projector on @xmath104 .
after evaluating the gram matrix @xmath112 , the following matrix is considered : @xmath113 where @xmath114 are the eigenvectors of @xmath112 , and the sum is over the eigenvalues @xmath115 not smaller than @xmath108 times the largest eigenvalue of @xmath112 .
then we evaluate @xmath116 , and denote @xmath59 the projector onto the @xmath117-dimensional range of @xmath66 .
note that the condition @xmath118 may not be strictly satisfied in this case ( however in our experiments we find that the violation of this relation is always very small , if any ) .
the kernel granger causality index for the gaussian kernel is then constructed as in the previous case , see equation ( [ causalityindex5 ] ) .
let @xmath119 , @xmath120 , be @xmath121 simultaneously recorded time series . in order to put in evidence the drive - response pattern in this system
, one may evaluate the bivariate granger causality , described in the previous sections , between every pair of time series .
it is recommended , however , to treat the data - set as a whole , thus generalizing kernel granger causality to the multivariate case , as follows .
we denote @xmath122 for @xmath123 and @xmath18 . in order to evaluate the causality @xmath124
, we define , for @xmath18 , @xmath125 containing all the input variables , and @xmath126 containing all the input variables but those related to @xmath127 . gram matrices @xmath82 and @xmath112 are then evaluated : @xmath128 and @xmath89 .
the target vector is now @xmath29=@xmath129 . along the same lines as in the bivariate case , for ip kernel or the gaussian one
, we then calculate the causality index as in equation ( 6 ) : it is denoted @xmath130 and measures the strength of causality @xmath131 , taking into account all the available variables . repeating these steps for all @xmath132 and @xmath133 ,
the casuality pattern in the data set is evaluated .
note that the threshold for the bonferroni s correction , in the multivariate case , must be lowered by the number of pairs @xmath134 .
as an example , we consider the following system of three logistic maps @xcite : @xmath135 here @xmath136 , @xmath137 and @xmath138 s are unit variance gaussian noise terms .
the causal relationships implemented in these equations are @xmath139 and @xmath140 . analyzing segments of length @xmath141
, we evaluate both the multivariate causality , as described in section 4 , and the bivariate causality for all pairs of maps .
we use the ip kernel with various values of @xmath91 ; the results are displayed in figure ( [ fig1 ] ) .
it is well known @xcite that performing pairwise evaluation for multivariate data has the drawback that one can not discern whether the influence between two time series is direct or is mediated by other time series .
this is what happens in the present example .
both the multivariate and the bivariate analysis reveal the influences @xmath139 and @xmath140 . on the other hand , the bivariate analysis reveals also the influence @xmath142 , which is actually mediated by @xmath32 : the multivariate analysis recognizes @xmath142 as non - significative .
in this section we simulate two dynamical networks and apply the multivariate granger analysis to estimate the topology structure of systems from time series data .
let us consider a coupled map lattice of @xmath13 nodes , with equations , for @xmath143 : @xmath144 where @xmath132 , @xmath145 and @xmath138 s are the same as in equation ( [ map ] ) , and @xmath146 represents the coupling @xmath147 .
we fix @xmath148 and construct couplings as follows .
we consider the well known zachary data set @xcite , an undirected network of @xmath149 nodes .
we assign a direction to each link , with equal probability , and set @xmath146 equal to 0.05 , for each link of the directed graph thus obtained , and zero otherwise .
the network is displayed in figure ( [ fig2 ] ) : here the goal is to estimate this directed network from the measurements of time series on nodes .
the multivariate granger analysis , described in the previous section , perfectly recovers the underlying network using the ip kernel with @xmath150 , @xmath151 and @xmath152 . note that while evaluating @xmath153 , for all @xmath154 and @xmath155 , 39270 pearson s coefficients @xmath156 are calculated .
their distribution is represented in figure ( [ fig3 ] ) : there is a strong peak at @xmath157 ( corresponding to projections that are discarded ) , and a very low number of projections with a rather large value of @xmath156 , up to about @xmath158 .
it is interesting to describe the results in terms of a threshold for correlations . given a threshold value @xmath156 , we select the correlation coefficients whose value is greater than @xmath156 .
we then calculate the corresponding causality indexes @xmath159 , and construct the directed network whose links correspond to non - vanishing elements of @xmath153 . in figure ( [ fig4])-top
we plot the total number of links of the reconstructed network , as a function of the threshold @xmath156 : the curve shows a plateau , around @xmath160 , corresponding to a directed network which is stable against variations of @xmath156 . at the plateau @xmath161 projections
are selected , which coincide with those selected by means of bonferroni s test . in figure ( [ fig4])-bottom
we plot the number of errors ( the sum of the number of links that exist in the true network and are not recovered , plus the number of links that exist only in the recovered network ) versus the threshold @xmath156 : the plateau leads to perfect reconstruction of the original network . we stress that a large number of samples is needed to recover the underlying network : in the typical case we find that the network is perfectly reconstructed if @xmath162 , whilst if @xmath19 is further lowered some errors in the reconstructed network appear .
moreover , it is important to observe that , although all couplings @xmath146 have the same magnitude , the causality strengths @xmath159 depend on the link , as it is shown in figure ( [ fig5 ] ) .
granger causality is a measure of the information being transferred from one time series to another , and it should not be identified with couplings . in this subsection
we consider time series from a model of genetic regulatory network made of genes linked by weighted connections ( inhibitory or excitatory ) @xcite .
the expression levels of all genes , organized in a vector @xmath163 , evolve as follows : @xmath164 where @xmath165 is a connectivity strength matrix corresponding to the network , @xmath166 , @xmath167 is the identity matrix and @xmath168 is a vector of random variables uniform in @xmath169 $ ]
. the values of @xmath163 are restricted by floor and ceiling function to range in @xmath170 $ ] : this constraint provides the nonlinear character of the model .
as the simulation runs , multivariate data are sampled every @xmath171 time steps . moreover
, the continuous data values are discretized into @xmath172 categories ( with equal bin sizes ) . in @xcite
dynamic bayesian network ( dbn ) models @xcite were trained to data of length @xmath19 to recover the structure of matrix @xmath165 : the values @xmath173 , @xmath174 and @xmath175 were found to lead to the best reconstruction of genetic networks by dbn .
the genetic network we consider here is an example from @xcite and consists of ten genes with connections described in figure ( [ fig6 ] ) : there are two independent regulatory pathways , one of which includes a large feedback structure . in figure ( [ fig7 ] ) the typical curve of expression for a gene in the network
is represented ( top ) ; the distribution of expressions , for the same gene , is bimodal ( bottom ) .
we simulate 100 times the system equations , starting from different initial conditions , and sample time series of length @xmath176 , @xmath173 and @xmath177 .
the typical recovered network by dbn , on this example , corresponds to one missing link ( from node 7 to 10 ) @xcite .
the linear multivariate granger approach , with @xmath178 , leads to perfect reconstruction of the network in 90 cases out of the 100 ; we obtain similar results on all the examples presented in @xcite . using ip ( with @xmath179 ) and gaussian kernels we obtain similar performances as the ones from the linear kernel .
it is interesting to stress that the possibility that one has to reconstruct the true genetic network depends on the sampling rate . in figure ( [ fig8 ] )
we plot the mean number of errors ( over 100 realizations ) in the reconstructed network as a function of @xmath171 for the linear kernel and gaussian kernel : the performance degrades as @xmath171 gets far from 5 .
let us now discuss the case of large @xmath171 : all granger causalities are recognized as non significative . on the other hand , at large @xmath171 , we find significant _
static _ linear correlations between time series of all pairs of genes belonging to the same pathway . in other words , referring to figure ( [ fig6 ] ) , the linear correlations of times series from every pair of genes extracted from @xmath180 is significant , as well as the linear correlation for every pair extracted from @xmath181 ; consistently the linear correlation , for pairs of genes from different pathways , is not significant .
we conclude that , at large @xmath171 granger causality analysis is not effective , but static correlation analysis may anyway put in evidence groups of genes belonging to the same regulatory pathway .
the granger causality approach for dynamical networks here presented requires a large amount of data samples to provide trustable answers .
however , there are situations ( frequent in bioinformatics ) where the number of samples is smaller than the number of variables ( genes ) : in these situations the multivariate granger approach is unfeasible . under the assumption of sparse connectivity
, it has been proposed to replace least squares methodology with a multivariate approach using minimization with respect to @xmath0 norm @xcite . here
we show that there are situations where , even in presence of sparse connectivity and limited data , use of bivariate granger causality is a better choice w.r.t @xmath0 minimization . indeed , in these cases
, the statistical robustness of the estimation of information flow between pairs of time series may still be good , with the drawback that some causality links , found by the bivariate approach , may not be direct but mediated .
we construct a network of @xmath182 nodes and @xmath182 links using the preferential attachment procedure @xcite ; we give randomly a direction to each link , with @xmath183 probabilities , thus obtaining a directed network .
let us denote @xmath184 the number of nodes from which a link pointing to @xmath154 starts .
we evolve a linear system on this network , with equations : @xmath185 the sum is over nodes such that @xmath147 is a link of the network ; @xmath138 s are unit variance gaussianly distributed noise terms ; @xmath186 is one , if @xmath187 , and @xmath188 otherwise . after a transient , we sample @xmath189 consecutive time points , with @xmath190 .
the @xmath0 approach we use is the following .
for each @xmath191 , we find the vector @xmath192 with minimum @xmath0 norm , among all those satisfying : @xmath193 @xmath194 .
the interaction @xmath147 is considered significant if the absolute value of @xmath195 exceed a threshold , fixed so that the total number of false positive connections is five .
subsequently we apply the bivariate linear granger approach , described in section iia , for each pair of nodes : also for granger s approach we fix a threshold for the correlation coefficients @xmath156 , see equation ( [ causalityindex3 ] ) , so that the total number of false positive connections is five . in figure ( [ fig9 ] )
we depict the number of true positive connections found by the two approaches , as a function of @xmath189 .
it is clear that here the bivariate granger approach outperforms @xmath0 minimization .
hela is the most famous cell culture line to date @xcite .
these are cells isolated from a human uterine cervical carcinoma in 1951 and used in biomedical research especially to culture viruses .
whilst the patient ultimately died of her cancer eight months after the operation , her cells have lived on , still surviving in laboratories today .
hela cells have somehow acquired cellular immortality , in that the normal mechanisms of programmed cell death after a certain number of divisions have somehow been switched off .
we apply our approach to the hela cell gene expression data of @xcite .
data corresponds to @xmath196 genes and @xmath197 time points , with an hour interval separating two successive readings ( the hela cell cycle lasts 16 hours ) .
the 94 genes were selected , from the full data set described in @xcite , on the basis of the association with cell cycle regulation and tumor development .
first of all , we perform the analysis of the static pairwise correlations between time series : 800 pairs of genes are significantly correlated .
drawing a link for each correlated pair leads to an undirected network depicted in figure ( [ fig10 ] ) : it is clear that there are two modules , and symbols in figure ( [ fig10 ] ) corresponds to the partition by the method of module identification described in @xcite .
the first module is made of 23 genes and corresponds to the regulatory network of the transcriptional factor nfkb @xcite ; it contains several well known activators and targets of nfkb @xcite , like , e.g. , a20 , icam-1 , il-6,vcam-1 , ikappaba , junb , mcp-1 , fgf2 , cyclin .
the second module , 62 genes , appears to be orchestrated by transcriptional factors p53 and stat3 .
note , however , that the two modules are not independent , as they form a highly related network .
the proto - oncongene c - myc appears to be central between the two modules : it has 12 significant static correlations with both modules .
after the discussion in section iv - b , we assume that the modular structure depicted in figure ( [ fig10 ] ) is the result of regulatory mechanisms acting on time scales much smaller than the sampling time . next , in order to detect causalities acting on the time scale of the sampling time , we apply bivariate granger causality analysis . for all pairs @xmath198
, we use the linear version ( @xmath96 , @xmath151 ) of our approach and evaluate the pearson correlation coefficient @xmath156 , equation ( [ causalityindex3 ] ) , for the causality @xmath199 . due to the small number of samples , we do not use t - test to evaluate the probability @xmath200 corresponding to @xmath156 : we generate a set of surrogates by permuting the temporal indices of the i - th times series while keeping fixed those of the j - th time series . the probability @xmath200 is identified with the fraction of times that an higher coefficient is obtained over @xmath201 random shufflings of time indices of the i-@xmath202 time series .
moreover , we use the false discovery rate ( fdr ) method @xcite instead of bonferroni s correction .
fdr works in the following way : the @xmath203 pearson coefficients are ordered , @xmath204 , according to their increasing @xmath205 values , and a parameter @xmath206 , which controls the fraction of false positive , is set to @xmath74 . the index @xmath207 is identified as the largest such that for all @xmath208 we have @xmath209 .
pearson coefficients @xmath210 are accepted for @xmath211 .
this procedure selects 19 causal relationships , out of 8742 ; they are listed in table i. ikappaba is the most abundant inhibitory protein for nfkb @xcite : our approach detects the significant causality ikappaba @xmath212 nfkb .
we find that nfkb is also casually related to iap , an anti - apoptotic gene , and b99 ( a direct target for transcriptional activation by p53 : here no significant interaction between b99 and p53 has been detected ) .
three causality relationships involve bcl - xl , the dominant regulator of apoptosis ( active cell suicide ) and tsp1 , a peptide shown in some tumor systems to be linked with angiogenesis .
notably table i also contains fibroblast growth factors , fgf7 and fgfr4 , the tumor necrosis factor killer /dr5 , the myeloid tumor suppressor gene pkig , the tumor protein tpd52-l and cyclin e1 , a gene which is overexpressed in many tumors . in @xcite data have been analyzed with the sparse vector autoregressive model , a multivariate l1 approach which depends on a regularization parameter , @xmath213 , fixed by cross - validation .
only one causality relationship , out of the 19 in table i , was revealed also in @xcite : a20@xmath212bcl - xl .
.causalities for hela gene network . [ cols="^,^,^ " , ]
our method of analysis of dynamical networks is based on a recent measure of granger causality between time series , rooted on kernel methods , whose magnitude depends on the amount of flow of information from one series to another . by definition of granger causality ,
our method allows analysis of networks containing cycles .
firstly we have demonstrated the effectiveness of the method on a network of chaotic maps with links obtained assigning a direction to the edges of the well known zachary data set , using a nonlinear kernel : perfect reconstruction of the directed network is achieved provided that a sufficient number of samples is available .
secondly we studied a simulated genetic regulatory network .
the results from our method were better than those from dbn approach .
however our performance was strongly dependent on the sampling time , as it occurred also using dbn method . in this example
, use of ip kernel , with ( @xmath214 ) , or gaussian kernels did not lead to improvement in the performance with respect to the linear kernel : this means that these kernel are not suitable to model the nonlinear constraint connected to the fact that expressions are confined in @xmath170 $ ] .
further work will be devoted to the search for kernels capable to capture this kind of nonlinearity : for a given application one should choose the proper kernel out of the many possible classes @xcite .
then we considered the case of sparse connectivity and limited data .
using an example consisting in a linear dynamical network on a graph grown by preferential attachment , we have shown that there are instances where the multivariate granger approach is unfeasible , but the application of bivariate granger analysis , to every pair of time series , leads to better results than those from a method based on l1 minimization .
finally we have analyzed a real data set of temporal gene expression samples from hela cells . the static correlation analysis between time
series , which is the result of regulation mechanisms with time scaler faster than the sampling rate , revealed the presence of two modules .
use of bivariate granger causality has put in evidence 19 causality relationships acting on the time scale of one hour , all involving genes playing some role in processes related to tumor development .
our result on hela data has very little overlap with those from the output of a method based on multivariate l1 minimization , but this is not surprising , as we observed the same fact also on the linear dynamical model of section v , where the true connectivity was known .
we remark that currently available data size and data quality make the reconstruction of gene networks from gene expression data a challenge . detecting cause - effects influences between components of a complex system
is an active multidisciplinary area of research in these years .
the kernel approach here presented , provides a statistically robust tool to assess drive - response relationships in many fields of science .
barabasi , _ linked : the new science of networks_. ( perseus publishing , cambridge mass . , 2002 ) ; s. boccaletti , v. latora , y. moreno , m. chavez and d .- u .
hwang , phys .
rep . * 424 * , 175 ( 2006 ) .. l.f .
abbott , c. van vreeswijk , phys . rev . *
e 48 * , 1483 ( 1993 ) .
k. wiesenfeld , physica * b 222 * , 315 ( 1996 ) .
d. bernardo , d. lorenz , j.j .
collins , science * 301 * , 102 ( 2003 ) . c.l .
tucker , j.f .
gera , p. uetz , trends cell biol . * 11 * , 102 ( 2001 ) .
h. jeong , b. tombor , r. albert , z.n .
oltvai , a.l .
barabasi , nature * 407 * , 651 ( 2000 ) .
newman , phys.rev.lett .
* 89 * , 208701 ( 2002 ) ; s. boccaletti , d.u .
hwang , m. chavez , a. amann , j. kurths , and l.m .
pecora , phys . rev .
* e 74 * , 16102 ( 2006 ) .
d. yu , m. righero , l. kocarev , phys.rev.lett . * 97 * , 188701 ( 2006 ) .
d. napoletani , t. sauer , phys . rev . *
e 77 * , 26103 ( 2008 ) .
granger , econometrica * 37 * , 424 ( 1969 ) ; for recent reviews see k. hlavackova - schindler , m. palus , m. vejmelka , j. bhattacharya , physics reports * 441 * , 1 ( 2007 ) , and m. lungarella , k. ishiguro , y. kuniyoshi , n. otsu , int . j. bifurcation and chaos * 17 * , 903 ( 2007 ) .
m. kaminski , m.z .
ding , w.a .
truccolo , s.l .
bressler , biological cybernetics * 85 * , 145 ( 2001 ) ; k.j .
blinowska , r. kus , m. kaminski , phys . rev .
* e 70 * , 50902(r ) ( 2004 ) ; a.k .
seth , network : computation in neural systems * 16 * , 35 ( 2005 ) .
r. ganapathy et al . , phys . rev .
* e 75 * , 16211 ( 2007 ) . l. faes , g. nollo , k.h .
chon , annals of biomedical engineering * 36 * , 381 ( 2008 ) .
m. dhamala , g. rangarajan , m. ding , phys.rev.lett .
* 100 * , 18701 ( 2008 ) . v. vapnik , _ statistical learning theory _
( john wiley & sons , inc . ,
j. shawe - taylor and n. cristianini , _ kernel methods for pattern analysis_. ( cambridge university press , london , 2004 ) .
d. marinazzo , m. pellicoro and s. stramaglia , _ kernel method for nonlinear granger causality _ , preprint arxiv:0711.2643 , accepted for publication in physical review letters .
m. palus , m. vejmelka , phys . rev . *
e 75 * , 56211 ( 2007 ) ; h. nalatore and m.z .
ding , phys . rev .
* e 75 * , 31123 ( 2007 ) ; k. ishiguro , n. otsu , m. lungarella , y. kuniyoshi , phys . rev . * e 77 * , 26216 ( 2008 ) .
h. akaike , ieee trans .
control * 19 * , 716 ( 1974 ) .
h. kantz , t. schreiber , _ nonlinear time series analysis _
( cambridge university press , cambridge , 1997 ) .
j. geweke , journal of the american statistical association , 77 , 304 ( 1982 ) .
n. ancona and s. stramaglia , neural comput .
* 18 * , 749 ( 2006 ) .
e. ott , _ chaos in dynamical systems_. ( cambridge university press , london , 2002 ) .
zachary , j. anthropol .
res . * 33 * , 452 ( 1977 ) .
j. yu , v.a .
smith , p.p .
wang , a.j .
hartemink , and e.d .
jarvis , bioninformatics * 20 * , 3594 ( 2004 ) . n. friedman , m. linial , i. nachman , and d. peer , j. comput .
* 7 * , 601 ( 2000 ) .
barabasi , r. albert , science * 286 * , 509 ( 1999 ) .
j.r . masters , nature reviews cancer * 2 * , 315 ( 2002 ) .
a. fujita et al . , bmc system biology * 1:39 * , 1 ( 2007 ) .
whitfield et al .
cell * 13 * , 1977 ( 2002 ) .
l. angelini , s. boccaletti , d. marinazzo , m. pellicoro , s. stramaglia , chaos * 17 * , 23114 ( 2007 ) .
j. benjamini , j. and y. hochberg , y. , j.r .
* b 57 * , 289(1995 ) . j. inoue , j. gohda , t. akiyama , k. semba , cancer sci . * 98 * , 268 ( 2007 ) . h. l. pahl , oncogene * 18 * , 6853 ( 1999 ) .
f. chen , v. castranova , x. shi , am . j. pathol .
* 159 * , 387 ( 2001 ) . ; the plot shows the number of links of the directed network constructed from the projections whose pearson s coefficient exceeds the threshold .
( bottom ) the total number of errors , in the reconstructed network , is plotted versus the threshold @xmath156 . at large @xmath156
the errors are due only to missing links , whereas at large @xmath156 the errors are due only to links that do not exist in the true network and are recovered .
[ fig4],width=377 ] ) obtained using linear kernel ( @xmath96 , empty circles ) and gaussian kernel ( stars ) , is plotted versus the sampling time @xmath171 . in the gaussian case
, @xmath215 is used .
the level of discretization is @xmath177 and @xmath178 .
[ fig8],width=377 ] minimization ( empty circles)on the preferential attachment dynamical network .
the error bar represents one standard deviation , evaluated over 50 realizations .
the number of false positive connections is always 5 .
[ fig9],width=377 ] | we propose a method of analysis of dynamical networks based on a recent measure of granger causality between time series , based on kernel methods .
the generalization of kernel granger causality to the multivariate case , here presented , shares the following features with the bivariate measures : ( i ) the nonlinearity of the regression model can be controlled by choosing the kernel function and ( ii ) the problem of false - causalities , arising as the complexity of the model increases , is addressed by a selection strategy of the eigenvectors of a reduced gram matrix whose range represents the additional features due to the second time series .
moreover , there is no _ a priori _ assumption that the network must be a directed acyclic graph .
we apply the proposed approach to a network of chaotic maps and to a simulated genetic regulatory network : it is shown that the underlying topology of the network can be reconstructed from time series of node s dynamics , provided that a sufficient number of samples is available . considering a linear dynamical network , built by preferential attachment scheme
, we show that for limited data use of bivariate granger causality is a better choice w.r.t methods using @xmath0 minimization .
finally we consider real expression data from hela cells , 94 genes and 48 time points .
the analysis of static correlations between genes reveals two modules corresponding to well known transcription factors ; granger analysis puts in evidence nineteen causal relationships , all involving genes related to tumor development . |
clusters of galaxies , the largest gravitationally - bound structures , are believed to form hierarchically in a sequence of cosmic structure formation where smaller groups of galaxies merge to form larger and richer systems ( geller & beers 1982 ; dressler & shectman 1988 ; girardi et al .
1997 ; kriessler & beers 1997 ; jones & forman 1999 ; schuecker et al . 2001 ; burgett et al .
the intra - group or intra - cluster medium ( icm ) of these clusters contains a hot ( t@xmath1k ) and tenuous ( n@xmath2 @xmath3 ) plasma which emits in x - rays mainly through bremsstrahlung and line emission ( kellogg et al .
1972 ; mitchell et al .
1979 ; also see reviews by sarazin 1988 ; mcnamara & nulsen 2007 ) .
although many rich clusters show smooth x - ray surface - brightness distributions with a central peak , a large fraction ( at least , @xmath030% ( forman & jones 1990 ) or even up to @xmath040%-60% ( jones & forman 1992 , 1999 ; schuecker et al . 2001 ) ) shows double or multiple peaks in their x - ray surface brightness maps . as x - ray surface - brightness is more sensitive to density than temperature
these observations effectively show peaks in the density distribution of the icm itself , which suggests the presence of subclusters or substructures which are possibly in the process of merging to form richer groups and clusters of galaxies ( e.g. nakamura et al .
presence of large inhomogeneities in the x - ray temperature of the icm ( greater than a few kev ) on scales less than 0.5 mpc provides additional observational evidence of mergers ( roettiger et al .
1996 ) . as dynamical evolution rapidly removes substructures , their existence proves that such clusters are dynamically young systems ( flin 2003 ) . for studies of cosmology and the formation and evolution of large - scale structures , it is important to study the physical properties of clusters of galaxies containing sub - groups which are young and in the process of merging
. mergers of subgroups can produce shocks , bulk gas flows and turbulence in the icm , and can disrupt cooling flows as has been seen in many clusters ( see fabian 1994 ; markevitch & vikhlinin 2007 ; owers et al .
2011 ; maurogordato et al .
mergers may also cause metal enrichment of the icm through enhanced ram - pressure stripping during mergers ( domainko et al .
2005 ) , thereby affecting the evolution of the galaxies and the icm .
subclustering information has also been used for estimating cosmological parameters such as the density parameter @xmath4 ( richstone et al .
1992 ; kauffmann & white 1993 , angrick & bartelmann 2011 ) . at radio wavelengths , besides the diffuse emission which may be in the form of halos and relics , the tailed radio sources associated with individual galaxies can provide further insights towards understanding these systems .
for example , the wide angle tail ( wat ) galaxies are usually associated with the dominant galaxy in a group or a cluster , with the jets flaring into tails of emission which can extend to over a mpc in size ( e.g. roettiger et al .
1996 ; douglass et al . 2008
; mao et al . 2010 ) .
their relatively high luminosity , close to the upper range of the fanaroff - riley i sources , and large sizes make these good tracers of galaxy clusters at moderate and high redshifts ( e.g. blanton et al .
2003 ; douglass et al .
2008 ) . since the dominant galaxies with which wats are associated are likely to have at most small peculiar velocities , their jets are unlikely to be bent by ram pressure caused by the motion of the galaxy ( burns 1981 ; eilek et al .
1984 ; odonoghue et al . 1990 ) .
cluster mergers where relative velocities between the systems can exceed @xmath01000 km s@xmath5 , causing significant bulk motion of the icm , have been invoked as the primary scenario for understanding the wat structures ( eilek et al .
1984 ; burns et al .
1994 ; roettiger et al . 1996 ) .
x - ray observatories , particularly chandra and xmm - newton , have brought to light new information leading to a more detailed understanding and new insights into the physical processes occurring in galaxy clusters ( weratschnig 2010 ) .
while chandra with its unprecedented spatial resolution of 0.5@xmath6 can study small - scale phenomena such as cold fronts , merger shocks and agn cavities , xmm - newton with its larger collecting area and sensitivity can study clusters to a much farther extent and can detect fainter x - ray signals . because of its large fov of @xmath7 and moderate resolution ( 6@xmath6 fwhm ) , xmm - newton can also study the nearby clusters very well .
we have chosen a3395 for a detailed study of its merging environment since it is a well - known bimodal merging cluster ( henriksen & jones 1996 ; markevitch et al .
1998 ; donnelly et al .
2001 ; flin & krywult 2006 ) and it has been observed with xmm - newton , chandra and australia telescope compact array ( atca ) . the paper is organized as follows .
detailed information about a3395 and a summary of its earlier x - ray observations with the roentgen satellite ( rosat ) and the advanced satellite for cosmology and astrophysics ( asca ) and the previous radio observations are given in
[ sec : a3395 ] .
the details of the recent x - ray and unpublished radio observations and their data analyses , are presented in [ sec : obs_n_data_red ] .
the results including the x - ray and radio morphology , x - ray surface brightness profiles , x - ray temperature , metallicity , density , entropy and pressure maps , mass , luminosity and cooling time estimates are provided in
[ sec : results ] .
a discussion of the results and the merging environment of a3395 is given in
[ sec : discussion ] .
a lambda cold dark matter cosmology with @xmath8 = 70 km s@xmath5 mpc@xmath5 and @xmath9 = 0.3 ( @xmath10 = 0.7 ) has been assumed , so that 1 arcsec corresponds to 0.96 kpc at z=0.0498 , the redshift of the cluster .
located nearby , at a distance of 211 mpc and redshift of 0.0498 , the positional coordinates of a3395 are : r.a.(j2000 ) @xmath11 , dec.(j2000)@xmath12 ( l=262.9589@xmath13 , b=-25.007@xmath13 ) ( from set of identifications , measurements , and bibliography for astronomical database ( simbad ) ) .
a3395 is a regular cluster with 54 members ( cluster members between m@xmath14 ( magnitude of @xmath15 brightest galaxy ) and m@xmath14 + 2 ) and richness class 1 ( abell , corwin , & olowin , 1989 ) .
it has a large extent with @xmath16 ( radius within which mean density of the cluster equals 180 times the critical density at the redshift of the cluster ) @xmath17 34.6@xmath18 ( markevitch et al .
1998 ) and hence can be studied with xmm - newton very well .
a3395 has been observed previously in x - rays with the rosat position sensitive proportional counter ( pspc ) by henriksen & jones ( 1996 ) .
they reported temperature of 2 to 4 kev for the cluster and calculated the cooling time for the cluster as @xmath19 gyr , which is much larger than the time predicted for the cluster to fully merge .
hence , they suggested that either there are no cooling flows at the centre of the cluster or the cooling flows are disrupted by the merger .
markevitch et al .
( 1998 ) reanalyzed the rosat pspc data and combined it with the analysis of asca observations .
they found some evidence for the presence of shock - heated gas in the cluster and suggested that a collision between the two subclusters may be responsible for heating the gas .
they calculated the emission - weighted gas temperature @xmath20 of a3395 ( excluding cooling flow and other contaminating components ) to be @xmath21 kev .
donnelly et al .
( 2001 ) reanalyzed both the rosat pspc and the asca data and reinforced the evidence for a merger in a3395 , as they detected a rise in the gas temperature ( @xmath22 above the average temperature ) in the region between the two main subcluster components . by using newtonian energy considerations donnelly et al . proved that the system of the two subclusters is in a bound state and determined the virial masses of the two main subclusters ( ne and sw ) as @xmath23 and @xmath24 respectively .
tittley & henriksen ( 2001 ) report the presence of a group of lower redshift galaxies ( z=0.048 ) , at @xmath25 northwest from the centre of the galaxy distribution , which is elongated in the nw - se direction .
they found an increase in the x - ray emission from this group and classified this as the third component of x - ray emission viz .
, a3395 nw ( centred at r.a.(j2000 ) @xmath26 , dec.(j2000)@xmath27 ) .
this is in addition to the two main x - ray peaks a3395 ne and a3395 sw , that coincide with subclusters , at slightly higher redshifts of 0.051 and 0.052 respectively . using asca and wide field of view rosat observations
, they also detected a filamentary structure that connects the clusters a3395 and a3391 , shown by the presence of excess x - ray emission in a region between the two clusters and aligned with the distribution of galaxies .
a3395 has also been observed in radio with the molonglo cross telescope at 408 mhz ( large et al .
1981 ) and with the molonglo observatory synthesis telescope ( most ) at 843 mhz ( jones & mcadam 1992 ; burgess & hunstead 2006 ) .
the observations showed a wat inside the cluster .
the cluster was also observed by atca at a frequency of 4.79 ghz ( gregorini et al .
reid ( 2000 ) used atca observations of this cluster at 1348 mhz and 2374 mhz , and detected eight radio sources of which five had optical counterparts .
there are 2 prominent radio sources in the sw subcluster , a wat at the centre ( close to the x - ray peak ) and a head - tail ( ht ) galaxy at the periphery .
we have used the atca data to understand the interaction of the wat radio emission with the icm .
a journal of the xmm - newton and chandra observations of a3395 is given in table [ tab : observation_table ] .
the cluster was observed with xmm - newton on 2007 january 24 ( table [ tab : observation_table ] ) .
the three epic cameras mos1 , mos2 ( turner et al . 2001 ) and pn ( strder et al . 2001 ) were operated in full frame mode with the thin1 filter .
the data have been obtained from the heasarc archives .
the raw mos1 , mos2 and pn images are shown in figure [ fig : raw_mos1_mos2_pn_chandra_raw_images ] . note that the observation has been performed after the loss of mos1 ccd # 6 due to a meteorite hit in 2006 june .
also , during the analysis , mos1 ccd # 5 was found to be in anomalous mode ( because of a strong enhancement of the background at @xmath28 kev ) and hence , has not been used anywhere in the imaging and spectral analysis .
all data analysis has been done using the standard procedures from the science analysis system ( sas ) software version 9.0 .
calibrated photon event files were produced from the raw data using the sas tasks * epchain and emchain * and the latest calibration files .
these files were then filtered for the good time intervals using the sas tasks * mos - filter * and * pn - filter*. good time interval found for mos1 , mos2 and pn ccds are 27.16 ks , 27.28 ks and 22.44 ks respectively ( see table [ tab : observation_table ] ) .
the histograms of lightcurves from all three detectors had very well defined gaussian shapes showing that the data are not much affected by the soft - proton contamination .
also the temporal filtering using the tasks * mos - filter * and * pn - filter * removes the soft proton contamination sufficiently . the residual soft proton contamination left after this step was removed from the data by adding powerlaws to the models ( snowden et al .
2008 ) used in the spectral analyses done in
[ sec : total - spec - analy ] , [ sec : azimuth_spec_analys ] and [ sec : box_thermodynamic_maps ] .
the details have been provided in [ sec : total - spec - analy ] . for removing the quiescent particle induced background and the cosmic background component we have used the blanksky observations , which consist of a superposition of pointed observations that have been processed with sas version 7.1.0 ( carter & read , 2007 ) .
blank - sky event files were obtained for all three detectors by submitting an xmm - newton epic background blank sky products request form with the request of a galactic column in the range 3.5 to @xmath29 @xmath30 .
the blanksky event files from the mos(pn ) detectors were then filtered for flares using an upper threshold of 0.35(0.40 ) counts second@xmath5 and then again using the selection criteria of @xmath31 ( @xmath32 ) and @xmath33 .
the files were then recast to have the same sky coordinates as a3395 .
the resulting event files were used for generating all the background images and background spectra for this paper . the high energy ( e@xmath010 - 12 kev ) count - rates for the mos1 , mos2 and pn detectors from the filtered blanksky images were found to be very close to those from the source images showing that the observations were not affected by flares .
we could not do the local background subtraction as the source fills almost the entire field of view of the detectors and it was impossible to find emission - free regions and therefore , using local background could have led to oversubtraction .
the mos1 and mos2 images were combined together using the sas tasks * merge * and * evselect * to increase the signal - to - noise ratio of the sources and reach fainter flux levels .
only those events that were spread over less than 4 pixels ( i.e. , pattern = 0 - 12 ) were selected for producing the combined image .
point source extraction was done simultaneously on the mos1 , mos2 combined image and the pn detector image using the sas task * edetect_chain * which is a combination of several sas tasks viz . * eboxdetect * ( in local mode ) , * esplinemap * , * eboxdetect * ( in map mode ) , and * emldetect*. when run in local mode * eboxdetect * produced source lists by collecting source counts in the cells of @xmath34 pixels and using a value of 8 for the minimum detection likelihood ( * eboxl_likemin * ) ( i.e. a value of e@xmath35 for the minimum value of the probability of the poissonian random fluctuation of the counts in the detection cell which would result in the observed number of source counts ) .
then * esplinemap * was used to generate background maps by using the source list generated by * eboxdetect * ( in local mode ) to remove point sources from the original merged image .
then * eboxdetect * ( in map mode ) uses these background maps to generate a new point source list ( with fainter sources included ) .
a total of 258 point sources were thus detected .
each of these detected sources was checked for each of the detectors and spurious sources ( sources that did not look like actual sources in the individual detector images ) were removed .
finally 77 sources for mos1 , 67 for mos2 , and 44 sources for pn were detected .
of these , 22 sources were found to be common in mos1 and mos2 and 13 were common in all three detectors .
these detected sources were then removed from individual mos1 , mos2 and pn images to create the cheesed images .
the cheesed mos1 and mos2 images were combined using the procedure described in snowden et al .
a contour map of diffuse x - ray emission from a3395 using the combined mos1 and mos2 image after the removal of point sources and after applying the smoothing function is shown in figure [ fig : combined_mos1_mos2_smoothedimage ] .
contours of the smoothed x - ray emission are also overlaid on the optical image of a3395 from the supercosmos survey in the @xmath36 band and shown in figure [ fig : a3395_supercosmos_optical_image ] .
the optical image shows both the bright central galaxies ( bcgs ) for the ne and sw regions .
the position of the bcg for the ne region coincides with the x - ray emission peak for the ne region whereas the bcg for the sw region is offset from the x - ray emission peak for the sw region by about @xmath37 which is equivalent to @xmath38 kpc at the redshift of the cluster .
a3395 was observed with chandra on 2004 july 11 ( obsid 4944 ) with acis - i detector for 22.2 ks ( table [ tab : observation_table ] ) . the raw chandra acis image
is shown in fig .
[ fig : raw_mos1_mos2_pn_chandra_raw_images ] .
the data were analyzed with the ciao version 4.3 and caldb version 4.4.0 .
no reprocessing of data and time - dependent corrections were required as the ascdsver ( keyword that stores the processing version information ) is ds 7.6.7.2 and time - dependent corrections have become a part of standard data processing after ds 7.3.0 .
the standard charge transfer inefficiency ( cti ) corrections have been applied .
point sources were detected using the ciao task * wavdetect * with the detection threshold fixed to the default value @xmath39 .
a total of 47 bright sources were detected ; 16 of these sources were common with the sources detected in the mos 2 detector image , discussed in the previous section .
the smoothed , point - source - removed and exposure - corrected chandra image of the cluster in the 0.3 - 7.0 kev band is shown in figure [ fig : chandra_smoothedimage ] .
a3395 was observed by a.d .
reid and r.w .
hunstead with atca on 1995 january 09 .
the pointing centre was 06@xmath40 26@xmath41 58.0@xmath42 , @xmath4354@xmath44 31@xmath18 12.0@xmath6 ( j2000 ) and centre frequencies were 1348 and 2374 mhz with bandwidths of 128 mhz .
the array was in the 6a configuration , with baselines ranging from 337 to 5939 metres .
the total integration time was 5.6 hours , consisting of multiple cuts spread over a wide range of hour angles .
primary flux density calibrator was b1934@xmath43638 and the phase calibrator was b0647@xmath43475 .
data reduction was carried out in miriad ( see sault et al .
1995 ) using standard techniques .
the restored beams were [email protected] arcsec@xmath46 along [email protected]@xmath47 at 1348 mhz , and [email protected] arcsec@xmath46 pa @xmath4332.5@xmath47 at 2374 mhz .
the rms noise is 1.4 mjy beam@xmath5 in the 1348 mhz image and 1.3 mjy beam@xmath5 in the 2374 mhz image .
a contour map of the 1348 mhz atca radio emission has been overlaid on the image from the supercosmos survey in the @xmath36 band and is shown in fig .
[ fig : a3395_supercosmos_optical_image ] .
figure [ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] shows the unsharp - masked image produced from the combined mos1 , mos2 detector image , by subtracting a large scale ( 100@xmath6 ) smoothed image from a small scale ( 15@xmath6 ) smoothed image .
five distinct regions can be seen : the a3395 ne , a3395 sw , a3395 w , a3395 nw and a filament joining the ne region to the w region . the w region is a small and relatively faint subclump of diffuse emission visible towards the west of the sw region . only two main x - ray peaks ne and sw , could be seen in the earlier x - ray images and are believed to be subclusters . figs .
[ fig : combined_mos1_mos2_smoothedimage ] and [ fig : chandra_smoothedimage ] show a strong gradient in the surface - brightness in the southeast part of the sw component evident from the compression of x - ray contours in this part , indicating that the sw component is moving along the southeast direction .
it appears that the small clump in the w region may also be a subcluster participating in the merger process . the chandra image of the cluster ( fig .
[ fig : chandra_smoothedimage ] ) does not fully cover the ne and w regions .
none of the three regions in a3395 cluster show a circular symmetry .
therefore , for the analyses that follow , we have made elliptical approximations for them .
the ne , sw , and w regions have been approximated as ellipses with semi - major - axis lengths 432.0@xmath6 , 280.8@xmath6 , and 194.4@xmath6 , ellipticities 0.87 , 0.81 , and 0.78 , and major - axis position angles @xmath48 , @xmath49 , and @xmath50 respectively , measured from the north in an anti - clockwise direction ( shown with the green ellipses in fig .
[ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] ) .
a3395 nw can be seen as a region of weak x - ray emission towards the north and northwest of the cluster in fig.[fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] .
this excess emission was first reported by tittley & henriksen ( 2001 ) .
the optical image of a3395 nw shows a clustering of galaxies , however , its x - ray emission is very diffuse and does not seem to clump .
the mean velocity of the group of galaxies associated with the nw , from the velocity data given in teague , carter , & gray ( 1990 ) is @xmath51 km s@xmath5 , while the same calculated from the velocity data given in donnelly et al .
( 2001 ) is @xmath52 km s@xmath5 . the data by donnelly et al .
put this group at almost the redshift of the ne and sw subclusters .
considering that a3395 nw lies along the filament joining the clusters a3395 and a3391 ( tittley & henriksen 2001 ) , it appears to be a part of the supercluster network .
merging clusters like a3395 can host both shock and cold fronts which can be confirmed by surface - brightness and temperature discontinuities .
the discontinuities are seen by a change in the slope of the profile . in order to search for these features , surface - brightness profiles for the ne , sw , and w regions were made .
we have used the raw mos1 , mos2 , pn detector images ( without any smoothing applied but with the point sources removed by applying the cheese masks generated by the @xmath53cheese@xmath54 task ) and the point - source - removed raw chandra acis detector image to derive the surface - brightness profiles .
we used 25 annuli in the ne , 10 annuli in the sw , and 9 annuli in the w regions centred at the intensity peaks in each of the ne , sw , and w regions , which were further divided into 12 sectors of @xmath55 each .
the ellipses are such that the @xmath56 ellipse for the ne region has a semi - major axis of @xmath57 , the sw region has a semi - major axis of @xmath58 , and the w region has a semi - major axis of @xmath59 . the major - axis position angles and ellipticities for the ellipses in ne , sw , and w regions were the same as given in section [ x - ray_morphology ] and the @xmath60 sectors have been arranged symmetrically about the major - axes . to analyze the surface - brightness in the region between ne and sw subclusters , ( which is not covered by the green ellipses in fig .
[ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] ) and also to avoid overlaps between the three subclusters , the area spanned by the ellipses in the ne subcluster is larger , whereas that for the ellipses in the sw and w subclusters is almost same as compared to the green ellipses in fig .
[ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] . in the sw region ,
certain sectors had holes ( due to point source removal ) at the centre . also , in the ne region , a few outermost annuli in the sectors with position angles from @xmath61 to @xmath62 had no mos1 data ( because of the missing ccd6 in mos1 ) , and with position angles from @xmath63 to @xmath64 had no chandra data .
in addition to this , the raw images had spurious discontinuities at the positions where the individual ccds in a detector overlap .
points from all affected annular sectors were removed from the plots , but the resulting gaps made it more difficult to identify discontinuities .
surface - brightness profiles for all four detectors for each and every sector were obtained and were fitted by using single or multiple ( for profiles showing more than one discontinuity ) power laws . in order to make confident detections of discontinuities
, we focused only on those discontinuities which are seen in all four detectors .
all sectors in the ne and sw regions , in general , show a fairly uniform decreasing profile , except for a few sectors that show a change in the slope .
the surface - brightness profiles of all sectors in the w region show a nearly constant profile .
only those surface - brightness profiles that show evidence for a significant change in slope are shown in figure [ fig : a3395_surface_brightness_profiles ] .
+ the @xmath61-@xmath65 sector ( figure [ fig : ne_surface_brightness_profiles_sect5 ] ) in the ne region shows a discontinuity at semi - major axis of @xmath66 .
the sectors @xmath64-@xmath67 ( figure [ fig : ne_surface_brightness_profiles_sect10 ] ) , and @xmath67-@xmath68 ( figure [ fig : ne_surface_brightness_profiles_sect11 ] ) , in the ne region , show a flattening of the slope in all four detectors at a semi - major axis of @xmath69 . starting from the centre of the ne component , this discontinuity is found at @xmath70 times the distance between the centres of the ne and sw regions ( @xmath71 )
. the observed flatness or increase in the surface - brightness in a region between the ne and sw subclusters is found to be due to the co - addition of the emission from the two subclusters .
this result was confirmed by analyzing the sum of the @xmath72-models fitted to the surface brightness profiles of the @xmath73 sector in the ne ( away from sw ) and the @xmath74 sector in sw ( away from ne ) , which showed a similar flatness in the same region between the nw and sw subclusters .
a weak discontinuity in the sector @xmath67-@xmath68 is also seen at a semi - major axis of @xmath75 .
all the sectors in the sw subcluster show a uniformly decreasing profiles .
the surface - brightness profiles in all the sectors of a3395 w show a nearly constant profile .
we also obtained the total surface brightness profiles of a3395 w over the full @xmath76 range by using 5 annuli centred at the emission peak of a3395 w and the filament by using 8 rectangular regions parallel to the length of the filament and placed almost symmetrically about it .
the surface brightness profile of the w subcluster shows a decreasing profile from the innermost to the outermost annulus with a peak at the centre ( figure [ fig : w_surface_brightness_profiles_full ] ) while that of the filament shows a significant hump above the background at the centre ( figure [ fig : bridge_surface_brightness_profiles_full ] ) .
average spectra of the various components of a3395 were extracted from both the xmm - newton and chandra data . for xmm - newton ,
these spectra were extracted from elliptical regions for the ne , sw and w components , and from polygon shaped regions for the nw component and the filament .
the ellipses were centred on the peaks of the surface - brightness .
the ellipse for the ne region had a semi - major axis of 432@xmath6 and an ellipticity of 0.87 , the sw region had a semi - major axis of 281@xmath6 and an ellipticity of 0.81 , and the w region had a semi - major axis of 194@xmath6 and an ellipticity of 0.78 .
for the spectral extraction using chandra data , the regions used for the sw and the filament were the same as those used for xmm - newton , while for the ne and w , maximum possible areas were picked by using a polygon shaped region , since chandra field does not cover these regions fully . as the nw region
was not covered by chandra , its spectral analysis was done using only xmm - newton detectors .
all these regions are shown in figure [ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] .
all spectra were extracted in the energy band 0.5 - 9.0 kev . for mos data , events with @xmath77
were used , whereas for pn data , events with @xmath78 were selected . for xmm - newton detectors , the response matrices and effective areas were generated using the tasks * rmfgen * and * arfgen*. the neutral hydrogen column density along the line of sight to the cluster ( i.e. , along @xmath79 , @xmath80 ) was taken to be @xmath81 @xmath30 based on leiden / argentine / bonn ( lab ) galactic hi survey ( kalberla et al .
2005 ) and redshift was frozen to the value of 0.0498 ( simbad astronomical database ) .
spectral analyses have been performed using the x - ray spectral fitting package * xspec * ( version 12.5.1 ) .
spectra extracted from all detectors were fitted using the * wabs * photoelectric absorption model ( morrison & mccammon 1983 ) and * apec * plasma emission model ( smith et al .
the relative elemental abundances used in * wabs * are as given by anders & ebihara ( 1982 ) . the mos1 , mos2 and pn spectra for each of the regions were fitted simultaneously using three separate * wabs*apec * models .
the values of abundance , temperature and * apec * normalizations for the models were linked together but were not frozen .
the mos1 , mos2 and pn spectra of the weak emission regions viz .
the nw , the w and the filament showed presence of residual soft proton contamination ( shown by prominent residues in the high energy end of the spectra ) and the residual instrumental al k-@xmath82 line at 1.49 kev .
the residual soft proton contamination was modeled by using separate powerlaw models with diagonal rmf files ( snowden et al .
2008 ) and the instrumental al k-@xmath82 line at 1.49 kev was modeled by adding gaussian components separately for mos1 , mos2 and pn ( snowden et al .
the powerlaw indices were found to be negative for most cases and hence were frozen to the value of 0.3 ( i.e. the minimum recommended in snowden et al .
2008 ) , while the powerlaw normalizations were left as free parameters . the centre position and the widths of the gaussian components were frozen to 1.49 kev and 0.02 kev respectively in all the spectra and their normalizations were left as free parameters .
the resulting spectra , along with the histograms of best fit model spectra are shown in figure [ fig : xmm_chandra_ne_sw_w_bridge_average_spectra ] and the best fit values of the temperature , abundance and * apec * normalizations from the xmm - newton and chandra spectral analyses are given in tables [ tab : xmm_a3395_total_regions_spectral_results ] and [ tab : chandra_a3395_total_regions_spectral_results ] respectively .
the confidence contours at the 68.3% , 90% and 99% confidence levels for the free parameters are shown in figure [ fig : xmm_chandra_ne_sw_s_w_chisq_cont ] . from the confidence contours produced from the spectral analysis using xmm - newton
, it can be seen that the values of the temperatures and abundances for the ne and the sw regions ; the ne and w regions and the nw and w regions are different only at a confidence level of 68.3% .
however , the temperature and abundance values for the sw and w regions hardly differ .
the temperature and abundance values for the filament appear to be distinct from the values for all other components except for the w region , which are distinct only at a confidence level of 68.3% .
the best fit parameters from chandra have errors much larger than those from xmm - newton because of fewer counts and hence , all four sets of confidence contours have large overlaps . while the results for all other regions are in agreement with those from xmm - newton ( within errors ) , the results from chandra for the ne region show a temperature slightly higher than that from xmm - newton .
this is possibly because the part of ne which is not covered by chandra has a lower than average temperature , as is indicated by the temperature map produced from xmm - newton data in [ sec : box_thermodynamic_maps ] .
this was also confirmed by using the same polygon shaped region for the extraction of spectrum from the ne with mos1 , mos2 and pn , as was used for chandra spectral analysis , and the results were in mutual agreement .
the background systematic errors ( assuming 10% ) do not affect the results significantly .
+ we also produced azimuthally averaged profiles of temperature , density , entropy , and pressure for the cluster by extracting spectra in elliptical annuli using both xmm - newton and chandra data .
the ne , sw , and w subclusters were divided into 7 , 5 and 4 elliptical annuli respectively ( shown in figure [ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] ) .
the centres of the ellipses were at the peak of the x - ray emission for all three regions .
ellipses in the ne regions had semi - major axis lengths of 84@xmath6 , 132@xmath6 , 184@xmath6 , 240@xmath6 , 300@xmath6 , 364@xmath6 , and 432@xmath6 .
similarly , ellipses in the sw region had semi - major axis lengths of 93.6@xmath6 , 140.4@xmath6 , 187.2@xmath6 , 234@xmath6 , and 280.8@xmath6 and w region had semi - major axis lengths of 64.8@xmath6 , 108@xmath6 , 151.2@xmath6 , and 194.4@xmath6 .
spectra were extracted from all the annuli for all three detectors of xmm - newton .
however , for chandra data , only the innermost 4 annuli in the ne and innermost two annuli in the w region were used as chandra did not cover the ne and w parts completely , while for the sw subcluster , spectra were extracted from all 5 annuli .
all the spectra were extracted in the energy band of 0.5 - 9.0 kev .
both the projected and deprojected profiles of temperature , density , entropy , and pressure were obtained .
the details of spectral analyses performed are the same as given in [ sec : total - spec - analy ] except that the elemental abundances were fixed to the respective average abundance value obtained from the spectral fitting of the ne , sw , and w regions using xmm - newton data i.e. , 0.42 , 0.35 , and 0.3 times the solar value ( @xmath83 ) ( table [ tab : xmm_a3395_total_regions_spectral_results ] . ) respectively , for the annuli belonging to the respective region .
the temperature profile was obtained directly from the spectral analysis and was used to derive the density , entropy , and pressure profiles . to derive the electron density n@xmath84
, we used the * apec * normalization , k@xmath1710@xmath85ei/(4@xmath86[d@xmath87(1+z)]@xmath88 ) ( expressed in units of @xmath89 in cgs system ) , where ei is the emission integral @xmath90dv .
we assume @xmath91 ( henry et al . 2004 ) , which under the assumption of constant density within each ellipsoidal shell gives , ei@xmath92v , where v@xmath17volume of ellipsoidal shell . for the volume estimation of the ellipsoidal shells we have assumed oblate ellipsoids ( i.e. , line of sight axis = major - axis of the ellipse ) .
then the volume of a shell , v@xmath93 , where @xmath94 and @xmath95 are the semi - major and semi - minor axis of the outer shell and @xmath96 and @xmath97 are those of the inner shell . from the commonly adopted definition
, entropy is given by s@xmath98 and electron pressure p@xmath99 ( gitti et al .
if the ions have the same temperature as that of the electrons then the total pressure is twice as large .
we admit that the volume estimates may introduce errors @xmath100 ( finoguenov et al .
2004 , and references therein ; henry et al .
2004 ) , but even these errors are quite small as compared to the errors in temperatures obtained from the deprojection analysis because of poor statistics of the data . for confirmation we also estimated the entropy and pressure for prolate ellipsoidal shells ( i.e. , line of sight axis = minor - axis of the ellipse ) but
the differences were very small ( @xmath100 ) .
projection effects along the line of sight can smooth out the variations in the measured quantities . to correct for this
we have performed the deprojection analysis on the same annular regions shown in fig .
[ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] by using the * xspec projct * model , which estimates the parameters in 3-d space from the 2-d projected spectra of ellipsoidal shells , using the * wabs*apec * model .
the model calculates the geometrical weighting factor according to which the emission is redistributed amongst the projected annuli .
the elemental abundances for the annuli belonging to the ne , sw , and w regions were fixed to 0.42 , 0.35 , and 0.3 times the solar value ( @xmath83 ) , same as in [ sec : projection_analysis ] .
the residual soft proton contamination and instrumental al lines were modeled by adding powerlaws and gaussian components to the models as described in [ sec : total - spec - analy ] with the only difference that a single model for all mos1 , mos2 and pn was used .
this was because the * projct * model requires all the spectra belonging to the same annulus to be part of the same group and therefore have to be modeled using the same model .
electron density n@xmath84 , entropy ( s ) , and electron pressure ( p ) have been calculated using the same relations as given in [ sec : projection_analysis ] .
the resulting projected and deprojected temperature and density profiles are shown in figure [ fig : ne_sw_w_projected_deprojected_temp_dens_profile ] , and the projected and deprojected entropy and pressure profiles are shown in figure [ fig : ne_sw_w_projected_deprojected_entr_press_profile ] .
best fit parameters and the derived values of other dependent parameters are given in tables [ tab : xmm_projected_annuli_spectral_results]- [ tab : chandra_deprojected_annuli_spectral_results ] .
the projected spectral analysis resulted in a nearly constant temperature profile for all three subclusters , while the projected density , entropy , and pressure profiles show a uniform decrease , increase , and decrease respectively for all three subclusters from the innermost to the outermost annulus .
the deprojected temperatures , entropies and pressures of all the three subclusters show a nearly uniform profile .
the deprojected density profiles from all three regions show a uniform decrease except for the outermost annulus where a slight increase in the density is seen .
this is a common artifact of deprojection analysis because the excess emission from shells with radii larger than that of the outermost annulus , gets added to the outermost annulus .
the projected profiles from chandra and xmm - newton are in complete agreement with each other although the former has larger errors as compared to the latter . for the deprojected profiles also , there is a good agreement in the results between xmm - newton and chandra with a few anomalies .
for example , the deprojected density for the innermost annulus in the ne region from chandra is greater while the density in the second annulus in the w region from chandra is lower as compared to those from xmm - newton .
however , the higher deprojected density and pressure from chandra in the fourth annulus ( outermost annulus for chandra ) in ne as compared to those from xmm - newton can easily be explained as the artifact of deprojection described earlier in this section .
we have made projected temperature , abundance , density , entropy , and pressure maps of a3395 , using box shaped regions for spectral analysis to improve the spatial resolution of the spectral parameters and to look for anisotropy in their spatial distribution . to make the results more robust ,
both xmm - newton and chandra data have been used and the whole cluster was divided into 139 boxes for xmm - newton and only 42 boxes for chandra . the number of boxes used for chandra were less because of fewer counts and also because chandra did not cover the outer parts of the cluster significantly due to smaller fov .
the sizes for the boxes were chosen adaptively to get sufficient counts in each region .
for xmm - newton , large size boxes ( @xmath101 or @xmath102 ) for the outermost parts , small size boxes ( @xmath103 ) for the innermost brightest parts , and medium sized boxes ( @xmath104 ) for the regions in between were made to get more than 700 total counts from all three detectors in each box . for chandra , large size boxes ( @xmath105 ) for outer parts and small size boxes ( @xmath106 ) for the inner parts were made to get more than 700 counts in each box .
hence , spatial resolution of the thermodynamic maps obtained with xmm - newton is better than that from chandra .
spectra from all boxes were fitted using * wabs*apec * model with fixed galactic absorption .
the residual soft proton contamination and instrumental al lines were modeled by adding powerlaws and gaussian components to the models as described in [ sec : total - spec - analy ] . for the boxes where
mos1 data could not be used ( i.e. , boxes lying in regions of mos1 missing ccd6 and anomalous ccd5 ) , only mos2 and pn spectra were used .
the electron density , entropy , and electron pressure were calculated using the same relations as in [ sec : projection_analysis ] .
the volume calculation for the box regions , however , was not as straightforward as in the case of ellipsoidal shells .
we had to assume spherical geometry which can be justified because the volumes considered here are very small as compared to those for the ellipsoidal shells .
we assumed the 139 box regions as projections of parts of spherical shells ( centred at the x - ray intensity peak of the subcluster nearest to the box ) with inner and outer radii ( r@xmath107 , r@xmath108 ) equal to the smallest and largest distance from the centre of their respective spheres .
the volume for each box region was estimated as d@xmath109 ( ehlert et al .
2010 ; henry et al .
2004 ) , where @xmath110 is the angular diameter distance and @xmath111 is the solid angle subtended by the region .
@xmath112 and @xmath113 are equal to the distances r@xmath107 and r@xmath108 expressed in angular units respectively .
+ the temperature , density , entropy , and pressure maps produced ( from both xmm - newton and chandra ) are shown in figures [ fig : a3395_box_temp_map ] , [ fig : a3395_box_density_map ] , [ fig : a3395_box_entropy_map ] , and [ fig : a3395_box_pressure_map ] respectively .
maps obtained from chandra had larger errors as compared to those from xmm - newton , and in all the maps fainter regions had larger errors .
the errors range from as low as @xmath015% for the innermost regions of ne and sw subclusters to as high as @xmath060% for the extreme outermost box regions and the highest temperature regions in the nw and towards the southeastern edge of the sw .
abundance values had large errors in the maps obtained from both xmm - newton and chandra and for most of the boxes only upper limit of abundance could be obtained , hence the abundance maps have not been shown .
maps from both xmm - newton and chandra show a high temperature and high entropy region where the ne and sw regions meet .
the sw , on an average shows a temperature slightly higher than the average temperature of the ne and even more higher temperature regions are seen in the w component , the region joining the ne and sw parts and the filament regions .
the ne and sw components have similar average entropies and the w component has slightly higher average entropy .
the filament and the region between the ne and sw components have even higher entropies .
the highest temperatures and entropies are seen at the south - eastern edge of the sw subcluster and in a few regions in the nw component lying above the filament , to the west of the ne subcluster .
although , the single * apec * model provided a good fit for the source spectra of all the regions , for some of the regions ( for e.g. , the region between the ne and sw ) they could also be very well fitted using powerlaw model ( in addition to the fixed index powerlaws used for modeling the residual soft proton contamination ) , thus pointing towards a completely non - thermal emission from shock accelerated particles .
none of the two models seemed to be statistically preferred .
fitting the spectra with * apec+power - law * or two * apec * models resulted in the values of the free parameters being insensitive to the fits .
in addition , the normalizations of the non - thermal ( power - law ) components in the * apec+power - law * fits were found to be negligible as compared to those of the thermal ( apec ) components .
the atca images at 1348 and 2374 mhz show that the most prominent source in the field is the wide - angle tailed ( wat ) source pks b0625@xmath43545 ( mrc b0625@xmath43545 ) which is identified with an elliptical galaxy at a redshift of 0.05174 ( teague , carter , & gray 1990 ) ( figure [ fig : a3395_wat_supercosmos_overlaid_atca_radio_contours ] ) .
both the oppositely - directed jets bend initially towards the west and then again bend sharply further down the flow at @xmath090 kpc from the host galaxy so that the emission is approximately along the north - south axis .
douglass et al .
( 2008 ) have observed similar bends in the wat source in abell 562 and have interpreted the bends further down the flow as the regions where the effects of buoyancy become important .
figure [ fig : a3395_wat_unsharp_mask_chandra_overlaid_atca_radio_contours ] shows the unsharp - masked chandra image of the cluster ( zoomed in to show the wat source ) produced by subtracting a large scale ( 80@xmath6 ) smoothed image from a small scale ( 4@xmath6 ) smoothed image , overlaid with the 1348 mhz radio continuum contours .
it may be worth noting that there is no evidence of x - ray cavities associated with the regions of radio emission .
the flux densities of the wat galaxy at a number of frequencies from low - resolution observations are listed in table [ tab : radio_flux ] , including the values listed by burgess & hunstead ( 2006 ) , and estimates from the atca images presented in this paper .
the flux density estimated from the atca image at 1348 mhz is very close to the measurement at 1410 mhz , suggesting that the interferometric observations at this frequency have not missed any significant amount of flux density .
however the expected flux density at 2374 mhz from the derived spectrum using the high - frequency points ( @xmath114800 mhz ) is higher than our measured value by @xmath100 . although the high - frequency points are consistent with a straight spectrum with a spectral index of @xmath01.05 , a good fit to the spectrum is obtained for the jaffe & perola ( 1973 ) model using the synage package ( murgia et al .
1999 ) . excluding the atca estimate at 2374 mhz where some flux density is missing ,
the spectrum is well fitted with an injection spectral index of 0.62 and a break frequency of @xmath013.9 ghz ( figure [ a3395_radio_spectrum ] ) .
higher frequency observations would be useful to determine this more reliably .
besides the wat galaxy , there is a head - tail ( ht ) source located at ra(j2000 ) : @xmath115 and dec.(j2000 ) : @xmath116 which is associated with a galaxy at a redshift of 0.05955 ( teague , carter , & gray 1990 ) and is shown in figure [ fig : a3395_supercosmos_optical_image ] .
the extent of the ht galaxy is smaller than the wat , with an angular size of @xmath117 including the diffuse emission , which corresponds to a linear size of @xmath090 kpc at the redshift of the galaxy .
the peak and total flux densities of the ht source at 1348 mhz are 25 mjy beam@xmath5 and @xmath0150 mjy .
the tail is not as well imaged at 2374 mhz where the peak and total flux densities are 8 mjy beam@xmath5 and @xmath070 mjy respectively .
most observations show that the peak and total flux densities of the ht source are 135 mjy beam@xmath5 and 299 mjy respectively at 843 mhz ( reid 2000 ) .
the ht galaxy is located at the periphery of the w region of a3395 .
ht galaxies are often detected in non - relaxed cluster but can also be found in the periphery of clusters and near regions of enhanced x - ray emission ( cf .
klamer , subrahmanyan & hunstead 2004 ; mao et al .
2009 , and references therein ) .
in addition , reid ( 2000 ) lists two radio sources towards the ne region , namely j0627@xmath435425 ( ra ( j2000 ) : @xmath118 , dec.(j2000 ) : @xmath119 ) identified with a dumbbell system at a redshift of 0.04512 ( teague , carter & gray 1990 ) , and j0627@xmath435426 ( ra ( j2000 ) : @xmath120 and dec ( j2000 ) : @xmath121 ) , associated with a galaxy at a redshift of 0.04348 .
both sources have been shown in figure [ fig : a3395_supercosmos_optical_image ] .
the peak and total flux densities estimated from the atca image at 1348 mhz are 14 mjy beam@xmath5 and 46 mjy respectively for j0627@xmath435425 .
the corresponding values for j0627@xmath435426 are 6.9 mjy beam@xmath5 and 17 mjy respectively .
the redshifts of these sources are much below the average redshifts ( z=@xmath122 ) of the main components of a3395 , hence these sources are most probably not cluster members .
x - ray luminosities for the ne , sw , w , nw and the filament regions using xmm - newton and for the ne , sw , and w and the filament regions using chandra in the energy range 0.5 - 9.0 kev were estimated from the flux values obtained from the spectral analysis of these regions described in [ sec : total - spec - analy ] . the fluxes ( @xmath123 ) were estimated by convolving the model used in [ sec : total - spec - analy ] with the * xspec * convolution model , * cflux * after freezing the * apec * normalization .
the x - ray luminosities ( @xmath124 ) were then obtained from the fluxes using the formula : @xmath125 where @xmath126 is the luminosity distance to the source . the luminosities ( l@xmath127 ) , derived from the flux values obtained from the spectral analysis done using
xmm - newton and chandra data are given in tables [ tab : xmm_a3395_total_regions_spectral_results ] and [ tab : chandra_a3395_total_regions_spectral_results ] respectively .
we also obtained the bolometric luminosities of the ne , sw , w and nw components .
for this purpose , we fitted the @xmath76 surface brightness profiles of the ne , sw and w subclusters using @xmath72-profiles and the average count rates were obtained from the same regions as in [ sec : total - spec - analy ] . for the nw component
the average count rate was obtained from an elliptical region ( an approximation to the polygon shaped region used for the nw in section [ sec : total - spec - analy ] ) with semi - major and semi - minor axis lengths of @xmath128 and @xmath129 respectively . by using the heasarc tool
* web portable , interactive , multi - mission simulator ( webpimms ) * , these count rates were converted to fluxes in the energy range of 0.01 - 100 kev from which bolometric x - ray luminosities were obtained using the relation given above in this section .
the results obtained from the @xmath72-model fitting along with the estimated x - ray bolometric luminosities of the ne , sw , w and nw components have been given in table [ tab : bol_xray_luminosity ] .
xue and wu ( 2000 ) obtained the l@xmath130-kt relations based on 274 clusters and 66 groups as @xmath131 and @xmath132 for clusters and groups of galaxies respectively , where l@xmath133 is in units of 10@xmath134 ergs s@xmath5 and t is in units of kev . on comparing the bolometric x - ray luminosities and temperatures of the ne , sw , w and nw components of a3395 with those obtained from these relations , we find that the ne , sw and w subclusters have bolometric x - ray luminosities within a factor of 1.5 - 2 of each other , and are close to the l@xmath130-kt relation for the rich clusters obtained by xue and wu ( 2000 )
. the temperatures of these regions are almost twice those of the rich clusters of similar luminosities .
however , the nw component is sub - luminous by about an order of magnitude , as compared to the rich clusters of similar temperatures and is hotter as compared to the isolated groups of similar luminosities . from these observations , it seems highly probable that all the components of the cluster have been heated up due to the ongoing merger processes .
the projected gas densities obtained from [ sec : projection_analysis ] for different annuli in the ne , sw , and w regions were fitted using a @xmath72-model i.e. , @xmath135 where @xmath136 is the central density and @xmath137 is the core radius .
the gas mass @xmath138 out to radii 0.5 mpc and 1 mpc for the ne , sw , and w regions were obtained by using the following formula ( see donnelly et al .
2001 ) : @xmath139 ^{(3/2)\rm \beta } ds\ ] ] where @xmath140 ; @xmath141 is the mass of a proton , and @xmath142= 0.609 is the average molecular weight for a fully ionized gas ( gu et al .
the values of @xmath143 , @xmath137 , @xmath144 , and @xmath145 based on fitting the density profiles with the above model are listed in table [ tab : gas_mass_estimate ] .
the deprojected densities have not been used for determining the gas masses as they had very poor @xmath72-model fits , even after ignoring the outermost annuli densities , which were mostly an overestimate .
the results , therefore , had large errors , especially for the w subcluster .
a rough estimate for the gas mass of the nw component was also made by assuming a constant density in an oblate ellipsoid made from the ellipse used for the nw region in
[ sec : luminosity_estimates ] .
the density was derived from the * apec * normalization obtained from the spectral analysis of nw region in [ sec : total - spec - analy ] using the relation given in
[ sec : projection_analysis ] and the gas mass was estimated to be @xmath146 m@xmath147 . using the central gas temperatures ( @xmath148 ) and densities ( n ) derived from the deprojection analysis in [ sec : deprojection_analysis ] using xmm - newton data and by using the equations from sarazin ( 1988 ) , we calculate the cooling times for the three regions ne , sw , and w of a3395 , assuming them to be subclusters , as : @xmath149^{-1 } \left[\frac{\rm t_{\rm g}}{10^{8}k } \right]^{1/2}\ ] ] the cooling times for the ne , sw , and w subclusters are @xmath150y , @xmath151y , and @xmath152y respectively , which are much longer than the hubble time .
this shows that none of the subclusters has any ongoing cooling flow in it .
the present x - ray observations of a3395 show that the cluster morphology is much more complex than the simple bimodal structure reported earlier .
we identify four distinct regions of strong diffuse x - ray emission , namely , the ne , sw , w , and the filament connecting the w to the ne part of the cluster .
in addition , a fifth component a3395 nw is seen as a weak excess emission in the northwest of the cluster , which was also detected by tittley & henriksen ( 2001 ) and is most probably a part of the supercluster filament connecting the clusters a3395 and a3391 . because of a larger fov , better sensitivity , and larger energy band coverage of the xmm - newton , we have been able to better constrain the values of the x - ray temperature and luminosities for the ne and sw subclusters , as compared to their values from earlier observations that used rosat and asca data .
the temperatures determined by us for the ne ( @xmath153 kev ) and sw ( @xmath154 kev ) subclusters are very close ( within uncertainties ) to the values reported earlier by markevitch et al .
( 1998 ) ( @[email protected] kev for the ne and @xmath156 kev for the sw subcluster ) based on asca data .
we have estimated the x - ray luminosities of all the components of the cluster in the energy range of 0.5 - 9.0 kev and the values are given in tables [ tab : xmm_a3395_total_regions_spectral_results ] and [ tab : chandra_a3395_total_regions_spectral_results ] .
a3395 ne is clearly the dominant subcluster with the highest x - ray luminosity and extent of emission .
our estimates of the bolometric x - ray luminosity for the ne subcluster is equal to , while that for the sw subcluster is slightly higher than , those from donnelly et al .
( 2001 ) , who estimated the bolometric x - ray luminosity within 1 mpc of the ne and sw subcluster cores based on the surface brightness profiles using rosat data .
in addition to the previously identified sw subcluster , there are strong indications for a separate clump w in the west that is either a clump of gas stripped - off the sw subcluster or is probably a separate subcluster merging with the a3395 sw .
this w region is particularly delineated in the unsharp - masked mos2 image shown in fig .
[ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] . note that , in contrast to donnelly et al .
( 2001 ) , our estimate of the luminosity for the sw subcluster does not include the w region .
we now examine whether the w clump of x - ray emission represents a smaller group of galaxies in the process of merging in a3395 .
a3395 w appears to be a distinct region of x - ray emission ( figs .
[ fig : combined_mos1_mos2_smoothedimage ] to [ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] ) .
the detection of an ht source at the periphery of the w subcluster , which is indicative of enhanced x - ray emission and a deep potential well , points towards a distinct atmosphere of the w subcluster .
the surface brightness profile of the w subcluster ( fig .
[ fig : w_surface_brightness_profiles_full ] ) which shows a peak at the centre , also strengthens the idea of a distinct identity of the w subcluster .
we have also re - examined the galaxy velocity sample for a3395 from donnelly et al .
2001 , which had velocity information for 157 cluster member galaxies to look for dynamical evidence for a3395 w being a separate group .
we divided the galaxies into the three subclusters by considering ellipses with major axes 0.4 , 0.3 , and 0.2 times @xmath157 ( = 34.6@xmath18 ) for the ne , sw , and w subclusters respectively and counting the number of galaxies in these regions .
the major - axis position angles and ellipticities for the ellipses in ne , sw , and w regions were the same as given in section [ x - ray_morphology ] .
this led to 71 galaxies in the ne , 32 galaxies in the sw , and 15 galaxies in the w subcluster , leaving 69 galaxies which could not be clearly identified as belonging to any particular subcluster .
also , there were some slight overlaps in the three sets of galaxies .
for example , ne and sw subclusters had 13 galaxies in common ; ne and w subclusters had 9 galaxies in common , and sw and w subclusters had 12 galaxies in common .
the velocity histograms obtained for the four sets are shown in figure [ fig : a3395_gala_vel_histogram ] , where the bin - size was chosen to be 300 km @xmath158 wide .
donnelly et al .
( 2001 ) had reported a high velocity hump in a3395 . from fig . [
fig : a3395_gala_vel_histogram ] , we observe that the galaxies in this hump belong to either the sw or the w parts or to none of the subclusters .
the average subcluster velocities and their dispersions were obtained by fitting gaussians to their velocity distributions and are given in table [ tab : a3395_gala_vel_distbn_results ] .
the velocity data in the w region shows a very flat and wide distribution and is not sufficient for fitting a gaussian .
in addition , the galaxies in the w subcluster seem to follow the velocity distribution of the sw subcluster . though the velocity distribution of the w subcluster had a very poorly fitted gaussian ,
we have force - fitted a gaussian to it for getting an estimate for the velocity dispersion and a lower limit of the virial mass used in the bound system analysis in
[ sec : bound_sys_anlys ] .
we estimated the virial mass for each of the subclusters assuming that the galaxies included in each subcluster are bound and the velocity dispersions are isotropic .
we used the following formula ( see beers , geller , & huchra ( 1982 ) ) : @xmath159 where @xmath160 is the velocity dispersion along the line of sight and @xmath161 is the harmonic mean projected separation between galaxy pairs .
the mean velocity ( @xmath162 ) , velocity dispersion ( @xmath163 ) , and the virial masses of the three subclusters thus estimated , are given in table [ tab : a3395_gala_vel_distbn_results ] .
compared with donnelly et al .
( 2001 ) , our estimate of the virial mass for the ne subcluster ( @xmath164 ) is slightly larger , while that for the sw subcluster ( @xmath165 ) is almost equal ( see [ sec : a3395 ] ) within uncertainties . a possible reason for a higher mass estimate for the ne region than donnelly et al .
is that our area for including subcluster member galaxies in the ne region ( @xmath166mpc@xmath88 ) is @xmath025% larger than the area used by them ( @xmath167mpc@xmath88 ) , which leads to a larger velocity dispersion and harmonic mean distance between galaxy pairs .
application of wavelet transform techniques to the positions of galaxies in a3395 by flin & krywult ( 2006 ) led to the detection of two subclusters which most probably correspond to the bimodal structure of the cluster at a scale of @xmath0527 kpc .
it should be noted that if the w component is indeed a possible subcluster then its length scale is only @xmath0250 kpc . however , considering the velocity distribution of galaxies in the w region , the w component is most probably either a clump that has been stripped - off the sw subcluster by the ram pressure or a subcluster that has merged and relaxed into the sw subcluster ( also see [ sec : disc_evid_fo_merg ] ) . in this section
, we have tested whether the ne , sw , and w subclusters ( pairwise ) make a bound system or not . from simple newtonian energy considerations for a bound system (
see donnelly et al .
2001 ) : + @xmath168 where @xmath169 is the relative radial velocity between the two subclusters ( the difference in mean velocities of the two subclusters from table [ tab : a3395_bnd_system_analysis ] ) , @xmath170 is the projected separation between the centres of the two subclusters , m is the sum of the masses of the two subclusters ( we used the lower limit of the sum of the two masses ) , and @xmath82 is the projection angle from the plane of the sky .
the values of @xmath169 and @xmath170 for the ne - sw , sw - w , and ne - w subcluster pairs are given in table [ tab : a3395_bnd_system_analysis ] .
plots are shown in figure [ fig : a3395_bnd_system_analysis ] .
the hyperbolic curve represents the quantity ( 2gm@xmath171)@xmath172 while the horizontal dashed line represents v@xmath173 ( with its 68% confidence region shown with the cross hatching ) .
all orbit solutions below the hyperbolic curve are bound while those above it are unbound .
thus , all three pairs of subclusters are most probably bound systems . amongst the three regions ,
the ne region is the most luminous and most massive ( tables [ tab : xmm_a3395_total_regions_spectral_results ] , [ tab : chandra_a3395_total_regions_spectral_results ] , [ tab : gas_mass_estimate ] and [ tab : a3395_gala_vel_distbn_results ] ) .
the region a3395 w is connected to a3395 ne via a filamentary structure which is at an average temperature of @xmath174 kev .
the 2-d thermodynamic maps show a high temperature of @xmath175 kev in a region between the ne and sw components along with a high entropy .
as the surface brightness profiles of the ne and sw subclusters did not show any abrupt discontinuity , direct evidence for any shock heating could not be obtained .
possibly , the high temperature is not due to shock heating but due to viscous dissipation , which has also been invoked as the possible reason for heating in the merging binary cluster a115 by gutierrez & krawczynski ( 2005 ) .
the temperature map also shows high temperature regions in the w component .
a part of the nw component just above the filament shows the highest temperature and entropy in the 2-d thermodynamic maps .
also , from [ sec : luminosity_estimates ] , we find indications of a possible heating of all the components of the cluster due to the ongoing mergers .
taking into account the facts that non - detection of the shock front can also be due to projection effects and that some of the high temperature regions could also be fitted by purely non - thermal models , the possibility of x - ray emission in these regions being purely non - thermal from shock accelerated particles can not be ignored .
two different merging scenarios for the cluster a3395 are possible .
if the w region is indeed a separate subcluster , it is possible that the cluster is going through its first merger and sw and w subclusters are most probably falling and merging with the more massive and luminous ne subcluster .
however , under this assumption , the origin of the filament joining the w to the ne subcluster can not be explained .
another possibility is that the two main subclusters ne and sw have already gone through their first merger and the filament and the w region are likely results of two different phases of ram - pressure stripping from the sw subcluster .
the observed strong gradient in the surface - brightness profile of the sw subcluster along the southeast direction is possibly due to an orbital motion of the sw subcluster around the ne subcluster . under this assumption ,
the outer hot and diffuse layers of the sw subcluster were perhaps stripped off by the cooler and denser gas of the ne subcluster during the first phase of ram - pressure stripping , resulting in the formation of the filament and the w region was stripped off in a later ( more recent ) phase of ram - pressure stripping where a relatively colder clump of the sw subcluster was stripped off the main subcluster by the hotter surrounding gas .
randall et al . (
2008 ) have analyzed a similar merging scenario of the m86 galaxy with the icm of the virgo cluster in which a hotter tail and a colder plume were formed in two different phases of ram - pressure stripping .
+ clusters hosting a wat with bent lobes are known to be the sites of ongoing mergers .
thus , the bent lobes of the wat source in a3395 sw are consistent with its being a merging cluster .
the wat sources in merging clusters have been known to show some additional interesting characteristics , viz . , the offset of the x - ray centroid from the position of the bright central galaxy hosting the wat ( sakelliou & merrifield , 2000 ) . in a3395
sw , we have found an offset of about @xmath176 kpc ( @xmath016.5@xmath6 ) between the wat hosting bcg and the x - ray centroid .
merging clusters hosting a wat also show an elongation of the icm distribution along the line that bisects the wat ( gmez et al .
1997 ) , which we do nt find in the icm distribution of the sw subcluster .
following these observations we propose a scenario where the subclusters sw and w are falling into and merging with the subcluster a3395 ne , although the actual merger geometries seem to be much more complex .
the radio properties of the wat radio galaxy have been well determined over a large frequency range .
we estimate the equipartition magnetic field and the minimum energy densities in the northern and southern peaks of emission which are closest to the core of the wat radio galaxy , by assuming a value of unity for the particle to electron energy density ratio and also for the filling factor of the relativistic plasma .
the formulae used to calculate the equipartition magnetic field ( b(u@xmath177 ) and the minimum energy densities ( u@xmath178 ) are ( see moffet 1975 , p. 211 ) : @xmath179 and @xmath180 respectively , where , @xmath181 @xmath182 @xmath183 and @xmath184 in cgs units , l is the total luminosity calculated from the total flux s , ( obtained by integrating the flux density ( @xmath185 ) from @xmath186mhz to @xmath187mhz ) , @xmath82 is the radio spectral index , v is the volume of the source region used , a is the ratio of the total particle energy to the energy in the electrons ( assumed to be 1 ) , and @xmath126 is the luminosity distance to the source .
the spectral index of the northern emission peak of the wat is @xmath00.7 and its deconvolved size has been estimated to be @xmath016@xmath458 arcsec@xmath46 from two - dimensional gaussian fits using the aips task jmfit . assuming a cylindrical geometry , the equipartition magnetic field is @xmath020 @xmath142 g and the minimum energy density is @xmath03.8@xmath4510@xmath188 erg @xmath3 , implying a pressure of @xmath01.3@xmath4510@xmath188 dynes @xmath30 . for the southern feature , @xmath82 is again @xmath00.7 , and for a deconvolved size of @xmath026@xmath4512 arcsec@xmath46 ,
the equipartition magnetic field is 14@xmath142 g and the minimum energy density is @xmath01.9@xmath4510@xmath188 erg @xmath3 , implying a pressure of @xmath00.6@xmath4510@xmath188 dynes @xmath30 .
the deprojected pressure near the centre of the sw subcluster is @xmath04@xmath4510@xmath188 dynes @xmath30 .
although an estimate of the volume will also be affected by projection effects , a change in the volume by @xmath020% will affect the pressure by only @xmath08% .
although the x - ray pressure appears somewhat larger than the internal pressure of the emission peaks of radio - emitting plasma , a ratio of particle to electron energy of 50 will increase the energy density and pressure by a factor of @xmath09.4 .
if this is the case the northern emission peak would be overpressured , while the southern emission peak would approximately be in pressure equilibrium with its environment .
the pressure of the emission - peaks would also increase if the radiating particles extend to lower energies and hence lower frequencies than have been assumed here .
we also estimate an average value of the equipartition magnetic field and pressure for the whole source with similar assumptions .
the total length of the wat radio source along its ridge line is @xmath189 , and the average deconvolved width is @xmath190 , although this varies significantly along the axis of the source . here , the radio spectrum has been integrated from 10 mhz to 1 ghz using the injection spectral index of 0.62 , and from 1 to 100 ghz using an estimated value of 1.05 for the high - frequency spectral index resulting from the steepening of spectra due to aging .
a cylindrical geometry has been assumed .
this yields an equipartition magnetic field of @xmath06@xmath142 g , which indicates a minimum energy density of 3@xmath4510@xmath191 erg @xmath3 and a minimum pressure of @xmath010@xmath191 dynes @xmath30 .
again , a contribution from heavier particles and/or integration to lower energies appears necessary to achieve pressure balance with the external environment where the deprojected pressure drops to @xmath010@xmath188 dynes @xmath30 at a distance of @xmath192 . as suggested by osullivan et al .
( 2010 ) , a possible source of additional pressure could be provided by the gas entrained and heated by the jets , which are mostly found in fri jets ( worrall 2009 ) . for a field strength of 5.7 @xmath142 g , a spectral break at @xmath013.9 ghz as suggested by synage ( murgia et al .
1999 ) fit for the jaffe & perola ( 1973 ) model , the spectral age of the wat is @xmath010 myr .
there appear to be no cavities in the x - ray maps near the location of the wat radio source ( figure [ fig : a3395_wat_unsharp_mask_chandra_overlaid_atca_radio_contours ] ) .
this is either because the observations are not deep enough to detect the cavities or because the surrounding hot thermal plasma has had time to leak into and fill the cavities .
the properties of the ht source are less well determined compared with the wat .
such ht sources are produced by galaxies moving at a high relative velocity into a high density of the icm .
the integrated flux densities discussed earlier suggest a rather steep spectral index of @xmath01.4 .
although there are considerable uncertainties in this value , the equipartition magnetic field with the same assumptions as those for the wat is @xmath04 @xmath142 g , while the minimum energy density is @xmath01.6@xmath4510@xmath191 erg @xmath3 , implying a pressure of @xmath00.5@xmath4510@xmath191 dynes @xmath30 .
this is again smaller than the deprojected pressure of the w subcluster where the pressure is @xmath01.5@xmath4510@xmath188 dynes @xmath30 .
the deprojected electron density in the central region of the w subcluster is close to 2@xmath4510@xmath193 @xmath3 , indicating a ram pressure , @xmath194 where @xmath195 is the velocity of the host galaxy relative to the icm , of @xmath03@xmath4510@xmath188 dynes @xmath30 . considering the galaxies in the w subcluster
, the ht host galaxy is moving at @xmath02000 km s@xmath5 relative to the median velocity of this group ( figure [ fig : a3395_gala_vel_histogram ] ) as would be expected for the tails to be bent by ram pressure of the icm .
x - ray observations of the cluster a3395 have revealed five main components in its x - ray morphology : the ne , sw , w subclusters , the nw region and a filament connecting the ne subcluster to the w subcluster .
the surface brightness profiles of the various components of the subcluster do not show any shock fronts in the cluster .
the 2-d thermodynamic maps of the cluster , however , provide evidence for high temperature regions at the interfaces of the various components of the cluster .
the very high temperature and entropy regions in the nw component , which is most probably a part of the supercluster filament connecting the clusters a3395 and a3391 , point to the merging environment in a3395 possibly being affected by a much larger supercluster network .
the x - ray bolometric luminosities of the ne , sw and w components , and the nw component are similar to those of the rich clusters and the isolated groups of galaxies respectively .
however , the temperatures of all the components of a3395 are unusually higher as compared to those of the rich clusters and groups of galaxies and suggest a possible heating of the whole cluster resulting from the ongoing mergers .
none of the ne , sw , and w subclusters shows any cooling flows , as it is very likely that the cooling flows have been disrupted by the mergers in the cluster .
the galaxy velocity distributions of the ne and sw subclusters have well defined gaussian fits and hence are confirmed as well defined subclusters in the process of merging .
although the morphology , the surface brightness profile and the presence of an ht galaxy in the w region show it as a separate subcluster , the galaxy velocity distribution of a3395 w could not be fitted with a gaussian because of very few galaxies in it .
another possibility is that the cluster may have already gone through its first merger and the filament and the w components have been stripped off the sw subcluster in two different phases of ram - pressure stripping . however , if the w component is indeed a separate subcluster , it is possible that the sw and w subclusters are falling into and merging with the more massive and luminous ne subcluster .
we have estimated the equipartition magnetic field and the minimum energy pressure for the wat and the ht source seen in the atca radio images near the centre of the sw subcluster and at the periphery of the w subcluster respectively .
neither source shows any cavities associated with it in the unsharp masked x - ray images , which could be either because the observations are not deep enough to detect the cavities or because the hot thermal plasma has leaked into and filled the cavities .
although the minimum energy pressure of both the radio sources is somewhat less than the external x - ray pressure , pressure equilibrium can be achieved by considering a larger contribution from heavier particles , integration to much lower energies , and additional pressure due to gas entrained and heated by the jets . from the spectral analysis of the source based on various radio observations ,
we have estimated a spectral age of @xmath19myr for the wat source .
deeper and higher resolution x - ray observations are required to properly understand the merging scenario and the geometry of the cluster , to quantify the structure of the shock fronts between the subclusters , and to understand the filamentary region in detail .
deeper radio observations of the ht source are required to better constrain the results from the imaging and spectral analysis of the source .
also , deeper radio observations at low frequencies ( @xmath0300 - 500 mhz ) are required to look for the extended sources of diffuse radio emission such as radio halos which are mostly found in merging clusters .
the galaxy velocity distribution of the w region requires observations of redshifts from a larger sample of galaxies in this region and its surroundings in order to derive the velocity dispersion and mass estimates with better precision , and to confirm the distinct identity of the w subcluster
. pointed observations from xmm - newton in the nw region and also along the filament between the clusters a3391 and a3395 ( discussed in tittley & henriksen 2001 ) can help in a better understanding of the merging environment of a3395 and its connection with the supercluster network .
the data for this research have been obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
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 ) .
we have also made use of data from the chandra x - ray observatory , managed by nasa s marshall center .
we thank the xmm helpdesk for their assistance on xmm - newton data analysis .
data were also obtained from the australia telescope compact array which is a part of the australia telescope national facility , funded by the commonwealth of australia for operation as a national facility managed by csiro .
we would like to thank keith arnaud for his help on various issues related to xspec .
finally , we thank the anonymous referee for his detailed comments and suggestions , which have helped us in improving the analyses and the presentation .
.best fit parameters obtained from the spectral analysis of the full ne , sw , w and nw regions plus the filament connecting the ne and w regions using * xmm - newton * data ( all regions as shown in fig . [ fig : a3395_unsharp_mask_comb_mos1_mos2_and_tot_spec_regions ] ) .
the spectra for each of the regions are fitted with a single - temp * apec * model for a fixed galactic absorption . the residual soft proton contamination and the instrumental al line at 1.49 kev
have been modeled by adding powerlaws and gaussians respectively ( separately for mos1 , mos2 and pn ) to the models .
best fit values for the temperature ( kt ) , elemental abundance relative to the solar values , normalization of the * apec * model , x - ray luminosity ( l@xmath127 ) , and minimum reduced @xmath196 are given along with the degrees of freedom ( dof ) . [ cols="^,^,^,^,^,^ " , ] band with overlaid x - ray contours from figure [ fig : combined_mos1_mos2_smoothedimage ] ( shown with black ) and atca 1348 mhz radio continuum contours ( shown with red ; levels are at 0.005 times 1 , 2 , 4 , 8 , 16 , and 32 jy beam@xmath5).,width=672 ] ) .
the overlaid x - ray emission contours are at 2 , 4 , 6 , 8 , 10 , 12 , 14 , 16 , 18 , 20 , 22 , and 24 times @xmath202 counts @xmath30s@xmath5pixel@xmath5 level above the mean background level ( = @xmath203 counts @xmath30s@xmath5pixel@xmath5).,width=672 ] ) smoothed image from a small scale ( 15@xmath6 ) smoothed image ) showing the four principal regions of emission : ne , sw , w , the filament connecting ne to w region ( marked with an f ) and the nw region ( reported by @xcite ) used for extraction of spectra using xmm - newton mos1 , mos2 and pn detectors in section
[ sec : total - spec - analy ] with green . for extraction of spectra
using chandra data polygon shaped approximations ( yellow ) were made for the ne and w regions as chandra did not cover these parts sufficiently , while regions used for the sw and filament regions were kept same .
all the annuli regions , shown with green plus blue ellipses in all three subclusters were used in the azimuthally averaged spectral analysis using xmm - newton and chandra data in [ sec : projection_analysis ] and [ sec : deprojection_analysis ] . only the innermost 4 annuli in the ne and innermost 2 annuli in the w region
were used for chandra data.,width=576 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in kev units shown in the bars alongside .
details of the spectral fittings are provided in [ sec : box_thermodynamic_maps].,title="fig:",width=288,height=336 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in kev units shown in the bars alongside .
details of the spectral fittings are provided in [ sec : box_thermodynamic_maps].,title="fig:",width=240,height=307 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of @xmath3 shown in the bars alongside .
details of the spectral fittings are provided in [ sec : box_thermodynamic_maps].,title="fig:",width=288,height=336 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of @xmath3 shown in the bars alongside .
details of the spectral fittings are provided in
[ sec : box_thermodynamic_maps].,title="fig:",width=240,height=307 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of kev cm@xmath88 shown in the bars alongside .
details of the spectral fittings are provided in
[ sec : box_thermodynamic_maps].,title="fig:",width=288,height=336 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of kev cm@xmath88 shown in the bars alongside .
details of the spectral fittings are provided in
[ sec : box_thermodynamic_maps].,title="fig:",width=240,height=307 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of erg @xmath3 shown in the bars alongside .
details of the spectral fittings are provided in [ sec : box_thermodynamic_maps].,title="fig:",width=288,height=336 ] and [ fig : chandra_smoothedimage ] respectively .
the scales are expressed in units of erg @xmath3 shown in the bars alongside .
details of the spectral fittings are provided in
[ sec : box_thermodynamic_maps].,title="fig:",width=240,height=307 ] | we present a detailed imaging and spectral analysis of the merging environment of the bimodal cluster a3395 using x - ray and radio observations .
x - ray images of the cluster show five main constituents of diffuse emission : a3395 ne , a3395 sw , a3395 nw , a3395 w , and a filament connecting ne to w. x - ray surface - brightness profiles of the cluster did not show any shock fronts in the cluster .
temperature and entropy maps show high temperature and high entropy regions in the w , the nw , the filament and between the ne and sw subclusters .
the ne , sw and w components have x - ray bolometric luminosities similar to those of rich clusters of galaxies but have relatively higher temperatures .
similarly , the nw component has x - ray bolometric luminosity similar to that of isolated groups but with much higher temperature .
it is , therefore , possible that all the components of the cluster have been heated by the ongoing mergers .
the ne subcluster is the most massive and luminous constituent and other subclusters are found to be gravitationally bound to it .
the w component is most probably either a clump of gas stripped off the sw due to ram pressure or a separate subcluster that has merged or is merging with the sw .
no x - ray cavities are seen associated with the wide angle tailed ( wat ) radio source near the centre of the sw subcluster .
minimum energy pressure in the radio emission - peaks of the wat galaxy is comparable with the external thermal pressure .
the radio spectrum of the wat suggests a spectral age of @xmath010myr . |
there is a strong tension , recently quantified by @xcite , between the value of the hubble constant @xmath0 derived by _
planck _ @xcite from anisotropies in the cosmic microwave background ( cmb ) : @xmath7 km
s@xmath5 mpc@xmath5 , and the value from local measurements : @xmath8 km s@xmath5 mpc@xmath5 @xcite . while the latter measurement is based on local measurements , the former infers a global value for the hubble constant within a cosmological model .
there remains disagreement about the local value of @xmath0 depending on the distance indicator used to measure it , which hints the discrepancy with _
planck _ could be the result of systematic errors .
@xcite calibrated the sne ia distances with three indicators : distance to ngc 4258 based on a megamaser measurement , parallax measurements to milk way cepheids ( mwc ) and cepheids observations and a revised distance to the large magellanic cloud ( lmc ) .
contrarily , calibrating the sne ia with the tip of red - giant branch , @xcite provides @xmath0 @xmath9 63.7 @xmath10 2.3 km s@xmath5 mpc@xmath5 .
this shows how crucial is the first - step calibration in the distance ladder to measure @xmath0 .
however , there are several local @xmath0 measurements with higher values .
@xcite found @xmath0 @xmath9 75.4 @xmath10 2.9 km s@xmath5 mpc@xmath5 by using cepheids in m31 . with a mid - infrared calibration for the cepheids
, @xcite derived @xmath0 @xmath9 74.3 @xmath10 2.1 km s@xmath5 mpc@xmath5 , and with 8 new classical cepheids observed in galaxies hosting sne ia @xcite got @xmath0 @xmath9 76.0 @xmath10 1.9 km s@xmath5 mpc@xmath5 . by using hii regions and hii galaxies as distance indicators , @xcite obtained @xmath0 @xmath9 74.3 @xmath10 3.1(random ) @xmath10 2.9 ( syst . ) km s@xmath5 mpc@xmath5 .
some of these are over 4@xmath11 away from the cmb - derived value .
[ figh0 ] for a plot of different measurements of @xmath0 .
0.1 in a variety of different physical effects could explain such a discrepancy .
it could just be cosmic variance : as we can observe the universe from only one position , we are not able to realize the global parameters from the local parameters , as in the local expansion rate for instance . if we live in an locally underdense region , a `` hubble bubble '' , a higher value for @xmath0 is obtained compared to the global value .
this effect was carefully addressed by @xcite through a modelling of the statistics of matter distribution which provides the distribution of the gravitational potential at the observer .
the outcome is that cosmic variance can alleviate the tension , but a complete elimination requires a very rare fluctuation @xcite .
another way to look at the problem is to consider that the discrepancy may indicate new physics , such as massive neutrinos @xcite , or alternative dark energy models @xcite .
recently , some analyses were performed trying to identify sources of systematic errors in order to remove or alleviate the tension .
for example , by using only the geometric maser distance to ngc 4258 of @xcite as an anchor , @xcite revisited @xcite analysis and derived @xmath12 km s@xmath5 mpc@xmath5 , while combining with lmc and mwc anchors the value is @xmath13 km s@xmath5 mpc@xmath5 , alleviating the tension .
the _ planck _
data were also reanalysed by @xcite , where it was claimed that the 217 ghz @xmath14 217 ghz detector is responsible for some part of the tension .
their new hubble constant without the 217 ghz @xmath14 217 ghz detector is slightly higher : @xmath15 km s@xmath5 mpc@xmath5 . with so many alternatives
, progress can be achieved by developing new ways to address the issue .
we point out here that @xmath1 data which are not calibrated on a @xmath0 estimate can be extrapolated to z=0 to provide an independent measurement of the global @xmath0 . here , the hubble function is reconstructed in order to derive @xmath0 from 19 @xmath1 measurements of passively evolving galaxies as cosmic chronometers @xcite .
many of these are at relatively moderate and high redshifts so intrinsically probe the global value for @xmath0 rather than the local one .
we use gaussian processes ( gp ) , which is a non - parametric method , to obtain the value of the hubble constant in a completely cosmological model - independent way , which is in principle not affected by the local systematics .
we show the value of the hubble constant derived in this way is lower than the standard local measurements .
we obtain @xmath0 @xmath9 @xmath16 km s@xmath5 mpc@xmath5 @xmath6 , in agreement with the cmb - inferred value
. a better understanding of systematic errors , especially the adopted stellar population synthesis model , is required : we show that to improve this result a big effort is necessary to decrease the errors substantially in future , and a focus on precision is worthier than the number of data .
the paper is organized as follows : in sec .
[ meth ] we describe gp as well as standard parametric methods adopted to constrain @xmath0 . in sec . [ const ] the bounds derived for the hubble constant are displayed , followed by forecasts of constraints in sec .
we finish the paper in sec .
[ conc ] with the conclusions .
a gaussian _ distribution _ is a distribution over random variables , while a gaussian _ process _ is a distribution over functions .
this allows one to reconstruct a function from data without assuming a parametrisation for it .
here we use gapp ( gaussian processes in python ) @xcite in order to reconstruct the hubble parameter as a function of the redshift from which we can infer @xmath0 .
this method has been applied for several purposes , for example the reconstruction of the equation of state of dark energy @xcite and to perform null tests of the concordance model @xcite .
the reconstruction is given by a mean function with gaussian error bands , where the function values at different points @xmath2 and @xmath17 are connected through a covariance function @xmath18 ( see @xcite for a discussion of choices of covariance functions ) .
this covariance function depends on a set of hyperparameters . here , as we expect that the hubble parameter and all its derivatives to be smooth , we consider the general purpose squared exponential ( sq . exp . ) covariance function which is given by @xmath19 in the above equation we have two hyperparameters , the first @xmath20 is related to typical changes in the function value while the second @xmath21 is related to the distance one needs to move in input space before the function value changes significantly .
we follow the steps of @xcite and determine the maximum likelihood value for @xmath20 and @xmath21 in order to obtain the value of the function . in this way
, we are able to reconstruct the hubble parameter as a function of the redshift from @xmath1 measurements . we discuss in sect .
[ sect_cov ] the impact of different covariance functions on our results .
in order to compare the results provided by non - parametric methods with standard analyses , we also consider two parametric models .
first of all , we take a flat xcdm model , where the universe is composed by dark matter and a fluid @xmath22 with equation of state @xmath23 , where the hubble parameter is given by @xmath24 where @xmath25 is the matter density parameter today . when @xmath26 this is the concordance @xmath27cdm model which we consider separately . in order to derive @xmath0 for the parametric models ,
we apply standard statistical procedures based on maximum likelihood methods .
we use 19 @xmath1 measurements @xcite from passively evolving galaxies as cosmic chronometers to derive the value of @xmath0 .
figure [ fig1 ] presents the results for the non - parametric approach adopted in this work .
we also plot the @xmath1 measurements with their respective errorbars .
the blue solid line refers to the reconstruction with gapp , with the shaded contours designating the @xmath28 errors .
when extrapolated to redshift @xmath3 , we obtain @xmath29 km s@xmath5 mpc@xmath5 @xmath6 .
note that this value is completely independent of a cosmological model , which makes it complementary but consistent with the _ planck _ value which is derived within the @xmath27cdm model . in this way , our result can also be used to shed light in the whole cosmological model .
this result goes in the direction of a model - independent approach with sne ia which also prefers lower values for @xmath0 @xcite . as a means to compare the non - parametric with standard parametric analyses , a flat @xmath27cdm model and a flat xcdm are also considered .
the hubble constant is found to be @xmath30 km
s@xmath5 mpc@xmath5 @xmath6 for the flat @xmath27cdm model and @xmath31 km s@xmath5 mpc@xmath5 @xmath6 for xcdm .
the bigger error for the xcdm case is derived as a consequence of the inclusion of an extra parameter .
the mean values are higher compared to the non - parametric approach , and in closer agreement with locally measured values , but also in agreement with the non - parametric result .
table [ table1 ] summarizes the constraints for @xmath0 with @xmath28 errors .
all methods prefer values for @xmath0 below 70 km s@xmath5 mpc@xmath5 , in contrast with some local determinations of @xmath0 . 0.1 in .@xmath0 constraints from 19 @xmath1 measurements .
[ cols="^,^,^,^ " , ] some tests were performed in order to evaluate the robustness of the result .
we split our analysis searching for three effects : ( i ) the impact of the covariance function in gapp , ( ii ) a possible presence of outliers driving @xmath0 for lower values and ( iii ) systematic errors from the stellar population synthesis ( sps ) models .
the freedom in the gp approach comes in the covariance function .
while in traditional parametric analyses we choose a model to characterise what is our prior belief about the function in which we are interested , with gp we ascribe in the covariance function our priors about the expected function properties ( e.g. smoothness , correlation scales etc . ) .
since we expect the hubble parameter and its derivatives to be smooth , the squared exponential covariance function was selected which is infinitely differentiable this implies that functions drawn from the process are also infinitely differentiable .
however , we considered other covariance functions to see how the results are affected . in order to do so
, we considered three covariance functions from the matrn family , namely the @xmath32 , @xmath33 and @xmath34 ( see @xcite for definitions and further discussion ) .
writing @xmath35 , each matrn function is @xmath36 times differentiable as are functions drawn from it , and the squared exponential is recovered for @xmath37 . increasing @xmath38 increases the width of the covariance function near the peak implying stronger correlations from nearby points for a fixed correlation length @xmath39 .
the results are shown in table [ table1 ] , where we see slightly higher values are derived for @xmath0 , together with slightly larger errors , for smaller @xmath38 .
in fact , for the matrn@xmath40 the tension with local @xmath0 disappears , although the result remains in fully agreement with _
planck_. interestingly , although there is some shift , the errors are relatively independent of the covariance function choice , especially when compared to the ones derived when one increases the number of parameters in parametric analyses ( the error more than doubles when allowing for @xmath41 to be free , compared to fixing it to @xmath42 ) , showing that gp provide very stable results within different reasonable assumptions . we checked if the high - redshift data were pivoting down the value of @xmath0 to smaller values . to do so , we removed all data points with redshifts greater than 1 , but again the results were completely consistent with the full sample , @xmath43 km s@xmath5 mpc@xmath5 , thus not removing the tension . by removing low - redshift points ,
first and second or third and fourth , again the results did not change significantly , with @xmath44 km s@xmath5 mpc@xmath5 and @xmath45 km s@xmath5 mpc@xmath5 , respectively .
we note that the points with smaller errors dominate the final error budget , as confirmed also by the analysis done in sect .
we also removed point by point in the analysis .
for the first 17 points , the results changed slightly , with a mean value between 64 and 65 and errors between 4 and 5 .
conversely , the high - redshift points showed the biggest departure once removed , with values @xmath46 ( 18th out ) and @xmath47 km s@xmath5 mpc@xmath5 ( 19th out ) .
higher values are derived and the error blows up , although in agreement with the full sample .
this shows how @xmath0 is sensitive to high - redshift values , where more data points in this redshift region might help to mitigate possible sytematic errors due to outliers .
one of the possible main sources of systematic errors in @xmath1 measurements comes from the adopted sps model .
the 19 points used here were derived with @xcite sps model ( bc03 ) . on the other hand , recently @xcite calculated @xmath1 for eight measurements considering bc03 and another sps model from ( * ? ? ?
* mastro ) .
we performed the reconstruction with gapp for this subset with both sps models .
for bc03 we derived @xmath0 in the range @xmath48 km s@xmath5 mpc@xmath5 , in good agreement with the full sample . on the contrary ,
the analysis with mastro provided @xmath49 km s@xmath5 mpc@xmath5 , in disagreement with _ planck _ and in good agreement with the value of @xcite . therefore ,
even with only eight data points , we identify the sps model as the main concern for our results .
another independent measurement for @xmath1 is given by baryon acoustic oscillations ( baos ) .
currently , there are 7 measurements from @xcite , @xcite , @xcite , @xcite and @xcite . combining with the other measurements , for gapp ( sq
we got @xmath50 km s@xmath5 mpc@xmath5 , for a flat @xmath27cdm model @xmath51 km s@xmath5 mpc@xmath5 and for a flat xcdm model @xmath52 , all values consistent with _
planck_. however , there are some drawbacks when using the bao data .
first , these data are not model independent .
they are based on the @xmath27cdm model to study the correlation functions and transform them to the @xmath1 values .
moreover , generally what is inferred is the combination @xmath53 , @xmath54 standing for the sound horizon whose value is given by _
@xcite . also important
is the point raised by @xcite warning that their values of @xmath1 are derived and so they should not be used to test models .
since the current errors do not allow a final decision about the value of @xmath0 , our next step is to study whether future data can settle the issue .
the procedure to analyse how future data can improve the determination of @xmath0 is split in two ways .
first , we consider how the increase of data points of same quality can change the constraints .
second , the errors for @xmath1 are shrunk and a comparison is made between the number of data and their quality . the current errors for @xmath1 measurements grow with redshift a few percent up to around 15 per cent . assuming that future data will provide measurements with the same errors , we update the method of @xcite to predict future data based on the recent measurements from @xcite . a value for @xmath1 is generated by @xmath55 , where @xmath56 and @xmath57 are respectively the simulated and fiducial values for the hubble parameter at redshift @xmath2 , and @xmath58 is a random number gaussianly distributed with mean zero and variance @xmath59 . to estimate @xmath59 , the uncertainties of the observational points
are restricted by two straight lines : @xmath60 and @xmath61 , with two `` outliers '' removed since they were not following the trend of the errors . assuming that the errors of future data will be between the two lines
, one can expect the mean line of the error to be @xmath62 .
therefore , the error of the simulated point is drawn from a gaussianly distributed random variable @xmath63 , where @xmath64 is chosen to assure the error is within @xmath65 and @xmath66 with 95.4% probability . 0.1 in figure [ fig2 ] presents
the expected future errors from considering data are equally spaced in the interval @xmath67 .
we present the expected error on @xmath0 from simulations with 64 , 128 , and 256 data points .
these numbers were chosen because with 64 @xmath1 measurements of same quality as today one can achieve the same constraints given by current sne ia @xcite .
the black crosses represent the errors with the gapp reconstruction , the red dots for the flat xcdm model and the blue triangles for the flat @xmath27cdm model .
the first panel in the left shows the behavior of future data with the same quality as today , and the others show the trend for smaller errors for the @xmath1 data , of 10% , 5% , and 3% ( see @xcite for an observational program to achieve such values ) .
some conclusions can be made from fig .
[ fig2 ] : * for future data with the same quality as today gapp performs very well , with errors smaller than the ones obtained with a flat xcdm model . *
current quality data provide better constraints to @xmath0 than a constant error of 10% in the whole redshift range , showing that lower redshift objects with higher precision compensate the low quality data at high redshifts . * for higher precision measurements gapp and the xcdm model provide the same constraints , showing that a non - parametric approach is powerful to study cosmological data . *
improvement in precision is more important than increasing the number of data of poorer quality .
we have applied gapp , a non - parametric smoothing method based on gaussian processes , to 19 @xmath1 measurements in order to constrain the hubble constant @xmath0 .
this method does not rely on a cosmological model , so its results can be used to infer the impact of systematic errors as well as the underlying cosmological framework .
we have obtained @xmath0 to be @xmath16 km
s@xmath5 mpc@xmath5 @xmath6 , a value which is in agreement with _ planck _ , but in disagreement with local measurements .
this supports the notion that either there are unidentified systematic errors in the local @xmath0 data , or the local value is indeed different from the global value . a better comprehension of systematic errors , especially a thorough analysis of the impact of sps models , can improve the robustness of our results .
simulations have shown that improvements in distant @xmath1 measurements can help pin down the global value of @xmath0 .
the authors are grateful to raul jimenez and martin hendry for useful comments .
vcb is supported by cnpq - brazil through a fellowship within the program science without borders .
ms is supported by the south africa square kilometre array project and the south african national research foundation ( nrf ) . | an intriguing discrepancy emerging in the concordance model of cosmology is the tension between the locally measured value of the hubble rate , and the ` global ' value inferred from the cosmic microwave background ( cmb ) .
this could be due to systematic uncertainties when measuring @xmath0 locally , or it could be that we live in a highly unlikely hubble bubble , or other exotic scenarios .
we point out that the global @xmath0 can be found by extrapolating @xmath1 data points at high-@xmath2 down to @xmath3 . by doing this in a bayesian non - parametric way we can find a model - independent value for @xmath0 .
we apply this to 19 measurements based on differential age of passively evolving galaxies as cosmic chronometers . using gaussian processes
, we find @xmath4 km s@xmath5 mpc@xmath5 @xmath6 , in agreement with the cmb value , but reinforcing the tension with the local value .
an analysis of possible sources of systematic errors shows that the stellar population synthesis model adopted may change the results significantly , being the main concern for subsequent studies .
forecasts for future data show that distant @xmath1 measurements can be a robust method to determine @xmath0 , where a focus in precision and a careful assessment of systematic errors are required .
[ firstpage ] cosmological parameters cosmology : observations cosmology : theory dark energy large - scale structure of universe
cosmology : distance scale |
statistical homogeneity and isotropy of the universe on sufficiently large scales is a crucial assumption which simplifies the study of large scale structure of the universe and cosmology as a whole .
this assumption which is known as the cosmological principle is one of the fundamental pillars of modern cosmology .
one can not derive this principle in a strictly mathematical sense and can only verify this by analyzing various cosmological observations and comparing them with the theoretical predictions based on it .
so far the most compelling evidence of isotropy comes from the near uniform temperature of the cmbr over the full sky @xcite .
this is further supported by multitude of evidences such as the isotropy in angular distributions of radio sources @xcite , isotropy in the x - ray background @xcite , isotropy of gamma - ray bursts @xcite , isotropy in the distribution of galaxies @xcite , isotropy in the distribution of supernovae @xcite and isotropy in the distribution of neutral hydrogen @xcite . but merely having isotropy around us does not guarantee homogeneity of the universe .
one can infer homogeneity from isotropy only when it is assured around each and every point with the hypothesis that the matter distribution is a smooth function of position @xcite .
unfortunately even the isotropy of the universe around us is not certain as there are mounting evidences in favour of significant anisotropy in the universe .
the cmbr which is considered to be the best supporting evidence for isotropy is not completely isotropic . over the years
many studies have reported power asymmetries and unlikely alignments of low multipoles @xcite .
recently @xcite reported a @xmath4 deviations from isotropy in the planck data .
there are other studies such as the hemispherical asymmetry of cosmological parameters @xcite , shape of the luminosity function of galaxies @xcite , hubble diagram of type ia supernovae @xcite , redshift - magnitude relation of type ia supernovae @xcite which find significant evidence for anisotropy . as a result
it is not straightforward to infer homogeneity of the universe by combining isotropy with the copernican principle .
most statistical tests to identify the transition scale to homogeneity rely on the number counts @xmath5 in spheres of radius @xmath6 and its scaling with @xmath6 which is expected to scale as @xmath7 for a homogeneous distribution . the conditional density @xcite measures the average density in these spheres which is expected to flatten out beyond the scale of homogeneity .
the multifractal analysis @xcite uses the scaling of different moments of @xmath5 to characterize the scale of homogeneity .
some of the studies carried out so far with these methods on different galaxy surveys claim to have found a transition to homogeneity on sufficiently large scales @xmath8 @xcite whereas some studies claim the absence of any such transition out to scale of the survey @xcite .
bulk flow measurements on large scales provide another important test for the large scale homogeneity .
the galaxy distribution as revealed by different galaxy redshift surveys ( sdss , @xcite ; 2dfgrs , @xcite ) indicates that the universe is highly inhomogeneous and anisotropic on small scales . the observed deviations from homogeneity and isotropy arise from peculiar velocities resulting from density perturbations .
any form of bulk flows are expected to disappear on sufficiently large scales where the density fluctuations are very small .
analysis of a large number of peculiar velocity surveys indicate statistically significant large cosmic flows on scales of @xmath9 @xcite .
combined studies with the wmap data and x - ray cluster catalog indicates strong and coherent bulk flow out to @xmath10 @xcite .
contrary to these claims a recent study of low - redshift type - ia supernovae suggest no excess bulk flow on top of what is expected in @xmath11cdm model @xcite .
if large scale bulk motions really exist then they indicate that there are significant density fluctuations on very large scales .
the results from these studies clearly indicate that there is no clear consensus in this issue yet .
the presence of inhomogeneities on very large scales has several important consequences .
inhomogeneities , through the backreaction mechanism can provide an alternate explanation of a global cosmic acceleration without requiring any additional dark energy component @xcite .
spherically symmetric radially inhomogeneous lemaitre - tolman - bondi ( ltb ) models can account for the distant supernova data in a dust universe without requiring any dark energy @xcite .
a large void can also mimic an apparent acceleration of expansion @xcite .
fortunately such models can be constrained with observations such as sne , cmb , bao and measurements of angular diameter distance and hubble parameter @xcite
. there would be a major paradigm shift in cosmology if the assumption of cosmic homogeneity is ruled out with high statistical significance by multiple data sets and it is important to test the assumption of homogeneity on multiple datasets with different statistical tools .
@xcite introduce a method based on shannon entropy @xcite for characterizing inhomogeneities and applied the method on some monte carlo simulations of inhomogeneous distributions and n - body simulations which show that the proposed method has great potential for testing the large scale homogeneity in galaxy redshift surveys .
@xcite applied this method to galaxy distributions from sdss dr12 @xcite and find that the inhomogeneities in the galaxy distributions persist at least upto a length scale of @xmath12 .
@xcite show that the methods for testing homogeneity based on the number counts are contaminated by confinement and overlapping biases which result from the systematic migration and overlap between the spheres with increasing radii .
these biases which are inevitable in any finite volume sample , artificially suppress inhomogeneities at each scale and particularly on large scales .
recently @xcite use the count in cells method in dark sky simulations @xcite to show the spurious effects in the measurement of correlation dimension due to the confinement and overlapping biases .
one can circumvent these biases by using number counts in independent spheres but the availability of only a very small number of independent spheres at increasing radii makes the statistics too noisy for most galaxy samples available including the sdss main galaxy sample . using count in cells method as well as new methods based on anomalous diffusion and random walks @xcite
show that the millennium run simulation @xcite volume is too small and prone to bias to reliably identify the onset of homogeneity .
we require galaxy distributions covering very large volume to serve the purpose .
the luminous red galaxy distribution ( lrg ) @xcite extends to a much deeper region of the universe as compared to the sdss main galaxy sample as they can be observed to greater distances as compared to normal @xmath13 galaxies for a given magnitude limit .
their stable colors also make them relatively easy to pick out from the rest of the galaxies using the sdss multi - band photometry .
the enormous volume covered by the sdss lrg distribution provides us an unique opportunity to test the assumption of homogeneity on large scales with unprecedented confidence .
we modify our method @xcite so as to carry out the analysis with non - overlapping independent regions upto sufficiently large length scales .
we also analyze the lrg data with overlapping spheres and compare the findings with that obtained using independent regions to asses the suppression of inhomogeneities on different length scales due to overlapping and confinement biases . beside analyzing the issue of cosmic homogeneity using the lrg distribution
we also address another important issue in the present work . some studies reported to have found periodicity in the observed distribution of galaxies and galaxy clusters @xcite .
both @xcite and @xcite find an apparent regularity in the galaxy distribution at a characteristic length scale of @xmath14 .
it has been suggested that in an oscillating universe the oscillation of the hubble parameter can cause such periodicity in the density fluctuations @xcite .
@xcite use the oscillations in the cluster correlation function to quantify such pictures of regularity in the galaxy distribution .
the correlation function being an isotropic statistics smooths out the anisotropic signal and may not be the best statistics to describe such anisotropic structures .
recently @xcite suggest a method which shows that shannon entropy can be also employed to test the isotropy of a distribution .
we shall use some inhomogeneous and anisotropic distributions with regularity in their distributions and test if the method presented here can describe such underlying regularities in the distributions .
a recent study by @xcite find quasi - periodical features in the distribution of lrgs .
recently @xcite detected two peaks in the distance distributions of rich galaxy groups and clusters at @xmath15 and @xmath16 around the abell cluster a2142 .
another study by @xcite find evidence for shell like structures in the sdss galaxy distribution where galaxy clusters have maxima in the distance distribution to other galaxy groups and clusters at the distance of about @xmath17 .
galaxies are distributed in a complex filamentary network which is often referred as the cosmic web .
filaments are known to be the largest known statistically significant coherent structures in the universe @xcite .
@xcite find that the filaments are statistically significant upto a length scales of @xmath18 in the lrg distribution and longer filaments , though possibly present in the data , are not statistically significant and are the outcome of chance alignments .
these studies suggest that there may be a characteristic scale present in the galaxy distribution other than the baryon acoustic oscillations ( bao ) feature observed at @xmath9 @xcite . in an earlier study @xcite find that the inhomogeneities in the sdss main galaxy distribution persist at least upto a length scales of @xmath18 and we would like to test if the inhomogeneities in the lrg distribution exhibit any regularity on even larger scales . throughout our work
, we have used the flat @xmath11cdm cosmology with @xmath19 .
@xcite propose a method based on the shannon entropy to study inhomogeneities in a 3d distribution .
shannon entropy @xcite is originally proposed by claude shannon to quantify the information content in strings of text .
it gives a measure of the amount of information required to describe a random variable .
the shannon entropy for a discrete random variable @xmath20 with @xmath21 outcomes @xmath22 is a measure of uncertainty denoted by @xmath23 defined as , @xmath24 where @xmath25 is the probability distribution of the random variable @xmath20 .
we modify the method proposed by @xcite which employs spheres of varying radii @xmath6 to measure the number counts inside them .
the spheres are allowed to overlap which suppress inhomogeneities and the degree of overlap gradually increases with increasing radii . in order to prevent overlapping of the measuring volumes which hinders measurement of inhomogeneity on large scales
we embed the galaxy distribution in a @xmath26 three dimensional rectangular grid .
this divides the entire survey region into a number of regular cubic voxels .
the voxels which are beyond the boundary of the survey region are marked and discarded from the analysis .
we also discard the voxels near the boundary which are partly inside the survey region to avoid any edge effects .
we identify the voxels which are fully inside the survey region and only consider them in our analysis .
the grid size @xmath27 is a variable which can be varied within a suitable range .
each choice of @xmath27 would result into a different number of analyzable voxels and galaxies within the survey region .
let @xmath28 be the number of resulting voxels which lie entirely within the survey region for a grid size @xmath27 .
we count the number of galaxies inside each of the @xmath28 voxels available for each choice of grid size @xmath27 .
let @xmath29 be the number counts inside the @xmath30 voxel where the voxel index @xmath31 runs from @xmath32 to @xmath28 . in general for a specific choice of the grid size
if we randomly pick up a galaxy it can reside only in one of the @xmath28 voxels available for that @xmath27 .
but which particular voxel does it belong to ?
the answer to this question has @xmath28 likely outcomes .
we define a separate random variable @xmath33 for each grid size @xmath27 which has @xmath28 possible outcomes each given by , @xmath34 with the constraint @xmath35 .
the shannon entropy associated with the random variable @xmath33 can be written as , @xmath36 where the base of the logarithm is arbitrary and we choose it to be @xmath1 .
@xmath37 will have the same value @xmath38 for all the voxels when @xmath29 is same for all of them .
this is an ideal situation when each of the @xmath28 voxels available contain exactly the same number of galaxies within them .
this maximizes the shannon entropy to @xmath39 for grid size @xmath27 .
we define the relative shannon entropy as the ratio of the entropy of a random variable @xmath33 to the maximum possible entropy @xmath40 associated with it .
the relative shannon entropy @xmath41 for any grid size @xmath27 quantifies the degree of uncertainty in the knowledge of the random variable @xmath33 .
the distribution of @xmath37 become completely uniform when @xmath42 is reached .
equivalently @xmath43 quantify the residual information and can be treated as a measure of inhomogeneity .
the fact that galaxies are not residing in any particular voxel and rather are distributed across the available voxels with different probabilities acts as the source of information .
if all the galaxies would have been residing in one particular voxel leaving the rest of them as empty then there would be no uncertainty and no information at all making @xmath44 or @xmath45 .
this fully determined hypothetical situation corresponds to maximum inhomogeneity . on the other hand
when all the @xmath28 voxels are populated with equal probabilities it would be most uncertain to decide which particular voxel a randomly picked galaxy belongs to .
this maximizes the information entropy to @xmath46 turning @xmath47 .
this corresponds to a situation when the distribution is completely homogeneous .
the galaxy distribution is expected to be inhomogeneous on small scales but with increasing grid size @xmath27 one would expect it to be homogeneous on some scale provided the universe is homogeneous on large scales . it may be noted here that one may prefer to use non - overlapping spheres to carry out the measurements but use of regular cubic voxels allow the optimal use of the available space . alternatively one can measure the inhomogeneity by using the kullback - leiblar(kl ) divergence or information divergence . in information theory kl divergence
is used to measure the difference between two probability distributions @xmath25 and @xmath48 .
+ @xmath49 let @xmath50 be the distribution corresponding to the data for which homogeneity is to be tested and @xmath51 is the distribution for a homogeneous poisson random distribution where both the distributions occupy same 3d volume with identical geometry and are represented by same number of points .
we simulate a monte carlo realization of a homogeneous poisson distribution within the survey region using same number of points as the galaxies in that region .
the monte carlo sample of the homogeneous random poisson distribution serve as our reference distribution for measuring the information divergence of the lrg distribution with respect to it .
the kl divergence between the actual and random data is then given by , + @xmath52 where @xmath53 and @xmath54 are the counts in the @xmath30 voxel for actual data and random data respectively .
+ + + + + + + +
the sloan digital sky survey ( sdss ) @xcite is a wide - field imaging and spectroscopic survey of the sky using a dedicated 2.5 m telescope @xcite with @xmath55 field of view at apache point observatory in southern new mexico .
the sdss has imaged the sky in 5 passbands , @xmath56 and @xmath57 covering @xmath58 square degrees and has so far made 12 data releases to the community .
the sdss galaxy sample can be roughly divided into two ( i ) the main galaxy sample and ( ii ) the luminous red galaxy ( lrg ) sample . for our work
, we have used the spectroscopic sample of lrgs derived from the data release 7 of sdss - ii @xcite .
sdss targets those galaxies for spectroscopy which have @xmath6 magnitude brighter than 17.77 ( @xmath59 ) . to select lrgs ,
additional galaxies are targeted using color - magnitude cuts in @xmath60 and @xmath31 which extends the magnitude limit for lrgs to @xmath61 .
the prominent feature for an early - type galaxy is the @xmath62 break in its sed . for @xmath63
this feature lies in the @xmath64 passband while it shifts to the @xmath6 band for higher redshifts .
hence selection of lrgs involves different selection criteria below and above @xmath65 , the details of which are described in @xcite .
the criteria for lower redshifts are collectively called cut - i ( eq . ( 4 - 9 ) in @xcite ) while those for the higher redshifts are called cut - ii ( eq .
( 9 - 13 ) in @xcite ) , with cut - i accounting for @xmath66 of the targeted lrgs .
a galaxy that passes either of these cuts is flagged for spectroscopy by the sdss pipeline as target_galaxy_red while a galaxy that passes only cut ii is flagged as target_galaxy_red_ii .
hence , while selecting the lrgs we require that the target_galaxy_red and target_galaxy_red_ii flags both be set . we obtain the @xmath64-band absolute magnitude from the @xmath6 band apparent magnitude accounting for the k - correction and passive evolution .
we use the prescription for @xmath67 correction from the table 1 in @xcite ( non - star forming model ) .
the cuts defined above are designed to produce an approximately volume limited sample of lrgs upto @xmath68 .
the comoving number density of lrgs falls sharply beyond @xmath68 .
it is obviously necessary to include lrgs that come from the main sample for @xmath69 .
but @xcite issue a strong advisory against selecting lrgs with @xmath70 for a volume limited sample as the luminosity threshold is not preserved for low redshifts .
we have restricted our lrg sample to the redshift range @xmath71 and @xmath64-band absolute magnitude range @xmath72 .
we identify a region @xmath73 and @xmath74 where @xmath75 and @xmath76 are the survey coordinates . combining these cuts
provide us our lrg sample which radially extends from @xmath77 to @xmath78 and consists of @xmath79 luminous red galaxies .
we have used the same data set in an earlier analysis of filamentarity of lrgs in @xcite .
the comoving number density of lrgs as a function of the radial distance @xmath6 in this sample is shown in figure 1 of @xcite .
we would like to test if our method can quantify the inhomogeneity and characterize the periodicity in a given spatial distribution .
we generate a set of monte carlo realizations for an anisotropic distribution with periodicity in their spatial distribution on certain length scale .
we decide here to generate the monte carlo realizations for a simple radial density distributions @xmath80 where @xmath81 . here
@xmath82 is a normalization constant .
these distributions are radially inhomogeneous poisson distributions which are anisotropic at all points other than the centre .
enforcing the desired number of particles @xmath83 within radius @xmath84 one can turn the radial density function into a probability function within @xmath85 to @xmath86 which is normalized to one when integrated over that interval .
so the probability of finding a particle at a given radius @xmath6 is @xmath87 which is proportional to the density at that radius implying more particles in high density regions .
we generate the monte carlo realizations of this distribution using a monte carlo dartboard technique .
the value of the function @xmath88 in @xmath89 is same and constant everywhere .
we label the maximum value of @xmath89 as @xmath90 which is same everywhere in this case .
we randomly choose a radius @xmath6 in the range @xmath91 and a probability value is randomly chosen in the range @xmath92 .
the actual probability of finding a particle at the selected radius is then calculated using expression for @xmath89 and compared to the randomly selected probability value . if the random probability is less than the calculated value , the radius is accepted and assigned isotropically selected angular co - ordinates @xmath93 within @xmath94 and @xmath95 within @xmath96 , otherwise the radius is discarded . in this way ,
radii at which particle is more likely to be found will be selected more often because the random probability will be more frequently less than the calculated actual probability .
the above @xmath93 and @xmath95 ranges are chosen so as to generate the distribution inside a cubic region .
we choose @xmath97 where @xmath98 is the length of the cube on each side .
the value of @xmath83 is chosen suitably for different @xmath98 so as to maintain a density which is higher than the mean density of lrgs .
we identify all the points whose co - ordinates @xmath99 lie within @xmath100 , @xmath101 and @xmath102 .
this gives us a cube of side @xmath103 containing a certain number of points .
we replicate this cube along x , y and z directions as many times as required so as to enable us to extract a region from this periodic distribution which has identical geometry as the lrg distribution .
we then randomly select as many points from the extracted region as there are lrgs in our sample .
we choose three different values of @xmath104 , @xmath105 and @xmath106 and generate @xmath1 monte carlo realizations of the periodic distributions for each value of @xmath98 .
we analyze each set of monte carlo realizations separately using the method described in section 2 .
a radially inhomogeneous distribution with no preferred scale must appear inhomogeneous on all scales and we want to make sure that the method proposed here efficiently capture the inhomogeneous character of such distribution on all scales .
we generate a radially inhomogeneous poisson distribution in a large cubic box where density falls of as @xmath107 from one of the corners of the box . the total number of points inside the box is decided so as to maintain a mean density comparable to the mean density of the lrg distribution .
@xmath1 such boxes are generated using the monte carlo technique discussed above and analyzed separately using the method presented in section 2 .
we show the inhomogeneity @xmath43 as a function of grid sizes or length scales for the radially inhomogeneous poisson distributions in .
the shows that the distribution is inhomogeneous on small scales which is partly due to its inherent inhomogeneous character and partly due to the poisson noise .
the observed inhomogeneity decreases with increasing grid sizes on small scales but suddenly starts increasing at @xmath108 and keep on increasing thereafter .
the contribution of poisson noise to the measured inhomogeneity decreases with increasing grid sizes due to the increase in the number counts inside them and becomes increasingly irrelevant on larger scales .
the degree of inhomogeneity on larger scales is primarily decided by the nature of the intrinsic inhomogeneity built in the distribution .
we see that for the grid sizes above @xmath109 the distribution appears to be increasingly inhomogeneous .
this indicates that there is no scale of transition to homogeneity in the distribution and the distribution can be regarded as intrinsically inhomogeneous .
it may be noted here that figure 2 of @xcite show that the measurement with overlapping spheres leads to suppression in inhomogeneity on large scales which may hide the real inhomogeneous nature of a distribution .
the method which employs independent voxels is far more reliable in a sense that there are no artificial reduction in inhomogeneity due to confinement and overlapping biases .
it is this advantage which allow us to capture the truly inhomogeneous nature of a distribution as shown in .
the method presented here is certainly advantageous in this sense and may help us to reliably measure the inhomogeneity present in a distribution on different scales . in different panels of we show how the inhomogeneities vary with length scales for inhomogeneous and anisotropic distributions which have a periodicity in their spatial distributions ( ) . the variations of inhomogeneity in a homogeneous poisson distribution is also shown together in each panel for comparison . in the top
left panel we show the inhomogeneity @xmath43 as a function of grid size for the distributions which are periodic on scales of @xmath110 .
we see that the periodic distributions and homogeneous poisson distributions both show inhomogeneity on small scales which decrease with increasing length scales .
but they vary differently as the degree of inhomogeneity on small scales would depend on the poisson noise as well as any intrinsic inhomogeneity built in the character of the distributions .
it is interesting to note that there is a sharp decrease in the inhomogeneity in the periodic distributions at @xmath109 .
we get @xmath47 when the grid size is @xmath111 which also is the periodicity of the distributions .
the inhomogeneity reappears again as the grid size exceeds @xmath112 and decreases once again and disappears exactly at @xmath113 and so on .
this can be easily explained from the fact that these periodic distributions are highly inhomogeneous below the scale of their periodicity and become homogeneous at the scale at which they are repeated .
once the grid size surpass this scale the distribution would again appear inhomogeneous by partial inclusion of the repeated patterns by the voxels .
this behaviour of @xmath43 continues with increasing grid sizes and we get successive minima at repeated intervals of @xmath112 showing precise homogeneity on those length scales .
the maximum amplitude of the inhomogeneity decreases in successive oscillations due to the decrease in the poisson noise with increasing grid sizes .
the error - bars shown here are the @xmath114 variations from the @xmath1 monte carlo realizations used in each case .
the extremely small size of the errorbars on the measured inhomogeneity at each node clearly indicates the presence of a periodicity in the distributions .
this characteristic behaviour of inhomogeneity in periodic distributions is markedly different from the smooth decrease of inhomogeneity with increasing length scales in the homogeneous poisson distributions .
we do not find this periodic variation in the homogeneous poisson distributions as there are no characteristic scales associated with them .
the variations of @xmath43 as a function of grid size for periodic distributions with periodicity of @xmath115 and @xmath116 are shown in the top right and bottom panels respectively .
we see similar periodic variations in the measured inhomogeneity at repeated intervals of @xmath117 and @xmath116 in the respective panels which clearly demonstrate the presence of periodicity in the distributions on those scales .
this shows that our method can identify the presence of any regularity in a spatial distribution and quantify the scale over which the patterns are repeated .
we show the inhomogeneity @xmath43 in the lrg distribution and a homogeneous poisson distribution as a function of length scales in the top left panel of when the data is analyzed with independent voxels .
we find that both distributions show inhomogeneity on small scales but the degree of inhomogeneity observed in the lrg distribution is higher than that of a homogeneous poisson distribution at each length scales .
this indicates that the lrgs are not distributed randomly and the amount of information encoded in the lrg distribution is higher as compared to a homogeneous poisson distribution at each scale .
the degree of inhomogeneity in both the lrg and poisson distribution decrease with increasing length scales but the inhomogeneity in the poisson distribution decays much faster as compared to the lrg distribution .
this can be explained from the fact that the information in the lrg distribution is sourced by the gravitational clustering and poisson noise whereas for the poisson distribution the contribution solely comes from the shot noise which becomes increasingly less important on larger length scales . in the top right panel
of we show the inhomogeneity @xmath118 in the lrg distribution as a function of length scales when the analysis is carried out with overlapping spheres .
we find that the distribution appears to be homogeneous on small scales .
the degree of inhomogeneity suddenly rises at a certain length scale after which it gradually decreases with increasing length scales as seen earlier in the top left panel of .
this characteristic behaviour giving rise to a bump in the inhomogeneity at a given scale arise due to the near uniformity in the distribution on length scale below its mean inter particle separation .
it may be noted that the peak of the bump is located at @xmath119 which is the mean intergalactic separation in our lrg sample .
a poisson distribution is expected to become homogeneous on a scale of the average size of voids in the distribution @xcite .
this is closely related to the mean inter - particle separation and one would expect a poisson distribution to become homogeneous above this length scale .
but we find a large degree of inhomogeneity in the lrg distribution on these length scales .
further we notice that the degree of inhomogeneity in the top right panel is smaller than that in the top left panel at each length scales .
these differences are caused by the suppression of inhomogeneities due to confinement and overlapping biases @xcite which result from the finite volume and progressive overlaps between the measuring spheres with increasing radii . in the bottom left panel of
we show the ratio of inhomogeneities measured at different length scales using independent voxels and overlapping spheres .
we find that the inhomogeneities are suppressed by a factor of @xmath120 upto @xmath121 and by a factor of @xmath1 on even larger scales in the overlapping measurements .
the ratio shows a sudden drop at @xmath122 due to the unavailability of sufficient independent voxels beyond this length scale .
the results clearly demonstrate the suppression of inhomogeneities on larger length scales when the measuring spheres are allowed to overlap .
all the methods for testing homogeneity which are based on the number counts @xmath5 in spheres of radius @xmath6 are equally affected by these biases .
therefore these biases should be properly taken into account in their applications . in the left panel of we show how the number of available voxels vary with length scales when independent voxels are used for the analysis .
clearly the number of voxels decreases with increasing grid sizes due to the finite volume of the survey region .
we see that the number of independent voxels drops below @xmath1 around @xmath123 and we decide not to consider the scales beyond this in our analysis .
the right panel of show the number of available centres as a function of the radius of the spheres when overlapping spheres are used for the analysis .
clearly in this method one can go upto a larger length scales albeit by compromising with larger degree of overlaps between the measuring spheres resulting into a greater degree of suppression in inhomogeneity .
we see that one can extend this analysis upto @xmath124 requiring at least @xmath125 overlapping spheres .
it may be noted in the top right panel of that @xmath118 decreases to @xmath126 at @xmath116 and show very little variation thereafter despite the increasing degree of overlaps between the measuring spheres .
the value of @xmath43 in the top left panel of also exhibit a quite similar behaviour where its value decreases to @xmath127 at @xmath116 and show a flattening which abruptly drops to @xmath128 at @xmath129 .
we do not attach any statistical significance to this sudden drop due to the non availability of sufficient number of voxels on that scale .
the degree of inhomogeneity at the plateau is somewhat different in the two methods .
this difference could be attributed to the fact that inhomogeneities are suppressed by a factor of @xmath120 at @xmath130 when overlapping spheres are used ( ) .
we note that @xmath125 independent voxels and @xmath131 centers of overlapping spheres are available at @xmath130 which trace a very small degree of inhomogeneity present on that scale
. the minuscule degree of inhomogeneity measured with a large number of voxels or spheres at the plateau indicates that the lrg distribution is nearly homogeneous on these length scales .
the lrg distribution is known to have a linear bias of @xmath132 on large scales @xcite and may have a larger variations in @xmath43 as compared to the homogeneous poisson distributions .
currently we do not have any estimates for the variations in @xmath43 as we have only one distribution for the lrgs . keeping this in mind
we conclude that the lrg distribution becomes homogeneous on scales of @xmath133 and beyond .
our results are consistent with the earlier findings that the distribution of galaxies in the sdss main sample becomes homogeneous on and above a scale of @xmath134 @xcite . in the bottom right panel of we show the kullback - leibler(kl ) divergence or the information divergence for the lrg distribution with respect to a homogeneous poisson distribution on different length scales when the measurements are done with independent voxels .
the definition and the significance of this measure is discussed in section 2 and this can be regarded as an alternative measure of inhomogeneity .
it is interesting to note that the kl divergence also decreases with increasing grid sizes which is quite similar to the decrease in @xmath43 and @xmath118 with increasing length scales and a similar flattening is observed at @xmath135 .
an important caveat to the present analyses arises due the fact that the comoving number density of lrgs varies with redshift and the lrg sample is considered to be only quasi volume limited @xcite .
these variations could result from both the intrinsic fluctuations in the galaxy density field and the observational selection effects @xcite .
we test how the presence of such fluctuations in number - redshift relation affect the scale of homogeneity inferred with our method . to test this
we construct @xmath1 mock lrg samples each from homogeneous and inhomogeneous poisson distributions and randomly exclude points from them so as to finally have the same comoving number density @xmath136 in them as the actual lrg sample .
the results for these analyses are shown in . in two panels
of we see a small increase in the degree of inhomogeneity when an lrg - like number - redshift relation is enforced in the homogeneous and inhomogeneous poisson distributions .
but we note that the variations of inhomogeneity with length scales in both cases still reflect the characteristics of the underlying distributions and the respective scales of transition to homogeneity in these distributions are not altered by the presence of such a number - redshift relation
. however it may be worth mentioning that this test and its outcome rely on a smooth variation in the radial density due to either selection effects or intrinsic fluctuations in the galaxy density field and it is in general difficult to test for any other sources of such fluctuations in the number - redshift relation of lrgs .
we do not see any periodic variation in the measured inhomogeneity @xmath43 of the lrg distribution ( ) in the range of length scales explored .
this is in sharp contrast with the variations observed in for periodic distributions .
this indicates that there are no underlying regularities in the distribution of the luminous red galaxies at least within the length scales probed in the present work .
one can extend this analysis to the lrg sample from the baryon oscillation spectroscopic survey of sdss - iii @xcite and future lrg sample such as from the eboss of sdss - iv @xcite to search for any periodic variation in inhomogeneity on even larger scales .
the authors would like to thank an anonymous reviewer for useful comments and suggestions .
the authors thank prajwal raj kafle for his suggestion for using kl divergence as an alternative measure of inhomogeneity .
b.p . would like to acknowledge iucaa , pune for providing support through the associateship programme .
b.p . would also like to acknowledge cts , iit kharagpur for the use of its facilities for the present work .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington . | we quantify the degree of inhomogeneity in the luminous red galaxy ( lrg ) distribution from the sdss dr7 as a function of length scales by measuring the shannon entropy in independent and regular cubic voxels of increasing grid sizes .
we also analyze the data by carrying out measurements in overlapping spheres and find that it suppresses inhomogeneities by a factor of @xmath0 to @xmath1 on different length scales . despite the differences observed in the degree of inhomogeneity both the methods show a decrease in inhomogeneity with increasing length scales which eventually settle down to a plateau at @xmath2 . considering the minuscule values of inhomogeneity at the plateaus and their expected variations we conclude that the lrg distribution becomes homogeneous at @xmath3 and beyond .
we also use the kullback - leibler divergence as an alternative measure of inhomogeneity which reaffirms our findings .
we show that the method presented here can effectively capture the inhomogeneity in a truly inhomogeneous distribution at all length scales .
we analyze a set of monte carlo simulations with certain periodicity in their spatial distributions and find periodic variations in their inhomogeneity which helps us to identify the underlying regularities present in such distributions and quantify the scale of their periodicity .
we do not find any underlying regularities in the lrg distribution within the length scales probed .
[ firstpage ] methods : numerical - galaxies : statistics - cosmology : theory - large scale structure of the universe . |
many intriguing and fascinating results were shown during this conference , which certainly require some revision of our classical understanding of the evolution of the first stellar generations in the universe .
let us briefly recall a few of them here : * the measured abundances of many elements at very low metallicity present a very small scatter ( cayrel et al .
this might indicate that already at this low metallicity , stars are formed from a well mixed reservoir composed of ejecta from stars of different initial masses .
* at least down to a metallicity of [ fe / h ] equal to -4 , there is no sign of enrichments by pair instability supernovae ( see the contribution by heger in this volume ) . * the n / o ratios observed at the surface of halo stars by israelian et al .
( @xcite ) and spite et al .
( @xcite ) indicate that important amounts of primary nitrogen should be produced by very metal - poor massive stars ( chiappini et al .
@xcite ) . * below [ o / h ] @xmath8 the c / o ratio presents an upturn indicating either that very metal poor stars produce more carbon , or less oxygen or both ( akerman et al .
@xcite ; spite et al .
@xcite ) * if most stars , at a given [ fe / h ] , present a great homogeneity in composition , a small group , comprising about 20 - 25% of the stars with [ fe / h ] below -2.5 , show very large enrichments in carbon with a great scatter , these are the c - rich extremely metal poor stars ( cemp ) . * the surface abundances of stars in globular clusters show an anti correlation between the abundances of sodium and oxygen , those of magnesium and aluminum ( see the contributions by gratton and charbonnel in this volume ) .
halo stars in the field do not show such anti correlations .
* the zero age main sequence in the very massive globular cluster @xmath9 cen can be decomposed in a blue and a red component .
the blue component is paradoxically more metal rich by about a factor two with respect to the red one .
the only way , presently , to understand such a behavior , is to suppose that the stars of the blue component are very helium - rich ( see the contributions by gratton and maeder et al . in this volume ) .
many more results could be added to the above list .
such a large amount of unexplained features points toward the need of exploring new lines of research .
some observational facts concern only a subsample of the stars ( for instance the c - rich stars , or the stars in globular clusters ) . in those cases
some special circumstances can be invoked in order to reproduce the observed features .
other observational facts as the c / o upturn or the primary nitrogen production point toward some general characteristics shared by a great part if not all the extremely metal poor stars .
classically much more massive stars can be formed at very low metallicity as a result of the absence of dust which acts as an efficient cooling agent ( see e.g abel et al .
the absence of cno elements modifies the way the h - ignition occurs in massive stars .
the lower opacity of metal poor material implies more compact stars and very weak stellar winds .
other characteristics might be different in pop iii stars , for instance the binary frequency , the amplitudes and the effects of magnetic fields .
here we want to focus on the effects of rotation . as explained in the contribution by maeder et al . in this volume , at low metallicity , the present models predict that stars have more chance to reach break - up during the main - sequence phase , are more efficiently mixed and may undergo strong mass loss through stellar winds . here
we want to present some new results illustrating this behavior and draw some consequences for nucleosynthesis at very low metallicity .
the case of rotating pop iii models is briefly discussed elsewhere in the present volume ( see the contribution by ekstrm et al ) . in the present paper , we shall concentrate on stars with a very small initial amount of metals .
it is interesting to point out that already a very small amount of metals makes a world of difference with respect to strictly pop iii stars .
this is illustrated in fig .
[ fig1 ] , presenting the chemical structure for two rotating 60m@xmath10 stars with @xmath11 = 800 km s@xmath12 .
three differences between the pop iii and the @xmath0 model can be seen : in the @xmath0 stellar model 1 ) the co core is smaller ; 2 ) the quantity of primary nitrogen is much greater ; 3 ) an extended convective zone is associated to the h - burning shell ( from about 20 to 34 m@xmath10 ) . in the @xmath0 stellar model , the cno content in the region of the star where the h - shell ignites is already sufficient for the h - burning to occur through the cno cycle .
this implies that from its birth the h - burning shell can compensate for a great part of the energy lost by the surface . typically , the h - shell luminosity is about one half the total one .
the star then adjusts its structure so that the he - burning core has only to compensate for the other half of the total luminosity . instead in the pop iii model
, the cno content in the region where the h - burning shell occurs is so low that the shell only succeeds in compensating for a very small fraction of the total luminosity ( typically a few percents ) , the rest having to be compensated by the he - burning core .
this explains the much bigger helium core in the pop iii model and the very modest h - burning shell . as a consequence in the pop iii model ,
much less primary nitrogen is produced and the h - shell remains radiative for a much longer period .
we do not know what was the initial rotation of the massive first stellar generations . on the other hand
, we can observe the rotational velocity of solar metallicity massive stars , and suppose that the very metal poor stars had , at the beginning of their evolution , about the same angular momentum as their solar metallicity counterparts . for pop iii stars this implies an initial velocity of about 800 km s@xmath12 .
the same initial velocity of 800 km s@xmath12 was considered for the very small metallicities @xmath0 and @xmath1 considered in the present paper .
the evolution we shall describe is more dependent on the choice of the initial velocity than on the precise choice of the initial metallicity .
both rotating models at @xmath0 and @xmath1 present the same qualitative behaviour , namely they both produce large amounts of primary nitrogen and lose a great part of their initial mass through stellar winds .
[ fig2 ] shows the evolution of the ratio @xmath13 at the surface during the ms phase . at both @xmath0 and @xmath1 ,
the fast models reach the break - up limit , the `` metal - rich '' model at an earlier stage than the `` metal - poor '' one .
this comes from the fact that when the metallicity increases , a given value of the initial velocity corresponds to a higher initial value of the @xmath14 ratio .
the stars then remain at break - up for the rest of their ms lifetimes and undergo an enhancement of their mass loss rates . as a consequence
, the @xmath0 model ends its ms life with 57.6 m@xmath10 , having lost 4% of its initial mass , while the @xmath15 model ends its ms life with 53.8 m@xmath10 , having lost 10% of its initial mass . despite the stars stay in the vicinity of the break - up limit during an important part of their ms lifetime
, they do not lose very important amounts of mass .
this is due to the fact that only the outermost layers of the stars are above the break up limit and are ejected .
these layers have low density and thus contain little mass .
the 60 m@xmath10 model at @xmath16 with @xmath11= 300 km s@xmath12 computed by meynet & maeder ( @xcite ) reaches the break - up velocity much later , only at the end of the ms phase . at @xmath15 ,
the velocity 300 km s@xmath12 appears thus as the lower limit for the initial rotation , allowing a 60 m@xmath10 star to reach the break - up limit during its ms phase .
this 60 m@xmath10 star ends its ms life with 59.7 m@xmath10 , having lost only 0.5% of its initial mass . during the ms phase ,
the surface of the rotating stars is enriched in nitrogen and * depleted in carbon * as a result of rotational mixing .
the n / c ratios are enhanced by more than two orders of magnitude at the end of the h - burning phase .
the total amount of cno elements remains however constant .
during the core he - burning phase , primary nitrogen is synthesized in the h - burning shell , due to the rotational diffusion of carbon and oxygen produced in the helium core into the h - burning shell ( meynet & maeder @xcite ) .
this is well illustrated in fig .
[ fig4 ] for the @xmath15 rotating model .
in contrast to what happens during the ms phase , rotational mixing , during the core he - burning phase , induces strong changes of the surface metallicity .
these changes only occur at the end of the core he - burning phase , when an outer convective zone appears and rapidly deepens in mass , dredging up newly synthesized elements to the surface . from this stage onwards the surface metallicity increases in a spectacular way .
typically our 60 m@xmath10 model at @xmath0 ends its life with a surface metallicity one million times higher than its initial metallicity , _
i.e. _ with @xmath17 , a metallicity greater that the metallicity of the large magellanic cloud ! this increase of the surface metallicity is entirely due to the arrival in great quantities at the surface of primary cno elements , carbon and oxygen being produced in the he - core and the nitrogen in the h - shell .
the consequence of such important surface enrichments on the mass loss rates remains to be studied in details using models of stellar winds with the appropriate physical characteristics ( position in the hr diagram and chemical composition ) . in absence of such sophisticated models
, we applied here the usual rule , namely @xmath18 , where @xmath2 is the metallicity of the outer layers . with this prescription ,
the stars lose a great part of its initial mass .
our 60 m@xmath10 at @xmath0 ends its life with only 24 m@xmath10 , the corresponding model at @xmath15 reaches a final mass of 37 m@xmath10 .
we have also computed a 7 m@xmath10 stellar model at @xmath15 with @xmath19 km s@xmath12 .
this model , in contrast with its more massive counterpart , never reaches the break - up limit during the ms phase .
the 7 m@xmath10 is more compact than the 60m@xmath10 model , this makes the gratton - pick term in the expression for the velocity of the meridional circulation smaller ( see the contribution by maeder et al . in this volume ) .
the angular momentum is thus less efficiently transported outwards than in the more massive model , making the reaching of the break - up limit more difficult .
interestingly , the model present a higher degree of mixing than the 60 m@xmath10 stellar model , because , due to the relative inefficiency of the angular momentum transport , the gradients of @xmath20 remain steeper , making the shear diffusion stronger .
this is in contrast to what happens at higher metallicity .
indeed it was shown ( see maeder & meynet @xcite ) that at a given metallicity and for a given initial velocity , the mixing was more efficient in the more massive stars .
this is correct as long as the gradients of @xmath20 do not depend too much on the initial mass .
here at very low @xmath2 , the gradients are very sensitive to the initial mass , being steeper in smaller initial mass stars .
the 7 m@xmath10 remains in the blue part of the hr diagram during the whole he - burning phase , preventing an outer convective zone to appear and to dredge - up the primary cno elements . only at the end of the core he - burning phase
, the star evolves to the red and approach the base of the asymptotic giant branch . at this point , an outer convective zone appears and produces an enormous enhancement of the surface metallicity .
this can be seen in fig .
[ fig5 ] , which compares the chemical structure of the 7 m@xmath10 at the early agb phase , with that of the 60 m@xmath10 model at the end of the core c - burning phase .
we see that in the outer layer of the 7 m@xmath10 the abundances of cno elements are at about the same level , while in the 60 m@xmath10 , the abundance of nitrogen is about at the same level as in the 7 m@xmath10 model , while those of carbon and oxygen are many orders of magnitude below . ,
@xmath21 and @xmath1 with and without rotation .
the contribution of the stellar winds have been accounted for when present.,width=377 ]
let us first discuss the cno yields of pop iii stellar models ( see ekstrm et al . in this volume ) .
the yields of carbon and oxygen appear to be little affected by rotation , while those of nitrogen are enhanced by 3 to 4 orders of magnitude in rotating models .
the yields for nitrogen are compared with those obtained in other models in fig .
we see that the pop iii yields from rotating model with @xmath22 km s@xmath12 are in general higher than those for @xmath15 with @xmath23 km s@xmath12 computed by meynet & maeder ( @xcite ) . however , according to chiappini et al .
( @xcite ) , the yields from massive stars , required to fit the n / o ratios in the unmixed sample of halo stars observed by israelian et al .
( @xcite ) and spite et al .
( @xcite ) , should be enhanced by a factor 50 ( 1.7 dex ) , _ i.e. _ should be much higher than predicted by the present pop iii stars for stars in the mass range between 15 and 85 m@xmath10 . unless , only very massive stars are formed , it appears thus difficult to explain the high n / o ratios observed in halo stars .
moreover it is likely that the production of primary nitrogen by massive stars extends over a range of metallicities and is not restricted to pop iii stars .
as shown above , a tiny amount of metal can boost in a very important way the amount of primary nitrogen produced . in that case , starting with a velocity of @xmath22 km s@xmath12 would allow to reach levels well above those required by chiappini et al .
( @xcite ) to fit the observed n / o in halo stars . thus rotating massive star models
have no difficulty in making the great amounts of primary nitrogen which seem to be required by the observations . more work is still needed in order to explore the range of initial conditions which would give the best agreement .
spectroscopic surveys of very metal poor stars ( beers et al .
@xcite ; beers @xcite ; christlieb @xcite ) have shown that carbon - rich extremely metal poor stars ( cemp stars ) account for up to about 25% of stars with metallicities lower than [ fe / h]@xmath24 .
a star is said to be c - rich if [ c / fe]@xmath25 .
the two most iron - deficient stars observed so far , he 0107 - 5240 , a giant halo star , and he 1327 - 2326 , a dwarf or subgiant halo star , are c - rich stars .
this might indicate that the frequency of c - rich stars increases when the metallicity decreases .
observations of c - rich non - evolved stars ( _ i.e. _ dwarf or subgiant ) indicate also that the pattern of abundances was already present in the cloud from which the star formed and is therefore not due to a mechanism occurring in the star itself .
many scenarios have already been proposed to explain the very peculiar abundances at the surface of these stars ( supernova with mixing and strong fall back , umeda & nomoto @xcite ; mixing of the ejecta of two supernovae , limongi et al .
@xcite ; mass transfer from an agb star , suda et al .
@xcite ) . here
we propose new scenarios based on rotating models .
let us first see if the cemp stars could be formed from material made up of massive star wind ( or at least heavily enriched by winds of massive stars ) . at first sight
such a model might appear quite unrealistic , since the period of strong stellar winds is rapidly followed by the supernova ejection , which would add to the wind ejecta the ejecta of the supernova itself .
however , for massive stars , it might occur that at the end of their nuclear lifetime , a black hole , swallowing the whole final mass , is produced . in that case
, the massive star would contribute to the local chemical enrichment of the interstellar medium only through its winds .
let us suppose that such a situation has occurred and that the small halo star that we observe today formed from the shock induced by the stellar winds with the interstellar material . what would be its chemical composition
? its iron content would be the same as that of the massive star since the iron abundance in the interstellar medium had no time to change much in the brief massive star lifetime .
also , the massive star wind ejecta are neither depleted nor enriched in iron .
the abundances of the other elements in the stellar winds for our two rotating 60 m@xmath10 at @xmath0 and 10@xmath26 are shown in fig .
[ fig8 ] . for the two metallicities considered here ,
the wind material of rotating models is characterized by n / c and n / o ratios between @xmath27 1 and 40 , and @xmath7c/@xmath5 c ratios around 4 - 5 .
these values are compatible with the ratios observed at the surface of cs 22949 - 037 ( depagne et al .
@xcite ) : n / c @xmath27 3 and @xmath7c/@xmath5c @xmath274 . the observed value for n / o ( @xmath270.2 ) is smaller than the range of theoretical values , but greater than the solar ratio ( @xmath27 0.03 ) .
thus the observed n / o ratio also bears the mark of some cno processing , although slightly less developed than in our stellar wind models . on the whole , a stellar wind origin for the material composing this star does not appear out of order in view of the above comparisons , especially if one considers the fact that , in the present comparison , there is no fine tuning of some parameters in order to obtain the best agreement possible .
the theoretical results are directly compared to the observations .
moreover only a small subset of possible initial conditions has been explored .
other cemp stars present however lower values for the n / c and n / o ratios and higher values for the @xmath7c/@xmath5c ratio . for these cases
it appears that the winds of our rotating 60 m@xmath10 models appear to be too strongly cno processed ( too high n / c and n / o ratios and too low @xmath7c/@xmath5c ratios ) .
better agreement would be obtained if the observed abundances also result from material from the co - core , ejected either by strong late stellar winds or in supernova explosion . to explore this possibility
we have study the case where wind ejecta are mixed with supernova ejecta and interstellar material .
at least two free parameters have to be introduced : 1 ) the dilution factor between the ejecta and the interstellar medium ; 2 ) the mass of iron in the supernova ejecta .
it is possible to adopt values of these two parameters so that the observed values of [ fe / h ] , [ o / fe ] and [ c / fe ] can be reproduced by both our non - rotating and rotating 60 m@xmath10 model .
however , the observed [ n / fe ] ratios can only be obtained using the rotating model .
the wind and supernova ejecta model lowers the n / c and n / o ratios improving the agreement with respect to the pure wind model , however at the cost of two free parameters ! another model has been suggested by suda et al .
( @xcite ) : the small halo star , observed today , was the secondary in a binary system whose primary went through the agb stage . at this stage
, part of the agb envelope has been accreted by the secondary greatly modifying its original surface abundances . as was the case for the massive star wind model seen above
, the iron must come from a previous star generation , since no iron is produced by the agb star . using our models computed in meynet & maeder ( @xcite ) for initial masses between 2 and 7 m@xmath10 at @xmath15 and with @xmath28 and @xmath29 km s@xmath12 , we can estimate the chemical composition of the envelope of intermediate mass stars at the beginning of the thermal pulse asymptotic giant branch phase .
the envelope is all the matter above the co - core .
the range of values for the cno elements given by these models are shown in fig .
[ fig9 ] ( vertical thin lines : no - rotation ; vertical thick lines , with rotation ) .
we have also computed a new 7 m@xmath10 with @xmath22 km s@xmath12 at @xmath15 ( continuous line with solid circles ) .
we see that the envelopes of agb stellar models with rotation show a chemical composition very similar to that observed at the surface of cemp stars . in particular , only rotating models are able to simultaneously explain the large abundances of c , n and o. it is however difficult to conclude that rotating intermediate mass star models are better than rotating massive star models in reproducing the abundance pattern of cemp stars .
probably , some cemp stars are formed from massive star ejecta and others from agb star envelopes .
interestingly at this stage , some possible ways to distinguish between massive star wind material and agb envelope do appear .
indeed , massive star wind material is characterized by very low @xmath7c/@xmath5c ratio , while intermediate mass stars seem to present higher values for this ratio .
agb envelopes would also present very high overabundances of @xmath6o , @xmath30o , @xmath31f and @xmath32ne , while wind of massive rotating stars present a weaker overabundance of @xmath6o and depletion of @xmath30o , @xmath31f and @xmath32ne . as discussed in frebel et
al . ( @xcite ) , the ratio of heavy elements as the strontium to barium ratio can also give clues as to the origin of the material from which the star formed . in the case of he 1327 - 2326 , frebel et al .
( @xcite ) give a lower limit of [ sr / ba ] @xmath33 , which suggests that strontium was not produced in the main s - process occurring in agb stars , leaving thus the massive star hypothesis as the best option , in agreement with the result from @xmath7c/@xmath5c in g77 - 61 ( plez & cohen @xcite ) and cs 22949 - 037 ( depagne et al . | rotating massive stars at @xmath0 and @xmath1 lose a great part of their initial mass through stellar winds .
the chemical composition of the rotationally enhanced winds of very low @xmath2 stars is very peculiar .
the winds show large cno enhancements by factors of @xmath3 to @xmath4 , together with large excesses of @xmath5c and @xmath6o and moderate amounts of na and al .
the excesses of primary n are particularly striking . when these ejecta from the rotationally enhanced winds are diluted with the supernova ejecta from the corresponding co cores , we find [ c / fe ] , [ n / fe],[o / fe ] abundance ratios very similar to those observed in the c rich extremely metal poor stars ( cemp ) .
we show that rotating agb stars and rotating massive stars have about the same effects on the cno enhancements .
abundances of s - process elements and the @xmath7c/@xmath5c ratio could help us to distinguish between contributions from agb and massive stars . on the whole
, we emphasize the dominant effects of rotation for the chemical yields of extremely metal poor stars . |
the concept of _ compact galaxies _ was introduced by zwicky @xcite , who has described them as galaxies barely distinguishable from stars " on the palomar sky survey plates . the term blue compact dwarf ( bcd )
galaxies @xcite identify those objects that show low luminosity , small linear dimensions , strong emission lines superposed on a blue continuum , and spectral properties that indicate low chemical abundances .
bcds form a subset of hii galaxies , a large number of which have been identified on objective prism surveys by @xcite , @xcite , @xcite and @xcite by the presence of strong emission lines , similar to giant hii regions ( ghiirs ) in our galaxy . here
, we will refer to hii / bcd galaxies as objects with a metallicity 7.0@xmath112+log(o / h)@xmath18.4 ( e.g. , @xcite ) , low luminosity ( m@xmath2 @xmath3 -18 ) and gas - rich objects ( e.g. , @xcite ) undergoing vigorous starburst activity in a relatively small physical size ( @xmath41 kpc ) .
the star - forming component , in these objects , typically contains multiple knots of star - formation with unresolved ensembles of young star clusters ( e.g. , @xcite , @xcite ) .
the hypothesis of these systems being young , forming their first generation of stars has been discarded by the detection of an evolved underlying stellar host with an age @xmath51 gyr , in the majority of the nearby hii / bcd population ( e.g. , @xcite,@xcite ) .
figure [ fig1 ] shows the optical spectrum of the galaxy tol 2146 - 391 obtained using integral field unit ( ifu ) observations with gemini / gmos . in this figure
we label the most important emission lines used in our studies , in particular , the strong balmer hydrogen recombination lines and collisionally excited emission lines , such as [ o ii ] @xmath63726,3729 , [ o iii ] @xmath04363 , [ o iii ] @xmath05007 , [ s ii ] @xmath66717,6731 , [ n ii ] @xmath06584 , which have been used for the determination of physical conditions ( e.g. , electron temperature and density ) and chemical abundances ( e.g. , oxygen , nitrogen , etc ) .
we also detect in some of our galaxies the high - ionization emission line he ii @xmath04686 .
although progress has been made in this field , important unsolved questions remain with regard to the mode of star formation ( e.g. , quasi - continuous vs fluctuating ) , and the triggering mechanism of ongoing starburst activity in hii / bcds .
it has been suggested @xcite that the cluster formation efficiency is lower in compact hii / bcd galaxies than the one found in more luminous galaxies .
these luminous systems generally show an irregular outer shape and kinematical signatures of merging in their interstellar medium ( ism ) . in some cases
, the formation of star cluster complexes occurs coevally @xcite , whereas in others star formation occurs in a propagating manner @xcite . in any case , the mechanism which may trigger the current star - formation in these galaxies is not well understood ; in particular the relative importance of intrinsic and environmental properties remains a subject of investigation .
another important issue is the chemical and kinematical imprints of star cluster formation and evolution on the spatially resolved properties of the ism in hii / bcd galaxies . as a natural consequence of star formation driven feedback ,
the newly synthesized elements will be dispersed and mixed across the ism via hydrodynamic mechanisms ( e.g. , @xcite ) , leading to _ chemical homogeneity in the oxygen abundance _
@xcite of hii / bcd galaxies ( e.g. , @xcite and references therein ) .
the nitrogen - to - oxygen ratio n / o has also been found to be rather constant ( log(n / o)@xmath7 - 1.6 ; @xcite ) at low metallicity ( 12+log(o / h)@xmath87.6 ) , suggesting primary production by massive stars @xcite as the main contribution to nitrogen enrichment at those very low metallicities .
a small fraction of hii / bcd galaxies fall in this very - low metallicity regime @xcite and are commonly referred to as extremely metal poor ( xmp ) galaxies or xmp bcds .
these galaxies are the best nearby candidates for cosmologically young objects , as various arguments imply that they have formed most of their stellar mass in the past 13 gyr @xcite . at intermediate metallicity ( 7.6@xmath112+log(o / h)@xmath18.3 ) the large observed spread in n / o has been attributed mainly to the loss of heavy elements via galactic winds @xcite , and/or to the delayed release of nitrogen by intermediate and/or massive stars and oxygen by massive stars @xcite .
however , the delayed - release scenario can not explain the presence of some hii / bcd galaxies with a high n / o ratio at low metallicities .
the most plausible explanation for the high n / o ratio observed in these objects is the chemical pollution of the ism by nitrogen released by massive wolf - rayet ( wr ) stars as is , apparently , the case of the well - studied bcd ngc 5253 @xcite . finally , at higher metallicities ( 12+log(o / h)@xmath98.3 ) the n / o ratio clearly increases with increasing oxygen abundance and the nitrogen content is mainly due to secondary production by intermediate - mass stars .
so far , an increasing number of hii / bcd galaxies has been studied with ifu spectroscopy ( see table [ tab1 ] where we provide an overview of the literature ) with main focus on the spatial properties of the ism .
recently , we have started a program investigating with ifu spectroscopy the physical conditions in the ism of the most compact hii / bcd galaxies , laying special emphasis on the extinction patterns , emission line ratios , oxygen and nitrogen abundances , kinematics and the relation on the intrinsic properties of star formation as well as possible evolutionary effects @xcite . to this end , we observed a sample of hii / bcd galaxies using the gmos - ifu on gemini south and north and , more recently , with vlt / vimos .
the gmos - ifu observations were performed using the gratings b600 and r600 in one slit mode , covering a total spectral range from @xmath103000 to @xmath107230 @xmath11 .
this observational setup provides a pattern of 500 hexagonal elements with a projected diameter of 0``.2 , covering a total 3''.5 @xmath125 `` field of view ( fov ) .
the vimos - ifu observations were obtained using the gratings hr@xmath13blue and hr@xmath13orange covering a spectral range from @xmath103710 to @xmath107700 @xmath11 .
our data yield a scale on the sky of 0.''33 per fiber , and cover a fov of 13 `` @xmath12 13 '' . in figure [ fig2 ]
we show the g - band acquisition image of the xmp bcd galaxy hs2236 + 1344 , in which we indicated the total field of view of 4``@xmath128 '' and the h@xmath14 map of the galaxy obtained from the composition of two different pointings with gmos - ifu . in figure [ fig3 ]
we show the h@xmath14 emission line map of the galaxies um 461 and tol 65 obtained using vimos - ifu .
table 2 lists the general parameters of our sample of galaxies .
[ cols="<,^,^,^,^,^,^ , > , < , > " , ] [ tab2 ]
using the reddening corrected emission line intensities of the spectra of each one of the spaxels we can derive the physical conditions ( electron temperature and density ) and the chemical abundances ( o and n ) across the ism of the galaxies .
we calculate oxygen abundances in regions where the [ oiii]@xmath04363 emission line has been detected assuming o / h = o@xmath15/h@xmath15 + o@xmath16/h@xmath15 , while nitrogen abundances are obtained assuming n / h = icf(n ) @xmath12 n@xmath15/h@xmath15 , with icf(n ) denoting the ionization correction factor ( o@xmath15+o@xmath17)/o@xmath15 . for the sake of illustration , in figure [ fig4 ]
we show the spatial distribution of 12+log(o / h ) in the gmos - ifu fov of the galaxy tol 2146 - 391 .
we can see in this figure that , despite a slight depression in the inner part of the galaxy , the oxygen abundance appears to be uniform across the galaxy ( see figure 18 in @xcite ) .
in @xcite we compare the spatial distribution of 12+log(o / h ) , found in @xcite , with the position of the star cluster / complexes detected in the galaxy um 408 by @xcite using high resolution near - ir k@xmath18-band images .
we found that the variation of the observed data points ( see figure 9 in @xcite and figure 1 in @xcite ) may not be statistically significant , indicating that these regions have identical chemical properties within the errors .
it is interesting to note that we observed a marginal gradient of decreasing abundance from the center outward in um 408 , indicating that the highest abundance values are found near the peak of h@xmath14 emission and extinction c(h@xmath19 ) , and coincident with the position of the brightest star cluster / complex @xcite . in any case , the absence of chemical overabundances in the ism of um 408 , tol 2146 - 391 , tol 0104 - 388 and hs2236 + 1344 and in the dwarf galaxies studied in the literature ( e.g. , @xcite ) indicates that the population of young star clusters is not producing localized oxygen overabundances .
the most likely explanation for this is that metals formed in the current star - formation episode reside in a hot gas phase ( t@xmath1010@xmath20 k ; @xcite ) ; thus , they are not observable in optical wavelengths . whereas metals from previous star - formation events are well mixed and homogeneously distributed through the whole galaxy . in tol 2146 - 391 , the 12+log(n / h ) radial distribution shows a slight decrease with radius .
this would argue in favor of heavy elements being produced in a previous burst of star - formation and dispersed within the ism by starburst - driven super - shells @xcite , while the depressed central region could be attributed to radial inflow of relatively low metallicity gas from large radii to the center , thus diluting the abundance of the gas in the nuclear region .
5 '' covered by our gmos - ifu observation .
right : 12 + log(o / h ) spatial distribution .
the isocontours display the h@xmath14 emission .
the maximum h@xmath14 emission is indicated in the maps by an x symbol .
we only considered spaxels with signal to noise ratio ( s / n ) @xmath5 3 in the [ oiii ] @xmath04363 line .
more details in @xcite . ] regarding the integrated properties of the galaxies , @xcite , @xcite and @xcite suggest that there is a dependence between n / o and the ew(h@xmath19 ) , in the sense of an increasing n / o ratio with decreasing ew(h@xmath19 ) .
izotov et al .
@xcite argue that this trend is naturally explained by nitrogen ejection from wr stars . in the following analysis
, we mainly concentrate on the spatially resolved physical properties of the ism in individual galaxies and their possible relation to the star formation process ( e.g. , the star formation history , burst parameter , wr star content ) . in figure 5
( see @xcite ) we show the log ( n / o ) versus ew(h@xmath19 ) and 12+log(o / h ) versus log ( n / o ) for all spaxels of the galaxies tol 0104 - 388 and tol 2146 - 391 . from that figure , it can be seen that the ew(h@xmath19 ) values are rather constant , with a very small variation of equivalent widths as the n / o ratio increases .
a comparison of log ( n / o ) versuss 12+log(o / h ) in tol 2146 - 391 ( figure 3b ) reveals that the log ( n / o ) values increase with the 12+log(o / h ) .
this data point distribution has similar patterns to those found in hii / bcd galaxies by @xcite of increasing n / o ratios with respect to the oxygen abundance .
the inner region of tol 2146 - 391 ( near the peak of h@xmath14 ) presents n / o ratios which are larger than those expected by pure primary production of nitrogen .
this might be a signature of time delay between the release of oxygen and nitrogen @xcite , or gas infall or outflow . in any case , for the metallicity of tol 2146 - 391 purely secondary nitrogen enrichment appears implausible . in the case of hs2236 + 1433 , we reported in @xcite evidence for a high n / o ratio in one of the three ghiirs of the galaxy
. but again , the spatial distribution of these abundances , at large scales , lead us to consider that oxygen and hydrogen are well mixed and homogeneously distributed over the ism of the galaxy . ) .
bottom panel : 12 + log ( o / h ) ratio versus log(n / o ) .
triangles correspond to the data points of tol 0104 - 388 and circles corresponds with the data points of tol 2146 - 391 .
more details in @xcite.,width=268 ] in summary , the results obtained in our studies suggest that the chemical properties ( o , n and n / o ) across hii / bcd galaxies are fairly uniform , although a slight gradient of o and n are observed in the ism of um 408 and tol 2146 - 391 , respectively .
we suggest that global hydro - dynamical processes , such as starburst - driven super - shells or / and inflow of gas might be governing the transport and mixing of metals across these galaxies , keeping the n / o ratio constant through the ism at large scales @xcite .
the origin of high - ionization emission lines , such as [ ne v ] @xmath03426 , [ fe v ] @xmath04227 and he ii @xmath04686 in starburst and hii / bcd galaxies has been a subject of study in the last years , given that photoionization models of hii regions generally fail to reproduce the observed intensities of these lines ( see , e.g. , @xcite ) .
several mechanisms for producing hard ionizing radiation have been proposed in the literature , such as wr stars @xcite , primordial ( zero - metallicity ) stars , high - mass x - ray binaries ( hmxb ; @xcite ) , radiative shocks @xcite and o stars at low metallicity @xcite . in @xcite and
@xcite we studied with gmos - ifu the spatial distribution of he ii @xmath04686 in the compact hii galaxies tol 0104 - 388 and tol 2146 - 391 , and in the xmp bcd galaxy hs 2236 + 1344 , respectively , in order to gain insights into the nature of their hard ionization radiation and its possible dependence on the properties of the ism @xcite .
based on a spaxel - by - spaxel analysis , instead of the integrated properties of the galaxies ( see figure 15 in @xcite ) , our results indicate that the spatial distribution of he ii @xmath04686 relative to h@xmath19 does not depend on the ew(h@xmath19 ) , oxygen abundance or log(n / o ) . in particular , the oxygen abundance appears to be constant through the whole extent of our sample galaxies , as already is observed in other hii / bcd galaxies ( e.g. , @xcite ; and references therein ) .
the opposite trend is found if we consider the integrated spectra of galaxies , in the sense that this emission line is stronger in galaxies at low metallicity @xcite .
the lack of a relationship between the hardness of the ionizing radiation and the ew(h@xmath19 ) , or age , @xcite suggests that the presence of high - ionization lines , in particular he ii @xmath04686 , is not due to a single excitation mechanism .
for instance , in @xcite , it was found for a sample of galaxies , with detected and non - detected wr features , the same dependence of i(he ii @xmath04686)/i(h@xmath19 ) on the ew(h@xmath19 ) .
this indicates that wr stars are not the sole origin of he ii @xmath04686 in star - forming regions ( see also @xcite ) . in galaxies with detected wr stars ,
the he ii @xmath04686 commonly appears to not be coincident with the location of the wr bumps ( e.g. , mrk 178 ) and these stellar features are not always seen when nebular he ii is observed ( e.g. , tol 2146 - 391 , hs2236 + 1344 , tol 65 ) .
the spatial offset between wr stars and he ii @xmath04686 , in mrk 178 , is interpreted by @xcite as an effect of the mechanical energy injected by wr star winds , so wr stars are not ruled out as the main source of the observed he ii @xmath04686 in that galaxy .
an examination of individual spaxels in our data cubes , and also in the integrated spectra of our sample galaxies ( e.g. , figure [ fig1 ] in this contribution ) , does not reveal any clear stellar wr features . in the case of the xmp bcd galaxy hs 2236 + 1344
, we detected the heii @xmath04686 emission line in only one of the ghiirs of the galaxy ( the brightest one ) . in this galaxy
, the he ii @xmath04686 line appears to be excited through point sources within a compact volume which , interestingly , does not coincide with the position where a high n / o abundance ratio has been observed .
we discuss , in @xcite , the possibility that the heii @xmath04686 emission line , in hs2236 + 1344 , is associated with wr stars , high - mass x - ray binaries ( hmxbs ) , o stars at low metallicities , and/or a low - luminosity active galactic nucleus .
however , since clear wr features have not been detected in that galaxy , wr stars are excluded as the primary excitation source of he ii @xmath04686 emission .
as far as the spatial distribution of oxygen abundances is concerned , we did not detect localized overabundances in any of our sample galaxies .
however , we find evidence for a marginal negative radial abundance gradient , with the highest abundances seen at the position of the brightest star cluster complexes ( peak of h@xmath14 emission ) , in the hii / bcd galaxy um 408 at least .
if real , the slight trend for an increasing 12+log(n / h ) abundance , in the galaxy tol 2146 - 391 , suggests rapid self - enrichment by the freshly produced heavy elements in the present starburst on scales of hundreds of pc , or , alternatively , metal pollution by a previous star formation episode . in any case , the oxygen and nitrogen appear to be well mixed across the ism of hii / bcd galaxies , suggesting efficient transport by expanding starburst - driven supershells and/or gas infall from the halo .
our spectroscopic ifu studies suggest a mixture of compact sources as the main excitation source for localized he ii @xmath04686 emission in hii / bcd galaxies , without clear wr signatures , with wr stars probably being of secondary importance . in the galaxy tol 2146 - 391 , we favor the idea of extended he ii @xmath04686 emission being primarily due to radiative shocks in the ism .
p.l . is supported by a post - doctoral grant sfrh / bpd/72308/2010 , funded by fct ( portugal ) , and p.p . by ciencia 2008 contract , funded by fct / mctes ( portugal ) and poph / fse ( ec ) .
we are very thankful to andrew humphrey for their very useful suggestions which have improved the paper .
we would like thank the anonymous referee for his / her comments and suggestions which substantially improved the paper .
we acknowledge support by the fundao para a cincia e a tecnologia ( fct ) under project fcomp-01 - 0124-feder-029170 ( reference fct ptdc / fis - ast/3214/2012 ) , funded by the feder program .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
we acknowledge the usage of the hyperleda database ( http://leda.univ-lyon1.fr ) .
the data presented in this paper have been obtained through the gemini programs gs-2004b - q-59 , gs-2005b - q-19 and gn-2010b - q-69 , and the eso - vlt program 090.b-0242 .
alloin d. , collin - souffrin s. , joly m. , & vigroux j. m. , 1979 , a&a , 78 , 200 berg , d. a. , skillman , e. d. , marble , a. r. , et al .
2012 , apj , 754 , 98 bordalo , v. , plana , h. , & telles , e. 2009 , apj , 696 , 1668 brinchmann j. , kunth d. , & durret f. , 2008 , a&a , 485 , 657 cairs l. m. , caon n. , papaderos p. , kehrig c. , weilbacher p. , & roth m. , zurita c. , 2009 , apj , 707 , 1676 cairs l. m. , caon n. , zurita c. , kehrig c. , weilbacher p. , & roth m. , 2009 , a&a , 507 , 1291 cairs l. m. , caon n. , zurita c. , kehrig c. , roth m. , & weilbacher p. , 2010 ,
a&a , 520 , 90 cairs , l. m. , caon , n. , garca lorenzo , b. , kelz , a. , roth , m. , papaderos , p. , & streicher , o. 2012 , a&a , 547 , 24 cresci , g. , vanzi , l. , sauvage , m. , santangelo , g. , & van der werf , p. 2010 , a&a , 520 , 82 croxall , k. v. , van zee , l. , lee , h. , et al .
2009 , apj , 705 , 723 dopita m. a. , & sutherland r. s. , 1996 , apjs , 102 , 161 edmunds m. g. & pagel b. e. j. 1978 , mnras , 185 , 77 filho , m. e. , winkel , b. , snchez almeida , j. , aguerri , j. a. , amorn , r. , ascasibar , y. , elmegreen , b. g. , elmegreen , d. m. , gomes , j. m. , humphrey , a. , lagos , p. , morales - luis , a. b. , muoz - tun , c. , papaderos , p. , & vlchez , j. m. 2013 , arxiv:1307.4899 garca - lorenzo , b. , cairs , l. m. , caon , n. , monreal - ibero , a. , & kehrig , c. 2008 , apj , 677 , 201 garnett d. r. 1990 , apj , 363 , 142 garnett d. r. , kennicutt r. c. , jr , chu y .- h .
, & skillman e. d. , 1991 , apj , 373 , 458 guseva n. g. , izotov y. i. , & thuan t. x. , 2000 , apj , 531 , 776 guseva , n. g. , izotov , y. i. , stasiska , g. , fricke , k. j. , henkel , c. , & papaderos , p. 2011 , a&a , 529 , 149 haro , g. 1956 , bott , 2 , 8 izotov y. i. , & thuan t. x. 1999 , apj , 511 , 639 izotov , y.i .
noeske , n.g .
guseva et al .
2004 , a&a , 415 , 27 izotov y. i. , schaerer d. , blecha a. , royer f. , guseva n. g. , & north p. , 2006a
, a&a , 459 , 71 izotov y. i. , stasiska g. , meynet g. , guseva n. g. , & thuan t. x. , 2006b , a&a , 448 , 955 james , b. l. , tsamis , y. g. , barlow , m. j. , westmoquette , m. s. , walsh , j. r. , cuisinier , f. , & exter , k. m. 2009 , mnras , 398 , 2 james , b. l. , tsamis , y. g. , & barlow , m. j. 2010 , mnras , 401 , 759 james , b. l. , tsamis , y. g. , barlow , m. j. , walsh , j. r. , & westmoquette , m. s. 2013 , mnras , 428 , 86 james , b. l. , tsamis , y. g. , walsh , j. r. , barlow , m. j. , & westmoquette , m. s. 2013 , mnras , 430 , 2097 kehrig , c. , vlchez , j. m. , snchez , s. f. , telles , e. , prez - montero , e. , & martn - gordn , d. 2008 , a&a , 477 , 813 kehrig , c. , prez - montero , e. , vlchez , j. m. , brinchmann , j. , kunth , d. , garca - benito , r. , crowther , p. a. , hernndez - fernndez , j. , durret , f. , contini , t. , fernndez - martn , a. , james , b. l. 2013 , mnras , tmp.1353 kobulnicky , h. a. & skillman , e. d. 1997 , apj , 489 , 636 kobulnicky , henry a. , skillman , evan d. , roy , jean - rene , walsh , j. r. , & rosa , michael r. 1997 , apj , 477 , 679 kobulnicky h. a. , & skillman e. d. , 1998 , apj , 497 , 601 kunth , d. , & sargent , w. l. w. 1983 , apj , 273 , 81 lagos , p. , telles , e. , & melnick , j. , 2007 , a&a , 476 , 89 lagos , p. , telles , e. , muoz - tun , c. , carrasco , e. r. , cuisinier , f. , & tenorio - tagle , g. 2009 , aj , 137 , 5068 lagos , p. , telles , e. , & carrasco , e. r. 2010 , iaus , 266 , 447 lagos p. , telles , e. , nigoche - netro , a. , & carrasco e. r. 2011 , aj , 142 , 162 lagos , p. & muoz - tun , c. 2011 , eas , 48 , 165 lagos p. , telles e. , nigoche - netro a. , & carrasco e. r. 2012 , mnras , 427 , 740 lagos , p. , papaderos , p. , gomes , j. m. , vega .
l. r. , & smith , a. v. 2013 , a&a , in prep .
lee , h. , skillman , e. d. , & venn , k. a. 2006 , apj , 642 , 813 lpez - snchez , a. r. , esteban , c. , garca - rojas , j. , peimbert , m. , & rodrguez , m. 2007 , apj , 656 , 168 lpez - snchez a. r. , & esteban c. , 2010 , a&a , 517 , 85 lpez - snchez , .
r. , mesa - delgado , a. , lpez - martn , l. , & esteban , c. 2011 , mnras , 411 , 2076 marquart , t. , fathi , k. , ostlin , g. , bergvall , n. , cumming , r. j. , & amram , p. 2007 , a&a , 474 , 9 markarian , b. e. , astrofizika , 3 , 24 , 1967 monreal - ibero , a. , vlchez , j. m. , walsh , j. r. , & muoz - tun , c. 2010 , a&a , 517 , 27 monreal - ibero , a. , walsh , j. r. , & vlchez , j. m. 2012 , a&a , 544 , 60 nava , a. , casebeer , d. , henry , r. b. c. , & jevremovic , d. 2006 , apj , 645 , 1076 kunth , d. & stlin , g. 2000 , a&arv , 10 , 1 papaderos , p. , loose , h .- h .
, thuan , t. x. , & fricke , k. j. 1996 , a&as , 120 , 207 papaderos , p. , izotov , y.i . , fricke , k.j .
, thuan , t.x . ,
guseva , n.g .
1998 , a&a , 338 , 43 papaderos p. , izotov y. i. , thuan t. x. , noeske k. g. , fricke k. j. , guseva n. g. , & green r. f. , 2002 , a&a , 393 , 461 papaderos , p. , guseva , n. g. , izotov , y. i. , & fricke , k. j. 2008 , a&a , 491 , 113 prez - montero , e. , vlchez , j. m. , cedrs , b. , hagele , g. f. , moll , m. , kehrig , c. , daz , a. i. , garca - benito , r. , & martn - gordn , d. 2011 , a&a , 532 , 141 sargent , w. l. w. & searle , l. 1970 , apj , 162 , 155 schaerer d. , 1996 , apj , 467 , 17 skillman , e. d. , kennicutt , r. c. , & hodge , p. w. 1989 , apj , 347 , 875 shirazi , m. , & brinchmann , j. 2012 , mnras , 421 , 1043 telles , e. , & terlevich , r. 1997 , mnras , 286 , 183 tenorio - tagle , g. 1996 , aj , 111 , 1641 thuan , t.x . & martin , g.e . 1981 , apj , 247 , 823 thuan t. x. , & izotov y. i. , 2005 , apjs , 161 , 240 walsh , j. r. & roy , j .-
1989 , mnras , 239 , 297 westmoquette , m. s. , james , b. , monreal - ibero , a. , & walsh , j. r. 2013 , a&a , 550 , 88 van zee l. , salzer j. j. , & haynes m. p. , 1998 ,
apj , 497 , 1 vanzi , l. , cresci , g. , telles , e. , & melnick , j. 2008 , a&a , 486 , 393 vanzi , l. , cresci , g. , sauvage , m. , & thompson , r. 2011 , a&a , 534 , 70 zwicky , i. f. 1964 , apj , 140 , 1467 zwicky , f. 1966 , apj , 143 , 192 | we review the results from our studies , and previous published work , on the spatially resolved physical properties of a sample of hii / bcd galaxies , as obtained mainly from integral - field unit spectroscopy with gemini / gmos and vlt / vimos .
we confirm that , within observational uncertainties , our sample galaxies show nearly spatially constant chemical abundances , similar to other low - mass starburst galaxies . they also show he ii @xmath04686 emission with properties being suggestive of a mix of excitation sources , with wolf - rayet stars being excluded as the primary one .
finally , in this contribution we include a list of all hii / bcd galaxies studied thus far with integral - field unit spectroscopy . |
coherent optical fiber links are the most precise technique to transfer time and frequency signals or to compare remote frequency standards , @xcite as they improve by more than four orders of magnitude the resolution of current satellite techniques @xcite .
thus , they are a key technology both for science and metrology , allowing for different outstanding applications . + presently , optical links are the only viable method for optical clocks comparison .
this is a prerequisite for an effective secondary representation of the second in the international system of units @xcite , as recommended by the confrence gnrale des poids et mesures , and for the possible redefinition of the second itself .
+ even cs fountains , presently the best realizations of the si second , could benefit from optical fiber links as an alternative to satellite techniques @xcite .
in fact , optical links would reduce by a factor 20 the measuring time needed to compare primary standards at their accuracy level .
+ moreover , optical links pave the way for a network of accurate clocks .
for instance , in europe there are about thirty atomic clocks between optical and fountain frequency standards .
this network would establish a unique facility for testing fundamental physics , relativistic geodesy and for improving global navigation satellite systems @xcite .
+ in addition , optical links could improve the synchronization in very long baseline interferometry ( vlbi ) antennas and particle accelerators @xcite .
+ most of these applications require fiber hauls of thousands of kilometers , and this is a challenge for performances , installation costs , and link characterization . +
this work presents the optical fiber link of implemented in italy , named lift ( italian link for time and frequency ) @xcite .
lift connects the italian metrological institute ( inrim ) to several italian scientific poles which would largely benefit from improved frequency references . in particular , as shown in the map in fig .
[ fig1 ] , the fiber link provides a reference signal to the institute for photonics and nanotechnologies in milan , to the institute for radio - astronomy in medicina ( bologna ) and to the university of florence - european nonlinear spectroscopy laboratory ( unifi - lens ) in florence for precision measurements in atomic and molecular physics .
lift will also connect italy to the forthcoming european fiber network .
+ the map of the coherent optical fiber link . view from north to south .
to is turin , mi milan , bo bologna , fi florence ( elaboration based on googleearth ) . ] to characterize the ultimate performance of the link in terms of residual phase noise and frequency uncertainty at the far fiber end , we need to compare the frequency of the delivered and of the original signal .
we do it by looping the link so that the signal goes from inrim to unifi - lens and then comes back to inrim , after travelling in the optical fiber .
in particular , we demonstrate that it is not necessary to use two independent fibers for this characterization .
this avoids the need of arming a separate twin link for the bare characterization purpose : the number of amplifiers is halved , and the reduction of the costs and of the infrastructure complexity benefits the effective implementation of continental long hauls .
+ different approaches have been proposed in the literature to analyse the link instability and inaccuracy @xcite .
we contribute to this discussion , also addressing some relevant issues in the use of the allan deviation estimator .
in particular , its theoretical evaluation from the observed phase noise spectrum , and the need of reducing the measurement bandwidth , as an alternative to the use of the modified allan deviation @xcite .
this is especially relevant with high phase noise or low control bandwidth , typical in long haul optical links . +
the infrastructure is based on a dedicated fiber long , with losses compensated by 9 bidirectional erbium doped fiber amplifiers ( bedfa ) . for a gain and on a bandwidth of ,
these devices exhibit an amplified spontaneous emission ( ase ) of at from the coherent carrier at , of at and of at .
this large ase is filtered on a bandwidth of by 8 telecom optical filters to prevent the amplifiers gain saturation and the onset of auto - oscillations @xcite .
table [ tab1 ] shows the amplifiers shelters location and the optical loss / gain for each fiber span , and fig .
[ fign ] shows the measured phase noise on the haul between inrim and some intermediate locations along the backbone .
+ llllll & from : & & & & + & & & /km & /db & /db + 1&inrim & turin ( city ) & 25 & -9 & + 2&turin ( city ) & santhi & 67 & -18 & 19 + 3&santhi & novara & 77 & -18 & 16 + 4&novara & lainate & 50 & -15 & 13 + 5&lainate & milan & 60 & -18 & 20 + 6&milan & piacenza & 67 & -16 & 17 + 7&piacenza & reggio emilia & 94 & -23 & 19 + 8®gio emilia & bologna & 74 & -19 & 16 + 9&bologna & rioveggio & 38 & -10 & 17 + 10&rioveggio & florence & 72 & -18 & 19 + 11&florence & unifi - lens & 18 & -7 & + & * total * & & * 642 * & * -171 * & * 156 * + the delivered optical radiation is provided by a fiber laser at , frequency - stabilized to an ultrastable fabry - perot cavity that reduces the laser linewidth to @xcite . to avoid the stability deterioration of the delivered signal due to temperature variations and mechanical stresses of the fiber ,
the link is phase - stabilized through the doppler noise cancellation technique @xcite .
the setup is shown in fig . [ fig2](a ) .
the ultrastable laser l1 is the local oscillator , sent to unifi - lens ; few milliwatts of optical power are coupled into the fiber . in the remote laboratory a part of the signal is extracted and represents the delivered ultrastable signal , the remainder is reflected back to inrim by a faraday mirror , which compensates for the fiber birefringence .
the acousto - optic fixed frequency shifter aom2 , at about , is used to distinguish the reflected signal from undesired backreflections along the fiber , that are not frequency - shifted . at inrim ,
the round - trip signal is phase - compared to the local laser on photodiode pd1 .
this beatnote contains the information about the noise added by the fiber on the round - trip .
it is used to stabilize the link with a phase - locked loop ( pll ) acting on the acousto - optic modulator aom1 , operating at about .
+ usually , the characterization of optical links is pursued by arming a second fiber in the same bundle with an independent set of amplifiers @xcite . in this work ,
we demonstrate a characterization technique that uses a single fiber and avoids the doubling of the amplifiers .
the setup is shown in fig . [ fig2](b ) .
the l1 radiation travels to unifi - lens and is reflected back to inrim , where it is extracted through the splitter s. a second independent ultrastable laser l2 works as a clean - up oscillator for the incoming signal and is phase - locked to it on a bandwidth of about by acting on the acousto - optic modulator aom3 , operating at .
the clean - up radiation travels the doubled link in the backward direction , to provide a double - round - trip signal , needed for the link stabilization .
the clean - up laser is required due to the relevant wideband noise added by the optical amplifiers and enables to achieve a signal to noise ratio of @xmath0 on a bandwidth of on the beatnote detected on pd1 , which is enough for the doubled link stabilization .
the clean - up radiation is up - shifted in frequency by with respect to the incoming one ; the large frequency separation between the signal in each of the 4 passes allows a good rejection of the single rayleigh scattering , which has a bandwidth of few kilohertz , and avoids any crosstalk between the signals .
the link performances are analysed by directly comparing l2 to l1 on the photodiode pd2 .
the pd2 beatnote is tracked by a voltage - controlled oscillator and its phase noise spectrum is shown in fig .
[ fig3 ] , both in the free running ( black line ) and in the doppler compensated link configuration ( red line ) . because the doubled link delay is @xmath1 , with @xmath2 the refractive index of the fiber , @xmath3 the speed of light in vacuum and @xmath4 , the loop bandwidth is about .
the noise of the compensated link achieves the fundamental limit ( blue line ) set by the round - trip delay @xcite and demonstrates that uncontrolled effects are not observed at this level .
we calculated the round - trip delay limit following the approach described in @xcite , where it is demonstrated that @xmath5 , being @xmath6 the noise of the fiber haul .
we calculated @xmath7 rather than @xmath8 , since in our case a single fiber is used instead of two independent ones ; a detailed derivation is given in the appendix .
+ phase noise spectrum of the free running ( black line , a ) and compensated ( red line , b ) optical fiber link , and the residual phase noise expected from theory ( blue line , c ) . ] in order to optimize the link performances the gain of each amplifier is tuned to notably reduce the double backreflections along the fiber , which would otherwise dominate the phase noise of the delivered signal . +
another detrimental effect is the loss of phase coherence ( cycles slips ) @xcite on the plls , that affects both the instability and the inaccuracy of the link .
the cycles slips rate mainly depends on the signal to noise ratio of the beatnotes . in our setup , the cycles slips are a few per hour , provided a polarization adjustment every few hours .
the phase of the beatnote on pd2 is measured with a dead - time - free digital phase recorder @xcite at the sampling rate of , equivalent to a bandwidth filter .
the cycles slips are barely visible on the phase data even when low - pass filtered at bandwidth , as shown in fig .
[ fig4 ] ( red line ) ; however if we filter the data on a bandwidth of ( black line ) their detection is quite easy .
+ particular care is devoted to the filtering process : instead of applying a simple average to the raw data , that acts like a first order low pass filter with cut - off frequency at half the reciprocal of the averaging time , we implemented a digital finite impulse response ( fir ) filter @xcite , with an out - of - bandwidth attenuation between .
the sharper filtering allows a factor 2 of improvement in the stability for the same equivalent bandwidth , thanks to a higher attenuation of the large noise bump between and , that is typical in optical links .
+ we realign the phase by subtracting an integer number of cycles , i.e. the minimum slip amplitude .
finally , the phase data are further filtered on a bandwidth ( smooth blue line in fig .
[ fig4 ] ) , decimated according to the nyquist theorem , and differentiated to obtain the instantaneous frequency . + to evaluate the frequency instability , the allan deviation estimator is used @xcite . in the literature ,
its expression is derived from the phase noise spectrum only in the cases where it is modelled by the relation @xcite : @xmath9 an analytical expression is not reported for a phase noise spectrum of the kind @xmath10 , such as the one observed in fig .
this has already been reported in optical fiber links @xcite and depends on the peculiar transfer function of the system , affected by the fiber delay @xcite .
it is important to note that in real systems , a low - pass filter with cut - off frequency @xmath11 is introduced to prevent noise divergence at high frequency .
assuming that the filter is infinitely sharp , we calculate the analytical expression of the allan variance in presence of the observed type of noise , by performing the integration : @xmath12 where @xmath13 is the averaging time and @xmath14 is the squared modulus of the allan variance transfer function @xcite .
for @xmath15 the integration leads to : @xmath16 during the link operation , we measured an allan deviation of @xmath17 on a measurement bandwidth and of @xmath18 on a measurement bandwidth .
this is in agreement with the prediction of eq .
( [ eq1b ] ) for the phase noise power spectrum observed in fig .
+ figure [ fig5 ] shows the allan deviation of the free running and compensated optical link on a measurement bandwidth , together with the stability of the compensated link on a bandwidth .
a shorter data sample with less cycle slips was chosen to evaluate the stability at measurement bandwidth , since the cycle - slips could not be efficiently realigned with this bandwidth .
the link achieves a short term instability of at in a bandwidth and an ultimate frequency resolution of in a bandwidth .
+ we evaluate the accuracy of the link at because a non - repeatable frequency offset at this level exceeds the obtained frequency resolution . in spite of this ,
the present frequency dissemination accuracy is beyond the most challenging requirements of any application and exceeds the world best oscillators @xcite .
+ the frequency instability , expressed as the allan deviation , of the 1284 km optical link in free running ( a , black ) and compensated ( b , red ) operation with 1 hz measurement bandwidth , and the allan deviation of the compensated link with 5 mhz measurement bandwidth ( c , blue ) .
data have been decimated according to the nyquist theorem . ]
this work describes the realization and characterization of a coherent optical fiber link for frequency transfer over a haul .
the characterization is pursued by doubling the link length to , so that both ends are in the same laboratory .
the loop has been established using a single fiber ; the expected delay - unsuppressed noise is only slightly affected with respect to the value calculated for independent fibers @xcite .
we avoided the arming of a second link ; this could be helpful , especially in view of long fiber hauls , as the number of optical amplifiers is considerably reduced .
+ in the data analysis , we addressed the use of the allan deviation associated to narrow - bandwidth filtering , which allows a resolution equivalent to that obtained using other statistical estimators @xcite .
+ on the doubled link , we obtain a short term allan deviation of at on the bandwidth of and an ultimate accuracy on the frequency transfer of at integration time .
this is also an upper limit for the real dissemination over the link .
+ the present infrastructure is now being used for the delivery of a high - accuracy frequency reference at lens , and will benefit the atomic physics experiments which are performed in this laboratory .
the haul will soon be upgraded to perform the frequency dissemination to the institute of photonics and nanotechnologies in milan and to the medicina radio - telescopes , near bologna , for the vlbi antennas synchronization .
extraction topologies such as those reported in @xcite are being considered as well as alternative techniques for the remote frequency comparisons @xcite , and raman optical amplification @xcite is being investigated as an alternative to the present architecture based on bedfa .
we thank gesine grosche and paul - eric pottie for technical help , giorgio santarelli for useful discussions , and the garr consortium for technical help with the fibers .
this work was supported by : the italian ministry of research miur under the progetti premiali programme and the prin09 - 2009zjjblx project ; the european metrology research programme ( emrp ) under sib-02 neat - ft .
the emrp is jointly funded by the emrp participating countries within euramet and the european union .
this appendix demonstrates the formula for the delay - unsuppressed noise on a link with both ends in the same laboratory , and where the same fiber is used to loop the link , instead of two independent fibers .
let us write the expression for the round - trip optical phase @xmath19 and for the forward signal optical phase @xmath20 , as a function of the fiber noise @xmath21 at time @xmath22 and position @xmath23 along the fiber and as a function of the phase correction needed for the link stabilization @xmath24 .
+ @xmath25 where @xmath26 is the loop length , in our case @xmath4 , and @xmath27 is the link delay . for the sake of clarity , we integrate the length only between @xmath28 and @xmath29 , as the two fiber halves are indeed the same fiber travelled in opposite directions .
+ let us now assume that the fiber perturbations evolve linearly with time ; this is justified for perturbations which act on timescales much longer than @xmath30 , as in the case of interest in this context . within this approximation , eq .
[ eqap1 ] is simplified into @xmath31 now , considering that in the closed feedback loop configuration @xmath32 , eq . [ eqap2 ] is rewritten as @xmath33 in the last equation , the evolution of the fiber noise is expressed as a function of its time derivative . for the fundamental theorem of the signal analysis ,
@xcite the noise power spectrum of the output of a linear and time - invariant system can be written in terms of the noise of the input ; in our case , this theorem can be applied to each fiber segment separately , i.e. : @xmath34 where @xmath35 is the contribution of a fiber segment with length @xmath36 to the compensated forward signal phase noise , @xmath37 and @xmath38 is the phase noise power spectrum of each fiber segment . assuming that the contributions of each fiber segment are independent
, we can perform the integration and end up with the stated result that @xmath39 where @xmath6 is the phase nosie of the -long link , and it has been used the relation @xmath40 99 k. predehl , g. grosche , s. m. f. raupach , s. droste , o. terra , j. alnis , th .
legero , t .
w. hnsch , th .
udem , r. holzwarth , h. schnatz , science * 336 * , 441 ( 2012 ) .
o. lopez , a. haboucha , f. kflian , h. jiang , b. chanteau , v. roncin , c. chardonnet , a. amy - klein , g. santarelli , opt . express * 18 * , 16849 ( 2010 ) .
s. droste , f. ozimek , th .
udem , k. predehl , t.w .
hansch , h. schnatz , g. grosche , and r. holzwarth , phys .
111 * , 110801 ( 2013 ) . o. lopez , a. kanj , p. pottie , d. rovera , j. achkar , c. chardonnet , a. amy - klein , g. santarelli , appl .
b * 110 * , 3 ( 2013 ) .
l. sliwczyski , p. krehlik , a. czubla , l. buczek , m. lipiski , metrologia * 50 * , 133 ( 2013 ) .
b. wang , c. gao , w.l .
chen , j. miao , x. zhu , y. bai , j.w .
zhang , y.y .
feng , t.c .
wang , sci .
* 2 * , 556 ( 2012 ) .
g. marra , r. slavik , h. s. margolis , s. n. lea , p. petropoulos , d. j. richardson , p. gill , opt
. lett . * 36 * , 511 ( 2011 ) .
m. fujieda , m. kumagai , s. nagano , a. yamaguchi , h. hachisu , and t. ido , opt . expr . * 19 * , 16498 ( 2011 ) s - c .
ebenhag , p. o. hedekvist , p. jarlemark , r. emardson , k. jaldehag , c. rieck , p. lthberg , ieee trans .
instrum meas .
* 59 * , 1918 ( 2010 ) .
hong , m. musha , m. takamoto , h. inaba , s. yanagimachi , a. takamizawa , k. watabe , t. ikegami , m. imae , y. fujii , m. amemiya , k. nakagawa , k. ueda , h. katori , opt .
lett . * 34 * , 692 ( 2009 ) .
j. vojtech , v. smotlacha , p. skoda , a. kuna , m. hula , s. sima , in _ proceedings of spie 8516 , remote sensing system engineering iv _ , san diego , california , 2012 , p. 85160h .
a. bauch , j .
achkar , s .
bize , d. calonico , r .
hlavac , l .
lorini , t .
parker , g. petit , d. piester , k. szymaniec , p. uhrich , metrologia * 43 * , 109 ( 2006 ) .
resolution 9 of the 23rd conference generale des poids et des mesures , in _ comptes rendus
de la 23e cgpm _ , 2007 , p. 431 . c. w. chou , d. b .
rosenband , d. j .
wineland , science * 329 * , 1630 ( 2010 ) .
j. mller , m. soffel .
s. a. klioner , j. geod .
* 82 * , 133 , ( 2008 ) .
g. cerretto , n. guyennon , i. sesia , p. tavella , f. gonzalez , j. hahn , v. fernandez , a. mozo , in _ proceedings of the european frequency and time forum _ , toulouse , france , 2008 .
y. he , b. j. orr , k. g. h. baldwin , m. j. wouters , a. n. luiten , g. aben , r. b. warrington , opt .
* 21 * , 18754 ( 2013 ) .
j. f. cliche , b. shillue , ieee contr .
syst . mag .
* 26 * , 19 ( 2006 ) .
kim , j. a. cox , j. chen and f. x. krtner , nature photonics * 2 * , 733 ( 2008 ) .
f. levi , r. ambrosini , d. calonico , c. e. calosso , c. clivati , g. a. costanzo , p. de natale , d. mazzotti , m. frittelli , g. galzerano , a. mura , d. sutyrin , g. m. tino , m. e. zucco , n. poli , proceedings of the joint uffc , eftf and pfm symposium ( 2013 ) , p.477 - 480 . c. delisle , j. conradi , j. lightwave tech . * 15 * , 749 - 757 ( 1997 ) . c. clivati , d. calonico , c. e. calosso , g. a. costanzo , f. levi , a. mura , a. godone , ieee trans .
. contr . * 58 * , 2582 ( 2011 ) .
ma , p. jungner , j. ye , j. l. hall , opt lett . * 19 * , 1777 ( 1994 ) .
a. williams , w. c. swann , and n. r. newbury , j. opt .
b * 25 * , 1284 ( 2008 ) .
udem , j. reichert , t. w. hnsch , m. kourogi , opt .
* 23 * , 1387 ( 1998 ) .
g. kramer and w. klische , in _ proceedings of the 2001 ieee international frequency control symposium and pda exhibition _
, seattle , washington , 2001 , p. 144 .
s. k. mitra , _ digital signal processing _
( mcgraw - hill , 2006 ) .
d. allan , in _ proceedings of the ieee _ * 54 * , 1966 , p. 221 .
e. rubiola , _ phase noise and frequency stability in oscillators _
( cambridge university press , 2009 ) . b. j. bloom , t. l. nicholson , j. r. williams , s. l. campbell , m. bishof , x. zhang , w. zhang , s. l. bromley , j. ye , nature * 506 * , 7175 ( 2014 ) .
i. ushijima , m. takamoto , m. das , t. ohkubo , h. katori , http://arxiv.org/abs/1405.4071 a. bercy , s. guellati - khelifa , f. stefani , g. santarelli , c. chardonnet , p - e .
pottie , o. lopez , a. amy - klein , journ . opt .
b * 31 * , 698 ( 2014 ) .
s. w. schediwy , gozzard , k. g. h. baldwin , b. j. orr , r. b. warrington , g. aben , a. n. luiten , opt . lett . *
38 * , 2893 ( 2013 ) . c. e. calosso , e. bertacco , d. calonico , c. clivati , g. a. costanzo , m. frittelli , f. levi , a. mura , a. godone , opt
. lett . * 39 * , 11771180 ( 2014 ) . c. clivati , g. bolognini , d. calonico , s. faralli , f. levi , a. mura , and n. poli , photon .
. lett . * 25 * , 1711 ( 2013 ) .
a. papoulis , international student edition , mcgraw hill and kogakusha company , ltd .
1965 , p. 347 | to significantly improve the frequency references used in radio - astronomy and precision measurements in atomic physics , we provide frequency dissemination through a coherent optical fiber link , that will be also part of a forthcoming european network of optical links .
on the frequency transfer , we obtained a frequency instability of at in terms of allan deviation on a measurement bandwidth , and an accuracy of .
the ultimate link performance has been evaluated by doubling the link to , demonstrating a new characterization technique based on the double round - trip on a single fiber .
the arming of a second fiber is avoided : this is beneficial to long hauls realizations in view of a continental fiber network for frequency and time metrology . the observed noise power spectrum is seldom found in the literature ; hence
, the expression of the allan deviation is theoretically derived and the results confirm the expectations . |
the existence of warped accretion discs , in which the orbital plane of the gas varies slowly with radius and possibly with time , is suggested by both observational evidence and theoretical reasoning . of the observational results ,
perhaps the best examples are certain x - ray binary stars , including her x-1 , in which the x - ray source appears to be periodically occulted by a precessing warped disc ( e.g. @xcite ; @xcite ) , and a few active galactic nuclei , including ngc 4258 , in which the warped shape of the disc is revealed by maser emission ( e.g. @xcite ; @xcite ) .
theoretical arguments indicate that a fluid disc will be warped by an external torque if it is misaligned with the equatorial plane of a spinning black hole or magnetized star at its centre ( @xcite ; @xcite ; @xcite ) , or with the orbital plane of a binary companion , planet or other satellite ( @xcite ) . even in an initially aligned system ,
the disc may develop a warp through a linear instability of the coplanar state , depending on tidal , radiation or magnetic forces ( @xcite ; @xcite ; @xcite ) .
interpreting the behaviour of warped discs is an important and challenging problem in astrophysical fluid dynamics .
the basic aim of theoretical approaches in this subject has been to understand how the shape of the disc evolves under the action of internal stresses and external torques , and to derive practical one - dimensional equations that describe this evolution . when the warp is sufficiently small the governing equations are linear and can be obtained from a perturbation analysis of a flat disc . @xcite and @xcite derived linearized equations for warped discs , superseding the pioneering but flawed analyses of @xcite and @xcite , and demonstrated that there are two basic dynamical regimes for warped keplerian discs in linear theory .
let the disc have angular velocity @xmath3 and vertical scale - height @xmath4 , and let @xmath5 denote the dimensionless effective viscosity parameter of @xcite .
to a first approximation , if @xmath6 the warp satisfies a diffusion - type equation with an effective diffusion coefficient @xmath7 , larger by a factor of @xmath8 than the effective kinematic viscosity of the disc . on the other hand ,
if @xmath9 the warp satisfies a wave - type equation and propagates with speed @xmath10 .
the rapid diffusive and non - dispersive wavelike behaviours are unique to keplerian discs as they result from a resonance between the orbital and epicyclic frequencies that couples the vertical motion associated with the warp to shearing horizontal epicyclic motions .
a possible limitation of these approaches is that the eulerian perturbation analysis on which the linearized equations are based requires that the vertical displacement be small compared to @xmath11 , and is therefore formally invalid for any observable warp .
it is appreciated that the linear theories may be valid for larger amplitudes but this can only be demonstrated using a lagrangian or semi - lagrangian method .
more recently it has also become possible to test these theories and to study aspects of warped discs using numerical simulations ( e.g. @xcite ; @xcite ) .
such simulations have focused mainly on the wavelike regime @xmath9 and have found reasonable agreement with @xcite for small - amplitude warps .
there are significant discrepancies , however , which suggest that nonlinear and dispersive effects can be important .
a particular concern has been that the shearing epicyclic motions associated with the warp in either regime are predicted to be very fast , comparable to or larger than the sound speed , for observable warps .
shocks may form if the warp has a short enough wavelength that these motions collide @xcite , but the horizontal motions may also be parametrically unstable and decay into smaller - scale inertial waves ( @xcite ; @xcite ) .
numerical simulations indicate that the shearing motions are also damped by magnetohydrodynamic turbulence in a quasi - viscous manner @xcite .
further studies of all these effects are needed to assess their importance .
in earlier work ( @xcite ; @xcite ) i derived a fully nonlinear theory of warped discs with effective viscosity , which agrees with @xcite in the appropriate limit and also bears a formal resemblance to the simplified nonlinear equations adopted by @xcite .
however , the theory of @xcite does not describe the regime of propagating bending waves in keplerian discs with @xmath12 .
it was noted there that the nonlinear dynamics of the wavelike regime is likely to be very different , and the purpose of the present paper is to investigate this important case . in section [ s :
leading ] of this paper i review the linear theory of long - wavelength bending waves in keplerian discs and introduce the detailed derivation that follows in section [ s : derivation ] .
the nonlinear evolutionary equation is analysed and solved numerically in section [ s : analysis ] and conclusions are presented in section [ s : conclusions ] .
the linearized equations for long - wavelength bending waves in a thin , inviscid , keplerian disc are ( e.g. @xcite ) @xmath13 @xmath14 where @xmath15 and @xmath16 are complex variables describing the warp and the internal torque .
specifically @xmath17 encodes the horizontal cartesian components of the unit vector @xmath18 normal to the plane containing an annulus of the disc at radius @xmath19 and time @xmath20 , while @xmath21 does the same for the horizontal internal torque in the disc ( divided by @xmath22 ) .
also @xmath3 is the angular velocity , @xmath23 is the surface density and @xmath4 is an effective density scale - height , defined by @xmath24 where @xmath25 is the central mass , @xmath26 is the density and @xmath27 is the vertical coordinate in the unperturbed disc . for a vertically isothermal disc @xmath11 is the usual gaussian scale - height . when @xmath28 is eliminated we obtain @xmath29 which implies that linear bending waves propagate non - dispersively at speed @xmath10 @xcite . if the model of the disc is such that the product @xmath30 is independent of @xmath19 , we obtain the classical wave equation @xmath31 in the conveniently rescaled spatial coordinate @xmath32 with plausible applications and numerical simulations in mind , we consider the case @xmath33 ; then @xmath34 and @xmath35 an outwardly propagating bending wave in this theory has the simple form @xmath36 where @xmath37 is related to the radial velocity amplitude through @xmath38 the function @xmath39 is unconstrained except by the initial conditions , and the wave propagates without change of form .
this long - wavelength theory of bending waves is subject to small corrections depending on the quantity @xmath40 , where @xmath41 is the radial wavenumber .
@xcite analysed the linear adiabatic wave modes in a vertically isothermal accretion disc in the case @xmath42 .
the bending wave of interest corresponds to @xmath43 and @xmath44 in their notation , and its dispersion relation in a keplerian disc can be expanded in the form @xmath45 where @xmath0 is the adiabatic index .
the first term on the right - hand side corresponds to the approximation of non - dispersive propagation as obtained in the long - wavelength bending - wave theory , while the second term indicates the primary dispersive correction .
the dispersion depends on the compressibility of the fluid and is a weak effect if the wavelength is long compared to @xmath11 .
nonlinearity will also modify the above theory .
if the nonlinearity is weak and of the same order of magnitude as the linear dispersion , we may expect a similar outwardly propagating wave solution to equation ( [ wave ] ) to exist , except that @xmath39 will no longer be an arbitrary function determined only by the initial conditions , but will evolve according to a nonlinear evolutionary equation . the situation is analogous to the classical problem of long water waves , where weak nonlinearity and dispersion combine in the korteweg de vries ( kdv ) equation ( e.g. @xcite ) . in that case
nonlinearity and dispersion have competing tendencies and solitary waves ( specifically , solitons ) are found .
another well known example is the nonlinear schrdinger ( nls ) equation , which arises in diverse fields including nonlinear optics .
if the sign of the nonlinear term is such as to steepen waves , the nls equation also admits solitons .
the kdv and nls equations are recognised as generic nonlinear equations that arise in many applications . since the bending wave propagates through an inhomogeneous medium
, it is not expected in general that its evolutionary equation will have constant coefficients .
however , it is found below that for the self - similar disc model in which @xmath46 and @xmath34 , which has reasonable scalings and is well suited for numerical simulations , the coefficients of the equation can indeed be made constant by an appropriate choice of coordinates .
the equations governing the dynamics of an ideal , compressible fluid are the equation of mass conservation , @xmath47 the adiabatic condition , @xmath48 and the equation of motion , @xmath49 where @xmath50 is the velocity , @xmath51 is the lagrangian time - derivative , @xmath52 is the pressure and @xmath53 is the gravitational potential . for a polytropic gas @xmath0 is a constant . consider a non - self - gravitating , keplerian disc around a point mass @xmath25 .
the time - dependent warping of the disc is best described in a nonlinear regime using the warped spherical polar coordinate system @xmath54 introduced by @xcite . in this scheme
the warped mid - plane of the disc is defined by the coordinate surface @xmath55 , and the warp consists of a tilt angle @xmath56 together with a twist angle @xmath57 .
the cartesian components of the unit tilt vector are given by @xmath58 . when expressed in warped coordinates ,
the governing equations become @xmath59 , \label{drho}\ ] ] @xmath60 , \label{dp}\ ] ] @xmath61 @xmath62= -{{1}\over{\rho r}}\partial_\theta p , \label{dut}\ ] ] @xmath63= -{{1}\over{\rho r\sin\theta}}\partial_\phi p , \label{dup}\ ] ] where @xmath64 are the components of velocity relative to the moving coordinate system , @xmath65 is the lagrangian time - derivative , @xmath66\partial_\theta-\left[-(\partial_r\beta)\cos\theta\sin\phi+(\partial_r\gamma ) ( \cos\beta\sin\theta+\sin\beta\cos\theta\cos\phi)\right ] { { 1}\over{\sin\theta}}\partial_\phi\ ] ] is a modified radial derivative , and @xmath67 are the absolute velocity components @xmath68 @xmath69 @xmath70 these equations involve no further approximation and are valid for arbitrary warps . as in @xcite ,
meaningful dynamical equations for @xmath71 and @xmath72 can be obtained only if the disc is thin and is defined to lie close to the surface @xmath55 .
this constraint is implied by the asymptotic analysis that follows .
we utilize the small parameter @xmath73 , such that the angular semi - thickness of the disc is @xmath74 .
the equations are to be expanded in a region of the @xmath75 plane that corresponds to a neighbourhood of the line @xmath76 followed by the nominal centre of an outwardly propagating bending wave at leading order .
( the problem of an inwardly propagating wave can be considered using the time - reversal symmetry of the problem . )
we therefore introduce the scaled variables @xmath77 , @xmath78 and @xmath79 to resolve the vertical structure of the disc and the evolution of the wave .
these are defined by @xmath80 and the equations are to be valid where @xmath77 , @xmath78 and @xmath79 are @xmath81 . thus @xmath77 is a scaled vertical coordinate relative to the mid - plane of the thin disc , @xmath78 is a scaled horizontal coordinate relative to the nominal centre @xmath76 of the travelling wave , and @xmath79 is a scaled time coordinate that follows the solution over the time - scale on which the radial location of the centre of the wave changes by a factor of order unity . the scaling of @xmath78 means that at any one time the wave is described within a region of radial extent comparable to the geometric mean of @xmath19 and @xmath11 ( figure 1 ) .
partial derivatives transform according to @xmath82 the radius @xmath19 can be expressed in terms of @xmath78 and @xmath79 and expanded in the form @xmath83=r_0(\tau)+\delta r_1(\xi,\tau)+\delta^2r_2(\xi,\tau)+\cdots,\ ] ] and this allows any function of @xmath19 to be expanded similarly .
in particular , the keplerian angular velocity is @xmath84=\omega_0(\tau)+\delta\omega_1(\xi,\tau)+\delta^2\omega_2(\xi,\tau)+\cdots.\ ] ] we then propose the asymptotic expansions @xmath85 , \label{beta_exp}\ ] ] @xmath86 @xmath87,\ ] ] @xmath88,\ ] ] @xmath89,\ ] ] @xmath90,\ ] ] @xmath91
. \label{v_phi_exp}\ ] ] here @xmath92 and @xmath93 are the density and pressure of the unperturbed disc in hydrostatic equilibrium , which satisfy @xmath94 when expressed in terms of @xmath78 and @xmath79 , these quantities have expansions @xmath95 @xmath96 and equation ( [ hydrostatic_global ] ) is satisfied at every order in @xmath97 , in particular @xmath98 the scaling of @xmath71 implies that the tilt angle of the disc is @xmath99 and therefore comparable to @xmath100 , while the scaling of @xmath101 implies that the radial velocity is comparable to the sound speed
. it is found below that , with this choice of scalings , the effect of nonlinearity in the solution is of comparable magnitude to the effect of linear dispersion .
the indexing of the variables in expansions ( [ beta_exp])([v_phi_exp ] ) is designed so as to give a certain order to the equations that are deduced below .
the parameter @xmath102 is unspecified and drops out of the analysis .
when these expansions are substituted into equations ( [ drho])([dup ] ) , we obtain a number of equations to be considered in turn .
equation ( [ dur ] ) at leading order [ @xmath81 ] is satisfied , as the keplerian rotation balances the gravitational force .
equation ( [ dut ] ) at leading order [ @xmath99 ] is satisfied by virtue of the vertical hydrostatic equilibrium , equation ( [ hydrostatic ] ) .
equation ( [ hydrostatic_global ] ) can be solved if the relation between pressure and density is specified . in order to make analytical progress we consider an isothermal model for the vertical structure , such that @xmath105 @xmath106 where @xmath107 and @xmath108 are the ( scaled ) density and pressure on the mid - plane . in order to satisfy equation ( [ hydrostatic_global ] ) , these quantities are related by @xmath109 the surface density of the unperturbed disc is @xmath110,\ ] ] with @xmath111 in order that @xmath112 , as adopted in section [ s : leading ] , we require that @xmath113 . the equations derived at higher orders have a consistently repeating formal structure that merits analysis .
the horizontal components of the equation of motion , equations ( [ dur ] ) and ( [ dup ] ) at @xmath114 , @xmath115 , involve the unknown quantities @xmath116 and @xmath117 , and have the form @xmath118 @xmath119 where the right - hand sides involve quantities known or partially known from the lower orders .
these equations may be combined to give @xmath120 where @xmath121 is the horizontal forcing combination @xmath122 the linear operator @xmath123 , subject to periodic boundary conditions in @xmath124 , has eigenvalues @xmath125 and eigenfunctions @xmath126 , where @xmath127 is any integer .
the special case @xmath43 is a null eigenfunction , or complementary function , corresponding to an epicyclic motion or eccentric distortion of the disc with an arbitrary dependence on @xmath77 .
the condition for equation ( [ horizontal_n ] ) to have a solution is @xmath128 i.e. that the forcing combination should contain no @xmath43 component .
( since @xmath121 is real there is no difference in considering @xmath43 or @xmath129 . )
the other three governing equations give rise to a vertical problem .
equation ( [ drho ] ) at @xmath130 , ( [ dp ] ) at @xmath131 and ( [ dut ] ) at @xmath132 , @xmath115 , involve the unknown quantities @xmath133 , @xmath134 and @xmath135 , and have the form @xmath136 @xmath137 @xmath138 the quantities @xmath133 and @xmath134 may be eliminated , using the hydrostatic condition ( [ hydrostatic ] ) , to obtain @xmath139 where @xmath140 is a linear operator defined by @xmath141 and @xmath142 is the vertical forcing combination @xmath143 we define the eigenvalues @xmath144 and eigenfunctions @xmath145 of @xmath140 as the solutions of the equation @xmath146 subject to the conditions that @xmath145 have periodicity @xmath22 in @xmath124 and be regular at the surfaces of the disc where the density and pressure vanish . with these boundary conditions @xmath140 is self - adjoint with weight function @xmath147 .
a vertically isothermal disc has no definite surface and the vertical boundary condition is instead that the wave energy flux tend to zero as @xmath148 .
the eigenfunctions and eigenvalues are then @xmath149 where @xmath150 and @xmath127 are integers , and @xmath151 denotes an hermite polynomial .. ] we will require only the first four , @xmath152 , @xmath153 , @xmath154 and @xmath155 .
the special case @xmath156 , @xmath157 is a null eigenfunction corresponding to a rigid tilt of the annulus at radius @xmath19 . for general ( irrational )
@xmath0 this is the only null eigenfunction .
the corresponding solvability condition on equation ( [ lvtn ] ) is @xmath158 the forcing terms required for the development of the solution to the desired order are listed in appendix a. our objective is to determine the spatiotemporal devlopment of the warp , and we must therefore obtain equations for @xmath159 and @xmath160 in terms of @xmath161 , @xmath162 , @xmath163 , etc .
this is done by solving the horizontal and vertical problems in turn up to the required order and extracting the relevant solvability conditions .
although straightforward in principle , this procedure involves arduous algebraic manipulations and was carried out with the aid of mathematica .
the first horizontal problem is unforced : @xmath164 .
the general solution is an epicyclic motion ( or eccentric distortion ) with an unknown dependence on position and time , @xmath165 based on our experience of the leading - order dynamics ( section [ s : leading ] ) we anticipate that in the absence of viscosity the solution for an outwardly travelling bending wave will be @xmath166 and @xmath167 ( cf . equations [ wave ] and [ ur ] ) .
this assumption simplifies the calculation , and is verified subsequently .
then we have @xmath168 the first vertical problem involves non - trivial forcing , and we find @xmath169 involving two different eigenfunctions , @xmath170 and @xmath171 , of the operator @xmath140 .
the solvability condition is satisfied , and equation ( [ lvtn ] ) has the solution @xmath172 ( we do not add any multiple of the complementary function @xmath173 , as this would amount simply to redefining the tilt @xmath174 . ) the quantities @xmath175 and @xmath176 may then be obtained from equations ( [ frhon ] ) and ( [ fpn ] ) after an integration with respect to @xmath124 , and we obtain @xmath177\zeta\right.}&\nonumber\\ & & \left.\qquad+\left({{3}\over{3-\gamma}}\right)\beta_0 ( \partial_\xi\beta_0\sin2\phi-\beta_0\partial_\xi\gamma_0\cos2\phi ) ( 1-\zeta^2)\right\},\end{aligned}\ ] ] @xmath178\zeta\right.}&\nonumber\\ & & \left.\qquad+\left({{3}\over{3-\gamma}}\right)\beta_0 ( \partial_\xi\beta_0\sin2\phi-\beta_0\partial_\xi\gamma_0\cos2\phi ) ( \gamma-\zeta^2)\right\},\end{aligned}\ ] ] where the axisymmetric parts are required to satisfy @xmath179 we may assume that @xmath180 vanishes , since it would otherwise correspond to an arbitrary redefinition of the unperturbed density of the disc
. then @xmath181 the second horizontal problem is also forced , and we find @xmath182 \left[\gamma-1+(\gamma+1)\zeta^2\right]-\left({{12}\over{3-\gamma}}\right)\beta_0 ^ 2 ( \partial_\xi\beta_0\cos3\phi+\beta_0\partial_\xi\gamma_0\sin3\phi)\zeta.\ ] ] again , the solvability condition is satisfied .
( this verifies that our choice for @xmath183 and @xmath184 was correct ; if we had not assumed their form in advance , the solvability condition at this order would have implied equation [ vh1 ] . )
equation ( [ horizontal_n ] ) then has the solution @xmath185 \left[\gamma-1+(\gamma+1)\zeta^2\right]-{{3}\over{4(3-\gamma)}}r_0\beta_0 ^ 2 ( \partial_\xi\beta_0\cos3\phi+\beta_0\partial_\xi\gamma_0\sin3\phi)\zeta \nonumber\\ & & \qquad-{{1}\over{2}}u_2(\zeta,\xi,\tau)\sin\phi+ { { 1}\over{2}}v_2(\zeta,\xi,\tau)\cos\phi.\end{aligned}\ ] ] the last two terms are the complementary functions , presently of unknown amplitude . the quantity @xmath186 may then be obtained from equation ( [ fphin ] ) , and we find @xmath187 \left[\gamma-1+(\gamma+1)\zeta^2\right]- { { 3}\over{2(3-\gamma)}}r_0\beta_0 ^ 2 ( \partial_\xi\beta_0\sin3\phi-\beta_0\partial_\xi\gamma_0\cos3\phi)\zeta \nonumber\\ & & \qquad-2r_0\beta_0 ^ 2(\partial_\xi\gamma_0)\zeta^2+\left[1-\left({{3}\over{3-\gamma}}\right)\beta_0 ^ 2\right]r_0 ( \partial_\xi\beta_0\sin\phi-\beta_0\partial_\xi\gamma_0\cos\phi)\zeta+ u_2\cos\phi+v_2\sin\phi.\end{aligned}\ ] ] the forcing for the second vertical problem is very complicated , but fortunately it is not necessary to solve it in detail .
the solvability condition alone provides an equation for the desired quantities @xmath159 and @xmath160 .
this involves much algebra , not written out here , and simplifies to @xmath188+{{\rm i}\over{2\omega_0}}{{3(\gamma-1)}\over{3-\gamma } } \left[\beta_0 ^ 2\partial_{\xi\xi}\beta_0 + 2\beta_0(\partial_\xi\beta_0)^2\right]\nonumber\\ & & -{{1}\over{\omega_0}}{{3(\gamma+1)}\over{3-\gamma } } \beta_0 ^ 2(\partial_\xi\beta_0)\partial_\xi\gamma_0-{{1}\over{2\omega_0}}\left({{\gamma+3}\over{3-\gamma}}\right ) \beta_0 ^ 3\left[\partial_{\xi\xi}\gamma_0 + { \rm i}(\partial_\xi\gamma_0)^2\right ] .
\label{scv2}\end{aligned}\ ] ] although this equation contains information about the evolution of @xmath161 and @xmath189 , it also involves the unknown quantities @xmath174 , @xmath190 , @xmath191 and @xmath192 .
further information must therefore be sought .
fortunately it is not necessary to solve the second vertical problem to obtain @xmath193 , @xmath194 and @xmath195 in detail .
the contribution of these quantities to the third horizontal problem , which is the last problem we consider , is of the form @xmath196 for the solvability condition we require only the @xmath43 component of this quantity . in principle
both the @xmath197 and @xmath198 components of @xmath193 could contribute , but the @xmath198 component gives exactly zero contribution .
now the axisymmetric part of @xmath199 is @xmath200 \left[\left({{2\gamma^2+\gamma-2}\over{\gamma}}\right)\zeta+ { { ( \gamma-1)(3\gamma+1)}\over{\gamma}}(\zeta^3 - 3\zeta)\right],\ ] ] involving the eigenfunctions @xmath201 and @xmath202 of @xmath140 , and so the axisymmetric part of @xmath193 is @xmath203 \left[{{(2\gamma^2+\gamma-2)}\over{\gamma(\gamma+1)}}\zeta+ \left({{\gamma-1}\over{\gamma}}\right)(\zeta^3 - 3\zeta)\right].\ ] ] this information is sufficient to deduce the final solvability condition in the form @xmath204=\cdots,\ ] ] where we do not write the right - hand side in full . by multiplying this equation by @xmath205 and integrating with respect to @xmath77 we
obtain @xmath206+{{\rm i}\over{2\omega_0}}\left({{3}\over{3-\gamma}}\right ) \left[\beta_0 ^ 2\partial_{\xi\xi}\beta_0 + ( 2-\gamma)\beta_0(\partial_\xi\beta_0)^2\right]\nonumber\\ & & -{{1}\over{\omega_0}}{{(15 + 6\gamma-\gamma^2)}\over{(\gamma+1)(3-\gamma ) } } \beta_0 ^ 2(\partial_\xi\beta_0)\partial_\xi\gamma_0- { { 1}\over{2\omega_0}}{{(\gamma+9)}\over{(\gamma+1)(3-\gamma ) } } \beta_0 ^ 3\left[\partial_{\xi\xi}\gamma_0 + ( \gamma+1){\rm i}(\partial_\xi\gamma_0)^2\right ] , \label{sch3}\end{aligned}\ ] ] which may be compared with equation ( [ scv2 ] ) .
the unknown quantities @xmath174 , @xmath190 , @xmath191 and @xmath192 may all be eliminated by taking the average of the two equations , yielding the desired evolutionary equation for @xmath161 and @xmath189 .
this is expressed most compactly in terms of the complex tilt variable @xmath207 : @xmath208 , \label{wtau}\ ] ] with dimensionless parameters @xmath209
@xmath210 we recall that the true complex tilt variable is @xmath211 .
equation ( [ wtau ] ) is a complex nonlinear equation describing the spatiotemporal development of the warp .
the first , linear term on the right - hand side corresponds to the linear dispersion evident in equation ( [ lp_dispersion ] ) derived from @xcite . appears because of the relations between @xmath79 and @xmath212 and between @xmath213 and @xmath19 . ]
the nonlinear terms are cubic and also depend on the compressibility of the gas .
the appearance of the factor @xmath214 in the denominator is significant and can be traced to the involvement of the mode @xmath171 , which is coupled resonantly if @xmath215 , even if the disc is not vertically isothermal @xcite . in reality
@xmath216 is the case of greatest relevance . in discussing the properties of equation ( [ wtau ] ) , however , it is instructive to contrast the cases of @xmath1 and @xmath2 .
we assume throughout that @xmath217 .
there is an explicit dependence on @xmath79 in equation ( [ wtau ] ) because the bending wave experiences a varying background as it propagates outwards through the disc .
however , this dependence can be conveniently eliminated by working with the rescaled time variable @xmath218 the reason that an equation with constant coefficients can be obtained is related to the self - similarity of the disc model .
the nonlinear terms can also be rescaled by defining @xmath219^{1/2}\psi.\ ] ] we then have @xmath220 , \label{dnls}\ ] ] where @xmath221 equation ( [ dnls ] ) is a derivative nonlinear schrdinger equation ( dnls ) .
the irreducible dimensionless parameters of the equation , @xmath222 and @xmath223 , satisfy @xmath224 and @xmath225 for @xmath226 , while @xmath227 and @xmath228 for @xmath2 .
equation ( [ dnls ] ) is invariant under translations of @xmath229 and @xmath78 and also under the ` gauge transformation ' @xmath230 , which corresponds to a trivial rotation of the coordinate system through an angle @xmath231 about the @xmath27-axis .
a further symmetry is @xmath232 , which corresponds to a ` reflection ' about the centre of the wave .
the coordinate @xmath229 , and the equation itself , are invariant under time reversal ( @xmath233 ) .
the trivial solution @xmath234 corresponds to a flat disc .
however , the equation is not invariant under @xmath235 , which might appear to correspond to an additional trivial rigid tilt of the disc .
the reason can be traced to equation ( [ wave ] ) where it is seen that the amplitude @xmath236 ( or @xmath237 ) determines the non - trivial horizontal velocities as well as the tilt , because the solution under consideration is a wave travelling in one direction .
it is readily verified that solutions exist in the form of uniform travelling waves @xmath238 , where @xmath239 and @xmath41 satisfy the nonlinear dispersion relation @xmath240 such a wave is linearly stable to long - wavelength disturbances if @xmath241\ge0,\ ] ] and to short - wavelength disturbances if @xmath242 in the rescaled variables @xmath78 and @xmath229 the ( negative ) linear dispersion coefficient is unity . since @xmath243 and @xmath244 in practice ,
the waves are stable if @xmath1 ( i.e. @xmath245 ) and the nonlinear terms increase their dispersion .
therefore it may be expected that solitary waves are not supported , and this appears to be confirmed by the following analysis . however , if @xmath2 ( i.e. @xmath246 ) , the waves are unstable except when @xmath247 , and the nonlinearity counteracts the linear dispersion .
following the usual analysis of nls solitons ( e.g. @xcite ) , we consider more general travelling - wave solutions of the form @xmath248 where @xmath239 is a frequency and @xmath249 is a wave speed to be determined .
note that @xmath249 is a slow speed in rescaled coordinates and is relative to the basic wave speed of @xmath10 .
writing @xmath250 with @xmath251 and @xmath53 real , we then find @xmath252 @xmath253\phi''+2\left[1+s(2a+b-1)r^2\right]r'\phi'=cr ' . \label{phipp}\ ] ] these equations are completely integrable .
equation ( [ phipp ] ) is linear in @xmath254 and has the solution @xmath255^{-q},\ ] ] where @xmath256 is an arbitrary constant and @xmath257 .
we have @xmath258 for @xmath226 , while @xmath259 for @xmath2 .
equation ( [ rpp ] ) may then be written in the form @xmath260 with @xmath261^{1-q}+\frac{h^2}{2r^2}\left[1+s(2a-1)r^2\right]^{1 - 2q}-\frac{1}{2}\omega r^2,\ ] ] and has the first integral @xmath262 there is an obvious mechanical analogy with a particle ( albeit of variable inertia ) moving in an effective potential . a solitary wave would have @xmath263 and @xmath264 as @xmath265 .
the analogous particle would roll in a potential well between @xmath266 and @xmath267 such that @xmath268 and @xmath269 for @xmath270 . in this problem , as @xmath263 , the effective potential diverges as @xmath271 , so this behaviour is possible only if @xmath272 . in this case @xmath273^{1-q}+\left[1+s(2a-1)r^2\right]^{1 - 2q}\right\}-\frac{1}{2}\omega
r^2,\ ] ] and @xmath274 tends to a constant as @xmath263 .
we exclude the trivial case @xmath275 .
it can then be shown that @xmath274 is a strictly concave function of @xmath276 ( i.e. @xmath277 ) when @xmath1 ( i.e. @xmath245 and @xmath278 ) .
therefore it is impossible for a potential well to be formed , and we conclude that solitary waves are not supported when @xmath1 . in contrast , when @xmath2 ( i.e. @xmath246 and @xmath279 ) , @xmath274 is a strictly convex function of @xmath276 that diverges to @xmath280 as @xmath276 approaches @xmath281 , and a potential well is always formed .
the squared amplitude @xmath282 can never exceed @xmath281 , so it appears that solitary waves are supported for small amplitudes @xmath283 and stable extended wavetrains for larger amplitudes @xmath284 .
an analytical approximation for the solitary waves in the case @xmath2 is possible when they are of small amplitude @xmath285 . in this limit @xmath286 with @xmath287 the solution of equation ( [ energy ] ) for @xmath288 is then @xmath289 and furthermore @xmath290 . for a specified height ( @xmath291 ) and width ( @xmath292 ) , the frequency @xmath239 and wave speed @xmath249
are determined by these relations .
equation ( [ dnls ] ) was solved numerically by applying a spatial discretization on a uniform grid , based on second - order centred differences , and solving the resulting coupled ordinary differential equations in time using a fifth - order runge
kutta method with adaptive stepsize .
illustrative solutions for the case @xmath216 are shown in fig . 2 .
the initial profile is a real gaussian of relatively large amplitude .
the solution of the linear schrdinger equation , obtained when the nonlinear terms are neglected , can be written in closed form and is shown in the right - hand panels ; it exhibits some linear dispersion .
when the nonlinear terms are included , as shown in the left - hand panels , the wave broadens much more rapidly and there is a greater emission of short waves .
( if their wavelength is comparable to or shorter than @xmath11 it is likely that they will behave differently from the predictions of equation [ dnls ] ) .
the computational domain is large enough that the boundary conditions do not affect the solution significantly for the ranges of @xmath78 and @xmath229 shown .
it was also confirmed numerically that small - amplitude solitary waves in the case @xmath2 propagate according to equation ( [ solitary ] ) .
it is straightforward to generalize the analysis to include a small effective viscosity .
suppose that the effective shear viscosity is given by @xmath293 , where @xmath5 is independent of @xmath27 and comparable to @xmath100 .
let @xmath294 with @xmath295 .
then the viscosity appears in the third - order horizontal forcing terms @xmath296 and @xmath297 , where its effect is to replace @xmath298 by @xmath299 .
indeed , any process that damps the shearing epicyclic motions at a rate @xmath300 would have an equivalent effect .
an effective bulk viscosity of comparable magnitude would not affect the analysis .
the final equation for @xmath236 is modified to @xmath301 where the dots indicate the inviscid dynamical terms found previously .
the dnls equation ( [ dnls ] ) is modified to @xmath302 when @xmath237 is small the effect of viscosity is to cause the solution to decay by a factor @xmath303 relative to the inviscid solution . in general , however , the nonlinearity of the equation means that it is not possible to find such an integrating factor .
we have derived a one - dimensional equation describing the evolution of weakly nonlinear and dispersive bending waves in a thin keplerian accretion disc with negligible viscosity and self - gravity .
the nonlinear and dispersive effects both depend on the compressibility of the gas through its adiabatic exponent @xmath0 .
perhaps surprisingly , in the physically realistic case @xmath1 , nonlinearity tends to broaden the waves and enhance their linear dispersion .
this is in contrast with the ( hypothetical ) case @xmath2 in which nonlinearity counteracts linear dispersion and solitary waves are supported .
linear theories of warped accretion discs ( @xcite ; @xcite ) previously made clear the distinction between the diffusive ( @xmath304 ) and wavelike ( @xmath9 ) regimes in keplerian discs .
since they were based on an eulerian perturbation analysis they could not predict the amplitude at which they would break down . in this paper
it is found that nonlinear effects cause a significant modification of the bending wave ( as the radial location of the wave changes by a factor of order unity ) when the dimensionless amplitude of the warp @xmath305 is @xmath306 and the horizontal velocities are comparable to the sound speed .
unfortunately it is difficult to explain in simple terms how these critical scalings arise or why the nonlinearity is steepening only when @xmath2 .
the nonlinearity arises through a mode - coupling process and is complicated further by the resonant nature of the epicyclic oscillations in a keplerian disc , arising from the coincidence of the orbital and epicyclic frequencies .
the analysis in this paper does not by any means present a complete description of the nonlinear dynamics of the wavelike regime .
we have focused on the evolution of a travelling bending wave and have not considered the interaction between inward and outward waves , nor the important case of steady warps maintained by an external torque .
furthermore , @xcite noted that shocks can form if the wavelength of the warp is short enough that the shearing epicyclic motions from different parts of the wave collide .
this effect typically requires @xmath305 to be @xmath81 .
@xcite also noted that the warp could decay through a parametric instability .
the critical amplitude for this effect is proportional to the effective viscosity of the disc . it would be interesting to compare the predictions of this analysis with the results of numerical simulations of warped discs .
if parametric instability sets in there should be a noticeable deviation from equation ( [ dnls ] ) . the best way to make
this comparison would be to set up a vertically isothermal disc with @xmath46 and @xmath112 , to introduce a warp and the accompanying horizontal velocities ( [ vh1 ] ) , and to follow the evolution of the complex tilt @xmath15 . the comparison with solutions of equation ( [ dnls ] ) could then be made by transforming variables from @xmath15 to @xmath307 , or vice versa .
i acknowledge the support of the aspen center for physics during the research programme ` astrophysical disks ' ( summer 2002 ) , where this work was mostly carried out , and of the kavli institute for theoretical physics , santa barbara , during the research programme ` physics of astrophysical outflows and accretion disks ' ( summer 2005 ) , where it was brought to completion . this research was supported in part by the national science foundation under grant number phy99 - 07949 .
bardeen j. m. , petterson j. a. , 1975 , apj , 195 , l65 clarkson w. i. , charles p. a. , coe m. j. , laycock s. , 2003 , mnras , 343 , 1213 gammie c. f. , goodman j. , ogilvie g. i. , 2000 , mnras , 318 , 1005 gerend d. , boynton p. e , 1976 , apj , 209 , 562 greenhill l. j. et al . , 2003 ,
apj , 590 , 162 lai d. , 1999 , apj , 524 , 1030 larwood j. d. , nelson r. p. , papaloizou j. c. b. , terquem c. , 1996 , mnras , 282 , 597 lipunov v. m. , shakura n. i. , 1980 , sov .
, 6 , 14 lubow s. h. , 1992 , apj , 398 , 525 lubow s. h. , ogilvie g. i. , 2000 , apj , 538 , 326 lubow s. h. , pringle j. e. , 1993 , apj , 409 , 360 miyoshi m. , moran j. , herrnstein j. , greenhill l. , nakai n. , diamond p. , inoue m. , 1995 , nature , 373 , 127 nelson r. p. , papaloizou j. c. b. , 1999 , mnras , 309 , 929 ogilvie g. i. , 1999 , mnras , 304 , 557 ogilvie g. i. , 2000 , mnras , 317 , 607 papaloizou j. c. b. , lin d. n. c. , 1995 , apj , 438 , 841 papaloizou j. c. b. , pringle j. e. , 1983 , mnras , 202 , 1181 papaloizou j. c. b. , terquem c. , 1995 , mnras , 274 , 987 petterson j. a. , 1978 , apj , 226 , 253 pringle j. e. , 1992 , mnras , 258 , 811 pringle j. e. , 1996 , mnras , 281 , 357 shakura n. i. , sunyaev , r. a. , 1973 , a&a , 24 , 337 torkelsson u. , ogilvie g. i. , brandenburg a. , pringle j. e. , nordlund . , stein r. f. , 2000 , mnras , 318 , 47 whitham g. b. , 1954 , linear and nonlinear waves , wiley , new york
for reference , we list here the forcing terms required for the development of the solution to the desired order . @xmath308,\ ] ] @xmath309\partial_\phi\rho_1 + { v_{\theta1}\over{r_0}}\partial_\zeta(\rho_{{\rm e}1}+\rho_1)}&\nonumber\\ & & \qquad- { { r_1v_{\theta1}}\over{r_0 ^ 2}}\partial_\zeta\rho_0+(\rho_{{\rm e}1}+\rho_1)\left\{-{{2}\over{r_0\omega_0}}[\partial_\xi v_{r1}- ( \partial_\xi\gamma_0)\partial_\phi v_{r1}]+ { { 1}\over{r_0}}\partial_\zeta v_{\theta1}\right\}\nonumber\\ & & \qquad+\rho_0\left\{-{{2}\over{r_0\omega_0}}[\partial_\xi v_{r2}- ( \partial_\xi\gamma_0)\partial_\phi v_{r2}- ( \partial_\xi\gamma_1)\partial_\phi v_{r1}]\right.\nonumber\\ & & \left.\qquad+ \left({{2\omega_1}\over{r_0\omega_0 ^ 2}}+{{2r_1}\over{r_0 ^ 2\omega_0}}\right ) [ \partial_\xi v_{r1}-(\partial_\xi\gamma_0)\partial_\phi v_{r1}]- { { 2v_{r1}}\over{r_0}}-{{r_1}\over{r_0 ^ 2}}\partial_\zeta v_{\theta1}- { { 1}\over{r_0}}\partial_\phi v_{\phi1}\right\},\end{aligned}\ ] ] @xmath310\right\},\ ] ] @xmath311\partial_\phi p_1 + { v_{\theta1}\over{r_0}}\partial_\zeta(p_{{\rm e}1}+p_1)}&\nonumber\\ & & \qquad- { { r_1v_{\theta1}}\over{r_0 ^ 2}}\partial_\zeta p_0+\gamma(p_{{\rm e}1}+p_1)\left\{-{{2}\over{r_0\omega_0}}[\partial_\xi v_{r1}- ( \partial_\xi\gamma_0)\partial_\phi v_{r1}]+ { { 1}\over{r_0}}\partial_\zeta v_{\theta1}\right\}\nonumber\\ & & \qquad+\gamma p_0\left\{-{{2}\over{r_0\omega_0}}[\partial_\xi v_{r2}- ( \partial_\xi\gamma_0)\partial_\phi v_{r2}- ( \partial_\xi\gamma_1)\partial_\phi v_{r1}]\right.\nonumber\\ & & \left.\qquad+ \left({{2\omega_1}\over{r_0\omega_0 ^ 2}}+{{2r_1}\over{r_0 ^ 2\omega_0}}\right ) [ \partial_\xi v_{r1}-(\partial_\xi\gamma_0)\partial_\phi v_{r1}]- { { 2v_{r1}}\over{r_0}}-{{r_1}\over{r_0 ^ 2}}\partial_\zeta v_{\theta1}- { { 1}\over{r_0}}\partial_\phi v_{\phi1}\right\},\end{aligned}\ ] ] @xmath312 @xmath313\partial_\phi v_{r1}+2\omega_1v_{\phi1 } } & \nonumber\\ & & \qquad-{{2}\over{\rho_0r_0\omega_0}}(\partial_\xi\beta_0\cos\phi+\beta_0\partial_\xi\gamma_0\sin\phi)\partial_\zeta p_0,\end{aligned}\ ] ] @xmath314\partial_\phi v_{r2}\nonumber\\ & & \qquad-\left[\omega_2+{{v_{\phi1}}\over{r_0}}-\partial_\tau\gamma_0 + \left(1-{{2v_{r1}}\over{r_0\omega_0}}\right)\partial_\xi\gamma_1- \left({{2v_{r2}}\over{r_0\omega_0}}-{{2v_{r1}r_1}\over{r_0 ^ 2\omega_0}}- { { 2v_{r1}\omega_1}\over{r_0\omega_0 ^ 2}}\right)\partial_\xi\gamma_0\right ] \partial_\phi v_{r1}\nonumber\\ & & \qquad+2\omega_1v_{\phi2}+2\omega_2v_{\phi1}+{{v_{\phi1}^2}\over{r_0}}- r_0\omega_0 ^ 2\zeta^2\nonumber\\ & & \qquad-{{2}\over{\rho_0r_0\omega_0}}\left\{\partial_\xi(p_{{\rm e}1}+p_1)+ ( \partial_\xi\beta_0\cos\phi+\beta_0\partial_\xi\gamma_0\sin\phi ) \partial_\zeta(p_{{\rm
e}1}+p_1)\right.\nonumber\\ & & \left.\qquad+ [ \partial_\xi\beta_1\cos\phi+(\beta_0\partial_\xi\gamma_1 + \beta_1\partial_\xi\gamma_0)\sin\phi]\partial_\zeta p_0- ( \partial_\xi\gamma_0)\partial_\phi p_1\right\}\nonumber\\ & & \qquad+2\left({{\rho_{{\rm e}1}+\rho_1}\over{\rho_0 ^ 2r_0\omega_0}}+ { { r_1}\over{\rho_0r_0 ^ 2\omega_0}}+{{\omega_1}\over{\rho_0r_0\omega_0 ^ 2}}\right)(\partial_\xi\beta_0\cos\phi+\beta_0\partial_\xi\gamma_0\sin\phi ) \partial_\zeta p_0,\end{aligned}\ ] ] @xmath315 @xmath316\partial_\phi v_{\theta1}}&\nonumber\\ & & \qquad-\left({{r_1}\over{\rho_0r_0 ^ 2}}+{{\rho_{{\rm e}1}+\rho_1}\over{\rho_0 ^ 2r_0}}\right ) \partial_\zeta p_1+{{\rho_1 ^ 2}\over{\rho_0 ^ 3r_0}}\partial_\zeta p_0 + 2\omega_0v_{\phi1}\zeta\nonumber\\ & & \qquad+2(2v_{r1}-r_0\omega_0)[\partial_\xi\beta_1\sin\phi- ( \beta_0\partial_\xi\gamma_1+\beta_1\partial_\xi\gamma_0)\cos\phi]\nonumber\\ & & \qquad+2(2v_{r2}-r_0\omega_1-r_1\omega_0)(\partial_\xi\beta_0\sin\phi- \beta_0\partial_\xi\gamma_0\cos\phi)\nonumber\\ & & \qquad-2(\partial_\phi v_{r1 } ) [ \partial_\xi\beta_1\cos\phi+(\beta_0\partial_\xi\gamma_1 + \beta_1\partial_\xi\gamma_0)\sin\phi]+2r_0\omega_0(\partial_\tau\beta_0\sin\phi- \beta_0\partial_\tau\gamma_0\cos\phi)\nonumber\\ & & \qquad-{{2}\over{\omega_0}}\left\{\omega_0\partial_\phi v_{r2}- \left(1-{{2v_{r1}}\over{r_0\omega_0}}\right)[\partial_\xi v_{r1}- ( \partial_\xi\gamma_0)\partial_\phi v_{r1}]- { { v_{\theta1}}\over{r_0}}\partial_\zeta v_{r1}\right\ } ( \partial_\xi\beta_0\cos\phi+\beta_0\partial_\xi\gamma_0\sin\phi)\nonumber\\ & & \qquad- r_0\left(1-{{2v_{r1}}\over{r_0\omega_0}}\right)^2 \{[\partial_{\xi\xi}\beta_0-\beta_0(\partial_\xi\gamma_0)^2]\cos\phi+ [ \beta_0\partial_{\xi\xi}\gamma_0 + 2(\partial_\xi\beta_0)\partial_\xi\gamma_0]\sin\phi\},\end{aligned}\ ] ] @xmath317 @xmath318\partial_\phi v_{\phi1}+ { { \omega_1v_{r1}}\over{2}}+{{3\omega_0r_1}\over{2r_0}}v_{r1},\ ] ] @xmath319\partial_\phi v_{\phi2}\nonumber\\ & & \qquad-\left[\omega_2+{{v_{\phi1}}\over{r_0}}-\partial_\tau\gamma_0 + \left(1-{{2v_{r1}}\over{r_0\omega_0}}\right)\partial_\xi\gamma_1- \left({{2v_{r2}}\over{r_0\omega_0}}-{{2v_{r1}r_1}\over{r_0 ^ 2\omega_0}}- { { 2v_{r1}\omega_1}\over{r_0\omega_0 ^ 2}}\right)\partial_\xi\gamma_0\right ] \partial_\phi v_{\phi1}\nonumber\\ & & \qquad+{{1}\over{2}}\omega_1v_{r2}+2\omega_2v_{r1}- { { v_{r1}v_{\phi1}}\over{r_0}}+ { { 3}\over{2}}\left[{{\omega_0r_1}\over{r_0}}v_{r2}+ \left({{2\omega_0r_2}\over{r_0}}-{{\omega_1 ^ 2}\over{\omega_0}}- { { \omega_0r_1 ^ 2}\over{r_0 ^ 2}}\right)v_{r1}\right].\end{aligned}\ ] ] | the nonlinear dynamics of a warped accretion disc is investigated in the important case of a thin keplerian disc with negligible viscosity and self - gravity .
a one - dimensional evolutionary equation is formally derived that describes the primary nonlinear and dispersive effects on propagating bending waves other than parametric instabilities .
it has the form of a derivative nonlinear schrdinger equation with coefficients that are obtained explicitly for a particular model of a disc .
the properties of this equation are analysed in some detail and illustrative numerical solutions are presented .
the nonlinear and dispersive effects both depend on the compressibility of the gas through its adiabatic index @xmath0 . in the physically realistic case @xmath1 ,
nonlinearity does not lead to the steepening of bending waves but instead enhances their linear dispersion . in the opposite case @xmath2
, nonlinearity leads to wave steepening and solitary waves are supported .
the effects of a small effective viscosity , which may suppress parametric instabilities , are also considered .
this analysis may provide a useful point of comparison between theory and numerical simulations of warped accretion discs .
[ firstpage ] accretion , accretion discs hydrodynamics waves . |
the large - scale spatial distribution of dark matter , as revealed from numerical simulations , shows a characteristic anisotropic web - like structure .
this _ cosmic web _ , which arises through the gravitational clustering of matter , is mainly due to the effects of the tidal field : in fact , the competition between cosmic expansion , the trace , and the traceless part of the tidal field imprints anisotropies in the large - scale matter distribution in much the same way that gravity and radiation pressure imprints baryonic acoustic oscillations ( bao ) on the cosmic microwave background ( cmb ) sky ( hu & sugiyama 1995 ; lee & springel 2010 ) . hence , the initial shear field plays a crucial role in the formation of large scale structures , and a number of studies in the literature have been devoted to this subject among the plethora of papers , see for example the classic works by zeldovich ( 1970 ) , icke ( 1973 ) , peebles ( 1980 ) , white ( 1984 ) , bardeen et al .
( 1986 ) , kaiser ( 1986 ) , bertschinger ( 1987 ) , bond & myers ( 1996 ) , bond , kofman & pogosyan ( 1996 ) and van de weygaert & bertschinger ( 1996 ) .
in addition , if the cosmic web originates from primordial tidal effects and its degree of anisotropy increases with the evolution of the universe , then studying this initial field is crucial in understanding the subsequent nonlinear evolution of cosmic structures ( springel et al . 2005 ; shandarin et al .
2006 ; desjacques 2008 ; desjacques & smith 2008 ; pogosyan et al . 2009 ) , the alignment of shape and angular momentum of halos ( west 1989 ; catelan et al . 2001
; lee & springel 2010 ; rossi , sheth & tormen 2011 ) , the statistical properties of voids ( lee & park 2006 ; platen , van de weygaert & jones 2008 ) , and more generally for characterizing the geometry and morphology of the cosmic web ( shen et al .
2006 ; van de weygaert & bond 2008 ; forero - romero et al .
2009 ; aragon - calvo et al .
2010a , b ; shandarin et al . 2010 ) .
the basic theory for the formation and evolution of structures is now well - understood , thanks to the pioneering work of doroshkevich & zeldovich ( 1964 ) , doroshkevich ( 1970 ) , zeldovich ( 1970 ) , and sunyaev & zeldovich ( 1972 ) the latter in the context of galaxy formation .
in particular , doroshkevich ( 1970 ) derived the joint probability distribution of an ordered set of eigenvalues in the tidal field matrix at random positions given the variance of the density field , corresponding to a gaussian potential ; we will refer to it as the _ unconditional _ probability distribution of eigenvalues .
in addition , zeldovich ( 1970 ) provided the fundamental understanding of anisotropic collapse on cosmological scales , and recognized the key role of the large scale tidal force in shaping the cosmic web .
subsequently , doroshkevich and shandarin ( 1978 ) calculated some statistical properties of the maxima of the largest eigenvalue of the shear tensor .
their study proved that the most probable formation process starts first with a one - dimensional collapse ( cosmic pancake formation ) .
the directions ( orientations ) for the one - dimensional collapsed sheets are determined by the largest eigenvalue of the deformation tensor , which can be attributed to the initial linear density perturbations ; moreover , the probability that two or even three of the initial eigenvalues are identical or nearly equal is extremely small , indicating that the collapse is triaxial .
later on , following a constrained field approach pioneered by bertschinger ( 1987 ) , van de weygaert & bertschinger ( 1996 ) developed an algorithm for setting up tailor - made initial conditions for cosmological simulations , which addressed the role of the tidal fields in shaping large - scale structures .
in the same period , bond , kofman & pogosyan ( 1996 ) developed a cosmic web theory which naturally explains the filamentary structure present in the cold dark matter ( cdm ) cosmology , due to the coherent nature of the primordial tidal field .
they realized that an `` embryonic '' cosmic web is already present in the primordial density field , and explained why in overdense regions sheet - like membranes are only marginal features . since then ,
because of the correspondence between structures in the evolved density field and local properties of the linear tidal field pointed out by the same authors , the statistics of the shear has received more attention in the literature .
for example , lee & shandarin ( 1998 ) computed some probability distributions for individual shear eigenvalues and obtained an analytic approximation to the halo mass function , and catelan & porciani ( 2001 ) explored the two - point correlation of the tidal shear components . however , a variety of studies in the physics of halo formation and cosmic web classification are based on doroshkevich s unconditional formulae for the ordered eigenvalues of the initial shear field associated with gaussian statistics ( doroshkevich 1970 ) , but those formulas can not differentiate between random positions and peak / dips as they neglect the fact that halos ( voids ) may correspond to maxima ( minima ) of the density field . according to bardeen et al .
( 1986 ) , if one assumes the cosmological density fluctuations to be gaussian random fields , the local maxima of such fields are plausible sites for the formation of nonlinear structures . hence , the statistical properties of the peaks can be used to predict the abundances and clustering properties of objects of various types , and in studies of the non - spherical formation of large - scale structures .
the study of bertschinger ( 1987 ) goes in this direction , by generalizing the treatment of bardeen et al .
( 1986 ) and proposing a path integral method for sampling constrained gaussian random fields .
the method allows one to study the density field around peaks or other constrained regions in the biased galaxy formation scenario , but unfortunately it is too elaborate and inefficient in its implementation .
instead , by applying the prescription of hoffman & ribak ( 1991 ) to construct constrained random fields , van de weygaert & bertschinger ( 1996 ) were able to show that it is possible to generate efficiently initial gaussian random density and velocity fields , and specify the presence and characteristics of one or more peaks and dips at arbitrary locations with the gravity and tidal fields at the site of the peaks having the required strength and orientation .
some other studies on constrained initial conditions were also pursued by van haarlem & van de weygaert ( 1993 ) , and by van de weygaert & babul ( 1994 ) .
the idea has been expanded in bond & myers ( 1996 ) , who presented a _
peak - patch _ picture of structure formation as an accurate model of the dynamics of peaks in the density field .
their approach goes further , as it involves the explicit formalism for identifying objects in a multiscale field ( i.e. it does not restrict to a single scale ) .
the model is even more precise for void patches , the equivalent framework for studying voids ( sahni et al .
1994 ; sheth & van de weygaert 2004 ; novikov , colombi & dore 2006 ; colberg et al .
2008 ) . in general , density peaks define a well - behaved point - process which can account for the discrete nature of dark matter halos and galaxies , and on asymptotically large scales are linearly biased tracers of the dark matter field ( desjacques & sheth 2010 ) .
therefore , it would be desirable to incorporate the peak ( or dip ) constraint in the statistical description of the initial shear field , in order to characterize more realistically the geometry and dynamics of the cosmic web .
the main goal of this paper is to do so , by providing a set of analytic expressions which extend the work of doroshkevich ( 1970 ) and bardeen et al .
( 1986 ) , and are akin in philosophy to that of van de weygaert & bertschinger ( 1996 ) .
this is achieved by constraining the hessian of the displacement field ( the matrix of the second derivatives ) to be positive ( negative ) definite , which is the case in the vicinity of minima ( maxima ) of the source of the displacement field .
the new _ conditional _ probability distributions derived in this study include the correlation between the potential and density fields through a reduced parameter @xmath0 , and naturally recover doroshkevich s ( 1970 ) _ unconditional _ formulae in the absence of correlation .
hence , the main focus of this work is to derive explicitly the joint probability distribution of the eigenvalues of the shear field , given the fact that positions are peaks or dips in the corresponding density field and not random locations . in this sense
the field is termed _ constrained _
( i.e. it has _ constrained _ eigenvalues , which are the result of looking only at peak / dip regions ) , and it is statistically described by a _
conditional _ probability distribution . note that even though the formalism is essentially restricted to one scale , the formulae derived here are useful in a variety of applications some of which will be discussed at the end of this paper .
the layout is organized as follows .
section [ new_formula ] provides the derivation of the new analytic expressions for the conditional distributions of eigenvalues of the initial shear field .
in particular , section [ notation ] illustrates the basic notation adopted ; section [ joint_eigen_new ] contains the derivation of the joint distribution of eigenvalues in the peak / dip picture , while section [ bbks_connection ] shows the reverse conditional distribution function . as a direct application of the new formulae , section [ individual_eigenvalues ] presents the individual distributions of eigenvalues subjected to the extremum constraint , along with some other related conditional probabilities .
this part extends previous work by lee & shandarin ( 1998 ) , which is briefly summarized in appendix [ lee_shandarin ] , and complements the study of van de weygaert & bertschinger ( 1996 ) .
finally , section [ conclusion ] highlights the main results and discusses ongoing and future applications , which will be presented in forthcoming publications .
in this section , i first introduce the basic notation adopted throughout the paper .
i then derive the conditional joint distribution of eigenvalues of the tidal field , under the constraint that the hessian matrix of the source of the displacement is positive ( or negative ) definite
. finally , i present the reverse probability function , useful in relating this work with that of bardeen et al .
( 1986 ) and van de weygaert & bertschinger ( 1996 ) .
let @xmath1 denote the displacement field , @xmath2 the potential of the displacement field , @xmath3 the source of the displacement field
. indicate with @xmath4 the lagrangian coordinate , with @xmath5 the eulerian coordinate , where @xmath6 use @xmath7 to denote the shear tensor of the displacement field , @xmath8 for the hessian matrix , @xmath9 for the jacobian of the displacement field ( @xmath10 ) .
clearly : @xmath11 @xmath12 @xmath13 @xmath14 the eigenvalues of @xmath7 are @xmath15 , those of @xmath8 are @xmath16 . in general , the ordering of the eigenvalues is assumed to be such that @xmath17 .
let the potential @xmath2 be a gaussian random field determined by the power spectrum of matter density fluctuations @xmath18 , with @xmath19 denoting the wave number and @xmath20 the smoothing kernel .
the density field described by the source @xmath21 is also a gaussian random field .
the correlations between these two fields are expressed by : @xmath22 @xmath23 @xmath24
where @xmath25 , @xmath26 , @xmath27 , @xmath28 is the kronecker delta , and @xmath29 @xmath30 @xmath7 and @xmath8 are real symmetric tensors , so they are specified by 6 components ( note however that only 5 are independent once the height of the density peak or dip has been set , because the trace of the shear field tensor is entirely constrained by the local density field value via the poisson equation ) .
label them as @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 ; the symbols @xmath37 or @xmath38 indicate the various couples , in a compact notation , where @xmath39 .
for example , if @xmath40 then @xmath41 , and so forth . in the subsequent derivations , for clarity all the various dependencies on @xmath42 s
are dropped .
this is done by introducing the `` reduced '' variables @xmath43 and @xmath44 , defined as @xmath45 and the `` reduced '' correlation @xmath46 where @xmath47 is the same as in eq .
( 4.6a ) of bardeen et al . (
1986 ) . in this notation ,
the eigenvalues of @xmath43 and of @xmath44 are @xmath48 s and @xmath49 s , respectively , with @xmath50 .
of course , the @xmath42 dependence can be restored at any time , if desired , by using the previous equations ( [ tilde_var ] ) and ( [ tilde_corr ] ) . the main goal of this section is to derive the joint distribution of the eigenvalues of @xmath9 when the ordering is @xmath17 , in regions where the source of displacement @xmath21 is a minimum ( or maximum ) .
being a minimum ( maximum ) of @xmath21 implies that the displacement field is such that the gradient of @xmath21 is zero , and the hessian matrix @xmath51 is positive ( or negative ) definite . in block notation ,
the covariance matrix @xmath52 of the 12 reduced components of @xmath53 is : @xmath54 where @xmath55 with @xmath56 a ( @xmath57 ) identity matrix and @xmath58 a ( @xmath59 ) null matrix .
the inverse of the reduced covariance matrix is simply : @xmath60 where @xmath61 the joint probability of observing a tidal field @xmath43 for the gravitational potential and a curvature @xmath44 for the density field ( from now on , the understood indices @xmath37 and @xmath38 are dropped ) is a multivariate gaussian in @xmath62 : @xmath63 therefore @xmath64 where @xmath65 the marginal distributions @xmath66 and @xmath67 are simply multidimensional gaussians with covariance matrix @xmath68 , and therefore they can be expressed using doroshkevich s formulae as @xmath69 with @xmath70 and similarly @xmath71 with @xmath72 the conditional distribution @xmath73 is also a multidimensional gaussian , with mean @xmath74 and covariance matrix @xmath75 , where @xmath76 carrying on the calculation indicated in ( [ cond ] ) yields : @xmath77 \label{doro_inter_extended}\ ] ] where @xmath78 and @xmath79 accounts for cross - correlation terms .
the previous expression ( [ doro_inter_extended ] ) generalizes doroshkevich s formula ( [ doro_standard ] ) to include the fact that halos ( voids ) may correspond to maxima ( minima ) of the density field .
note that , if @xmath80 , doroshkevich s formula ( [ doro_standard ] ) is indeed recovered as expected in the limit of no correlations between the potential and density fields .
equation ( [ doro_inter_extended ] ) is written in a doroshkevich - like format , and it is one of the key results of this paper .
it is also possible to express ( [ doro_inter_extended ] ) in terms of the constrained eigenvalues @xmath81s of the ( @xmath82 ) matrix .
the result is : @xmath83 \cdot where in terms of constrained eigenvalues : @xmath84 with the partial distributions expressed by doroshkevich s unconditional formulae as @xmath85 @xmath86 and @xmath87 which is obvious from ( [ k_def ] ) , with @xmath88 s and @xmath89 s the corresponding unconstrained eigenvalues of the matrices @xmath43 and @xmath44 .
equation ( [ doro_eigen ] ) is another key result of this work .
this relation is easily derived from ( [ doro_inter_extended ] ) because the volume element in the six - dimensional space of symmetric real matrices is simply given , in terms of constrained eigenvalues , by @xmath90 where @xmath91 is the volume element of the three - dimensional rotation group @xmath92 , and also of the three - sphere ( see appendix b in bardeen et al .
in addition , from ( [ k_def ] ) it is direct to see that one has @xmath93 and @xmath94 in the system where both @xmath43 and @xmath44 are diagonal ; therefore @xmath95 , with @xmath96 the constrained eigenvalues .
the fact that the system in which @xmath43 is diagonal is also the system in which @xmath44 is diagonal has a simple geometrical explanation : in order to obtain the conditional distribution ( @xmath82 ) , a standard procedure for multivariate gaussians ( when transforming from uncorrelated to correlated variates ) is to use a cholesky decomposition .
however , this decomposition does not sample symmetrically with respect to the principal axes . an alternative method which takes care of the alignment with the principal axes ( and used here ) is the following : decompose spectrally the covariance matrix of ( @xmath82 ) , @xmath97 , i.e. find the eigensystem @xmath98 and diagonal eigenvalue matrix @xmath99 of the covariance matrix such that @xmath100 ; then set @xmath101 , with @xmath102 a set of independent zero mean unit variance gaussians .
it is direct to show that @xmath103 is indeed the correct covariance matrix , and this scheme takes care of the alignment with the principal axes . equation ( [ what_we_want ] ) expresses the probability of observing a tidal field @xmath43 in regions where the curvature @xmath44 of the density field is positive / negative ( i.e. peak or dip regions ) .
one may be also interested in the reverse joint conditional distribution , namely @xmath104 .
obtaining its expression is easily achieved from the previous algebra by applying bayes theorem , so that : @xmath105 where now ( symmetrically ) @xmath106 the distribution @xmath107 is also a multidimensional gaussian with mean @xmath108 and covariance matrix @xmath75 , where @xmath109 along the lines of the previous calculation , it is direct to obtain @xmath110 \label{bbks_inter_extended}\ ] ] where in analogy with ( [ doro_inter_extended ] ) one has now : @xmath111 in addition , if @xmath112 are the constrained eigenvalues of ( @xmath113 ) , then @xmath114 \cdot where in terms of constrained eigenvalues @xmath115 with the @xmath116 s simply given by @xmath117 note an interesting symmetry : @xmath118 , @xmath119 , @xmath120 , @xmath121 are always the traces of their corresponding matrices .
new analytic formulae derived from ( [ bbks_inter_extended ] ) and ( [ bbks_eigen ] ) , which generalize the work of bardeen et al . ( 1986 ) , will be presented in a forthcoming publication .
the previous relations are also useful in making the connection with the work of van de weygaert & bertschinger ( 1996 ) . for example , equation ( [ constrained_eigen_bis ] ) confirms their finding of the correlation between the shifted mean values for density and shear ( see for example equation 108 in van de weygaert & bertschinger 1996 , which provides the reverse mean value of the shear given a density field shape , and their section 4.4 for more details ) .
of the initial shear field in the peak / dip picture ( equations [ zeta1 ] , [ zeta2 ] and [ zeta3 ] ) , for different values of the reduced correlation parameter @xmath0 as indicated in the panels.,scaledwidth=100.0% ] a direct application of the main formulae derived in section [ joint_eigen_new ] is to compute the individual distributions of eigenvalues ( along with some other related conditional probabilities ) , given the peak / dip constraint .
starting from equation ( [ doro_eigen ] ) , with some extra complications due to the inclusion of correlation terms , it is possible to extend the work of lee & shandarin ( 1998 ) to account for those regions where the source of the displacement is positive ( negative ) see the appendix [ lee_shandarin ] for a compact summary of their results
. introducing the following notation : @xmath122 \nonumber \\ n'(x , y|r ) & = & ( x - y)[8(1-r^2)+3(3x - y)(3y - x)]\cdot { \rm exp}\big[-{15 \over 16 ( 1-r^2 ) } \big ( 3 x^2 -2 xy + 3 y^2 \big ) \big ] \cdot { \rm erfc } \big [ { \sqrt{3 } \over 4 } { 1 \over \sqrt{1-r^2 } } ( x-3y)\big ] \nonumber\end{aligned}\ ] ] and dropping the understood @xmath123 symbol ( but note that the eigenvalues are still `` reduced variables '' ) , the two - point probability distributions of the constrained eigenvalues of the shear field are given by : @xmath124 \\ p(\zeta_2,\zeta_3|r ) & = & l ' \big [ l'(\zeta_3,\zeta_2|r ) + n \cdot n'(\zeta_3 , \zeta_2|r ) \big ] \nonumber \\
p(\zeta_1,\zeta_3|r ) & = & l ' \big \ { l'(\zeta_1,\zeta_3|r ) + l'(\zeta_3,\zeta_1|r ) + n \cdot \big [ n'(\zeta_1 , \zeta_3|r ) + n'(\zeta_3 , \zeta_1|r ) \big ] \big \}.
\nonumber\end{aligned}\ ] ] performing the various partial integrations , it is straightforward to obtain : @xmath125 \nonumber \\ & & + { 3 \sqrt{3 \pi } \over \sqrt{1-r^2}}~ { \rm exp } \big ( - { 15 \zeta_1 ^ 2 \over 4 ( 1-r^2 ) } \big ) { \rm erfc } \big ( -{\sqrt{3 } \zeta_1 \over 2 \sqrt{1-r^2 } } \big ) \big ] \nonumber \end{aligned}\ ] ] @xmath126 \nonumber\end{aligned}\ ] ] @xmath127 \nonumber \\ & & - { 3 \sqrt{3 \pi } \over \sqrt{1-r^2}}~ { \rm exp } \big ( - { 15 \zeta_3 ^ 2 \over 4 ( 1-r^2 ) } \big ) { \rm erfc } \big ( { \sqrt{3 } \zeta_3 \over 2 \sqrt{1-r^2 } } \big ) \big ] .
\nonumber \end{aligned}\ ] ] plots of these distributions are shown in figure [ extended_zetas ] , for different values of the reduced correlation @xmath0 .
note the symmetry between @xmath128 and @xmath129 , evident from ( [ zeta1 ] ) and ( [ zeta3 ] ) .
in particular , @xmath130 is simply a gaussian with variance @xmath131 , and so the conditional distribution @xmath132 is a gaussian with the same variance and shifted mean @xmath133 ( recall that @xmath134 ) .
in addition , it is easy to show that , with @xmath135 , one obtains @xmath136 \cdot { \rm exp } \big [ - { 3 \over 8 ( 1-r^2 ) } ( 3 \delta^2 -10 \delta \zeta_3 + 15 \zeta_3 ^ 2 ) \big ] \nonumber\\ & & + { 3 \sqrt{5 } \over 2 \pi } { 1 \over ( 1-r^2 ) } { \rm exp } \big [ -{3 \over ( 1-r^2 ) } \big ( \delta^2 -5 \delta \zeta_3 + { 15 \over 2 } \zeta_3 ^ 2 \big ) \big ] .\end{aligned}\ ] ] by integrating the previous equation over @xmath137 , it is direct to verify that @xmath138 is a gaussian with zero mean and variance @xmath139 , namely : @xmath140
\equiv p(k_1|r).\ ] ] since @xmath141 ( i.e. eq .
[ eigen_doro_ext ] ) , the previous expression implies that @xmath142 is therefore a gaussian with mean @xmath143 and variance @xmath139 .
in addition : @xmath144 moreover , the probability distribution of @xmath145 confined in the regions with @xmath146 is : @xmath147 .
\nonumber\end{aligned}\ ] ] using the previous formula , it is direct to prove that @xmath148 figure [ k1_positive ] shows the conditional probability distribution @xmath149 ( i.e. equation [ k1_positive_eq ] ) , for different values of @xmath0 .
the dotted vertical lines display the average values of those functions , as given in ( [ k1_positive_ave ] ) , while the vertical solid lines indicate their maximum values .
in fact , it is easy to see that the maximum of @xmath150 is reached when @xmath151 . this result
is readily obtained by computing the derivative of @xmath150 , and by finding the corresponding zeroes .
an example of this calculation is shown in figure [ k1_positive_max ] , where an arbitrary value of @xmath152 is considered . given by equation ( [ k1_positive_eq ] ) , for different values of @xmath0 , as indicated in the panels .
solid vertical lines show the maximum of each distribution ( @xmath153 ) , dotted lines represent their corresponding average values as given by equation ( [ k1_positive_ave]).,scaledwidth=100.0% ] ) .
the derivative @xmath154 is shown , for an arbitrary value of the reduced correlation @xmath0 , i.e. @xmath152 . the maxima correspond to those points where @xmath155 , indicated with solid vertical lines in the figure.,scaledwidth=50.0% ] finally , the probability that a given region with density @xmath145 will have all positive eigenvalues @xmath156 s is : @xmath157 .
\nonumber\ ] ] note that , in absence of correlations between the potential and density fields ( i.e. when @xmath80 ) , all the previous expressions reduce consistently to the unconditional limit of lee & shandarin ( 1998 ) see again appendix [ lee_shandarin ] .
several other results on probability distributions along these lines were also already derived by van de weygaert & bertschinger ( 1996 ) : here we confirm their findings that the inclusion of the condition for a peak or dip in the gaussian density field involves a @xmath158 factor , and that the conditional distributions have shifted mean and reduced variance ( see again their section 4 ) .
since the initial shear field associated with gaussian statistics plays a major role in the formation of large scale structures , considerable analytic work has been based on the joint distribution of its eigenvalues
i.e. doroshkevich s formulae ( equation [ doro_standard ] ) . however , doroshkevich s equations neglect the fact that halos ( voids ) may correspond to maxima ( minima ) of the density field .
the main goal of this work was to provide new analytic expressions , in the context of the peak / dip picture ( bardeen et al .
1986 ; bond et al . 1991 ) , which include the fact that the eigenvalues of the linear shear field are related to regions where the source of the displacement is positive ( negative ) .
these new conditional probabilities , derived in section [ new_formula ] , are equations ( [ doro_inter_extended ] ) , ( [ doro_eigen ] ) , ( [ bbks_inter_extended ] ) and ( [ bbks_eigen ] ) : they represent the main results of this paper . written in doroshkevich - like
format , they naturally reduce to doroshkevich s ( 1970 ) unconditional relations in the limit of no correlation between the potential and the density fields ( i.e. when @xmath80 ) . as a first direct application of ( [ doro_eigen ] )
, the individual conditional distributions of eigenvalues were obtained in section [ individual_eigenvalues ] ; these relations extend some previous work by lee & shandarin ( 1998 ) see appendix [ lee_shandarin ] for a compact summary of their results .
much more analytic work can be carried out using these new formulae , especially in connection with the statistics of peaks developed by bardeen et al .
( 1986 ) : their calculations can be extended within this framework ( see section [ bbks_connection ] ) , and results of this extension will be presented in a forthcoming publication along with more insights on the equations derived in section [ joint_eigen_new ] . to obtain the conditional distribution @xmath159 ( i.e. equation [ what_we_want ] ) , in principle one needs to sample numerically the probability ( [ cond ] ) as done in lavaux & wandelt ( 2010 ) . however , the new analytic results of this paper suggest a simpler generalized excursion set algorithm , which will be also presented in a forthcoming study . the algorithm allows for a fast sampling of ( [ cond ] ) , and permits to test the new relations derived here ( i.e. sections [ new_formula ] and [ individual_eigenvalues ] , and appendix [ lee_shandarin ] ) against mock data .
in addition , along these lines it is possible to extend the main shape distributions involved in the triaxial formation of nonlinear structures ( i.e. ellipticity @xmath160 , prolateness @xmath161 , axis ratios @xmath162 and @xmath163 , etc . ) .
halos and voids are in fact triaxial rather than spherical ( i.e. rossi , sheth & tormen 2011 ) , and the initial shear field plays a crucial role in their formation . for example , in the ellipsoidal collapse framework the virialization condition depends on ellipticity and prolateness , which are directly related to the eigenvalues of the external tidal field . starting from @xmath164 , it is then straightforward to turn this joint conditional distribution into @xmath165 to include the peak constrain , and then to characterize @xmath166 , @xmath167 , @xmath168 , @xmath169 and so forth .
applications to halos and voids will be also presented next , including implications for the skeleton of the cosmic web .
the extension to non - gaussian fields , along the lines of lam et al .
( 2009 ) , is a natural follow - up of this work and will be presented in a forthcoming publication as well . in the context of cosmic voids , other interesting uses of the new formulae ( [ doro_inter_extended ] ) , ( [ doro_eigen ] ) ,
( [ bbks_inter_extended ] ) and ( [ bbks_eigen ] ) involve the monge - ampre - kantorovitch reconstruction procedure
see for example lavaux & wandelt ( 2010 )
. the analytic framework described here can be also useful in several observationally - oriented applications , and in particular for developing algorithms to find and classify structures in the cosmic web .
for example , bond , strauss & cen ( 2010 ) presented an algorithm that uses the eigenvectors of the hessian matrix of the smoothed galaxy distribution to identify individual filamentary structures .
they used the distribution of the hessian eigenvalues of the smoothed density field on a grid to study clumps , filaments and walls .
other possibilities include a web classification based on the multiscale analysis of the hessian matrix of the density field ( aragon - calvo et al .
2007 ) , the skeleton analysis ( novikov et al . 2006 ) , as well as a morphological ( zeldovich - based ) classification ( for instance klypin & shandarin 1983 ; forero - romero et al
more generally , the fact that the eigenvalues of the hessian matrix can be used to discriminate different types of structure in a particle distribution is fundamental to a number of structure - finding algorithms ( forero - romero et al .
2009 ) , shape - finders algorithms ( sahni et al .
1998 ) , and structure reconstruction on the basis of tessellations ( schaap & van de weygaert 2000 ; romano - diaz & van de weygaert 2007 ) , etc .
in addition , the classification of different environments should provide a framework for studying the environmental dependence of galaxy formation ( see for example blanton et al .
to this end , recently park , kim and park ( 2010 ) extended the concept of galaxy environment to the gravitational potential and its functions as the shear tensor .
they studied how to accurately estimate the gravitational potential from an observational sample finite in volume , biased due to galaxy biasing , and subject to redshift space distortions , by inspecting the dependence of dark matter halo properties on environmental parameters ( i.e. local density , gravitational potential , ellipticity and prolateness of the shear tensor ) .
it would be interesting to interpret their results with the theoretical formalism developed here , and ultimately to study the gravitational potential directly from a real dataset such as the sdss main galaxy sample within this framework .
the formalism presented in this paper is restricted to one scale ( i.e. peaks and dips in the density field , as in bardeen et al .
1986 ) , but the extension to a multiscale _ peak - patch _ approach along the lines of bond & myers ( 1996 ) is duable and subject of ongoing work .
this will allow to account for the role of the peculiar gravity field itself , an important aspect not considered here but discussed for example in van de weygaert & bertschinger ( 1996 ) .
in fact , these authors introduced the peak constraints to describe the density field in the immediate surroundings of a peak , and then addressed the constraints on the gravitational potential perturbations ; in particular , they constrained the peculiar gravitational acceleration at the position of the peak itself , in addition to characterizing the tidal field around the peak . including all these effects in our formalism
is ongoing effort .
finally , the interesting and more complex question of the local expected density field alignment / orientation distribution as a function of the local field value ( or the other way around see bond 1987 ; lee & pen 2002 ; porciani et al . 2002 ; lee 2011 ) can be addressed within this framework , and is left to future studies .
i would like to thank ravi k. sheth especially for bringing to my attention the work of lee & shandarin ( 1998 ) , and changbom park for his always wise advices along with numerous discussions .
i would also like to thank the referee , rien van de weygaert , for his insightful report and useful suggestions which have been all included in the paper .
part of this work was carried out in june 2011 during the `` apctp - ieu focus program on cosmology and fundamental physics '' at postech in pohang ( korea ) , and completed during the `` cosmic web morphology and topology '' meeting at the nicolaus copernicus astronomical center in warsaw ( poland ) , on july 12 - 17 , 2011 where i had stimulating conversations with bernard jones and sergei shandarin .
i would like to thank all the organizers of these workshops , and in particular changrim ahn for the former and wojciech a. hellwing , rien van de weygaert and changbom park for the latter .
99 aragn - calvo , m. a. , platen , e. , van de weygaert , r. , & szalay , a. s. 2010 , apj , 723 , 364 aragn - calvo , m. a. , van de weygaert , r. , & jones , b. j. t. 2010 , mnras , 408 , 2163 aragn - calvo , m. a. , jones , b. j. t. , van de weygaert , r. , & van der hulst , j. m. 2007 , a@xmath170a , 474 , 315 bardeen , j. m. , bond , j. r. , kaiser , n. , & szalay , a. s. 1986 , apj , 304 , 15 bertschinger , e. 1987 , apj , 323 , l103 blanton , m. r. , eisenstein , d. , hogg , d. w. , schlegel , d. j. , & brinkmann , j. 2005 , apj , 629 , 143 bond , n. a. , strauss , m. a. , & cen , r. 2010 , mnras , 409 , 156 bond , j. r. , kofman , l. , & pogosyan , d. 1996 , nature , 380 , 603 bond , j. r. , & myers , s. t. 1996 , apjs , 103 , 1 bond , j. r. 1987 , nearly normal galaxies . from the planck time to the present , 388 - 397 , ed
. s. m. faber , springer - verlag catelan , p. , kamionkowski , m. , & blandford , r. d. 2001 , mnras , 320 , l7 catelan , p. , & porciani , c. 2001 , mnras , 323 , 713 colberg , j. m. , et al .
2008 , mnras , 387 , 93 desjacques , v. 2008 , mnras , 388 , 638 desjacques , v. , & sheth , r. k. 2010 , phys .
d , 81 , 023526 desjacques , v. , & smith , r. e. 2008 , phys .
d , 78 , 023527 doroshkevich , a. g. 1970 , astrofizika , 6 , 581 doroshkevich , a. g. , & zeldovich , y. b. 1964 , soviet ast .
, 7 , 615 doroshkevich , a. g. , & shandarin , s. f. 1978 , soviet ast . , 22 , 653 forero - romero , j. e. , hoffman , y. , gottlber , s. , klypin , a. , & yepes , g. 2009 , mnras , 396 , 1815 hoffman , y. , & ribak , e. 1991 , apj , 380 , l5 hu , w. , & sugiyama , n. 1995 , apj , 444 , 489 icke , v. 1973 , a@xmath170a , 27 , 1 kaiser , n. 1986 , mnras , 222 , 323 klypin , a. a. , & shandarin , s. f. 1983 , mnras , 204 , 891 lam , t. y. , sheth , r. k. , & desjacques , v. 2009 , mnras , 399 , 1482 lavaux , g. , & wandelt , b. d. 2010 , mnras , 403 , 1392 lee , j. 2011 , apj , 732 , 99 lee , j. , & springel , v. 2010 , jcap , 5 , 31 lee , j. , & park , d. 2006 , apj , 652 , 1 lee , j. , & pen , u .- l . 2002 ,
apj , 567 , l111 lee , j. , & shandarin , s. f. 1998 , apj , 500 , 14 novikov , d. , colombi , s. , & dor , o. 2006 , mnras , 366 , 1201 park , h. , kim , j. , & park , c. 2010 , apj , 714 , 207 peebles , p. j. e. 1980 , research supported by the national science foundation .
princeton , n.j . , princeton university press , 1980 .
435 p. platen , e. , van de weygaert , r. , & jones , b. j. t. 2008 , mnras , 387 , 128 pogosyan , d. , pichon , c. , gay , c. , prunet , s. , cardoso , j. f. , sousbie , t. , & colombi , s. 2009 , mnras , 396 , 635 porciani , c. , dekel , a. , & hoffman , y. 2002 , mnras , 332 , 339 romano - daz , e. , & van de weygaert , r. 2007 , mnras , 382 , 2 rossi , g. , sheth , r. k. , & tormen , g. 2011 , mnras , 1032 shandarin , s. , habib , s. , & heitmann , k. 2010 , phys .
d , 81 , 103006 shandarin , s. , feldman , h. a. , heitmann , k. , & habib , s. 2006 , mnras , 367 , 1629 sahni , v. , sathyaprakash , b. s. , & shandarin , s. f. 1998 , apj , 495 , l5 sahni , v. , sathyaprakah , b. s. , & shandarin , s. f. 1994 , apj , 431 , 20 schaap , w. e. , & van de weygaert , r. 2000 , a@xmath170a , 363 , l29 shen , j. , abel , t. , mo , h. j. , & sheth , r. k. 2006 , apj , 645 , 783 sheth , r. k. , & van de weygaert , r. 2004 , mnras , 350 , 517 springel , v. , et al .
2005 , nature , 435 , 629 sunyaev , r. a. , & zeldovich , y. b. 1972 , comments on astrophysics and space physics , 4 , 173 van de weygaert , r. , & bond , j. r. 2008 , a pan - chromatic view of clusters of galaxies and the large - scale structure , 740 , 335 van de weygaert , r. , & bertschinger , e. 1996 , mnras , 281 , 84 van de weygaert , r. , & babul , a. 1994 , apj , 425 , l59 van haarlem , m. , & van de weygaert , r. 1993 , apj , 418 , 544 west , m. j. 1989 , apj , 347 , 610 white , s. d. m. 1984 , apj , 286 , 38 zeldovich , y. b. 1970 , a@xmath170a , 5 , 84
starting from doroshkevich s formulae for the unconditional distribution of the ordered eigenvalues @xmath17 ( eq . [ doro_original ] ) , lee & shandarin ( 1998 ) derived a set of useful probability functions .
their expressions can be considerably simplified by using various symmetries , and by introducing the following notation : @xmath171 \nonumber \\ n(x , y ) & = & ( x - y)[8 + 3(3x - y)(3y - x)]\cdot { \rm exp}\big[-{15 \over 16 } \big ( 3 x^2 -2 xy + 3 y^2 \big ) \big ] \cdot { \rm erfc } \big [ { \sqrt{3 } \over 4 } ( x-3y)\big ] .
\nonumber\end{aligned}\ ] ] dropping the understood @xmath123 symbols for the eigenvalues ( although all the quantities are still `` reduced '' , i.e. the @xmath42 dependence is not shown ) , the two - point probability distributions are : @xmath172 \\ p(\lambda_2,\lambda_3 ) & = & l \big [ l(\lambda_3,\lambda_2 ) + n \cdot n(\lambda_3 , \lambda_2 ) \big ] \\
p(\lambda_1,\lambda_3 ) & = & l \big \ { l(\lambda_1,\lambda_3 ) + l(\lambda_3,\lambda_1 ) + n \cdot \big [ n(\lambda_1 , \lambda_3 ) + n(\lambda_3 , \lambda_1 ) \big ] \big \}.\end{aligned}\ ] ] performing the various integrations : @xmath173 \nonumber \end{aligned}\ ] ] note the symmetry between @xmath176 and @xmath177 . in addition , with @xmath178 , one gets @xmath179\cdot { \rm exp } \big [ - { 3 \over 8 } ( 3 \delta^2 -10 \delta \lambda_3 + 15 \lambda_3 ^ 2 ) \big ] + { 3 \sqrt{5 } \over 2 \pi } { \rm exp } \big [ -3 \big ( \delta^2 -5 \delta \lambda_3 + { 15 \over 2 } \lambda_3 ^ 2 \big ) \big ] .\ ] ] by integrating the previous expression over @xmath180 , it is direct to verify that @xmath181 is a zero - mean unit - variance gaussian distribution .
in addition : @xmath182 moreover , the probability distribution of @xmath183 confined in the regions with @xmath184 is : @xmath185 .
\nonumber\end{aligned}\ ] ] using the previous formula , it is direct to show that @xmath186 it is also easy to see that the maximum of @xmath187 is reached when @xmath188 : this can be readily achieved by computing the derivative of the previous expression and by finding the corresponding zeroes , as done in the main text ( see figure [ k1_positive_max ] ) .
finally , the probability that a given region with density @xmath183 will have all positive eigenvalues is : @xmath189 .
\nonumber\ ] ] | the initial shear field , characterized by a primordial perturbation potential , plays a crucial role in the formation of large scale structures .
hence , considerable analytic work has been based on the joint distribution of its eigenvalues , associated with gaussian statistics .
in addition , directly related morphological quantities such as ellipticity or prolateness are essential tools in understanding the formation and structural properties of halos , voids , sheets and filaments , their relation with the local environment , and the geometrical and dynamical classification of the cosmic web . to date , most analytic work has been focused on doroshkevich s unconditional formulae for the eigenvalues of the linear tidal field , which neglect the fact that halos ( voids ) may correspond to maxima ( minima ) of the density field .
i present here new formulae for the constrained eigenvalues of the initial shear field associated with gaussian statistics , which include the fact that those eigenvalues are related to regions where the source of the displacement is positive ( negative ) : this is achieved by requiring the hessian matrix of the displacement field to be positive ( negative ) definite .
the new conditional formulae naturally reduce to doroshkevich s unconditional relations , in the limit of no correlation between the potential and the density fields . as a direct application
, i derive the individual conditional distributions of eigenvalues and point out the connection with previous literature .
finally , i outline other possible theoretically- or observationally - oriented uses , ranging from studies of halo and void triaxial formation , development of structure - finding algorithms for the morphology and topology of the cosmic web , till an accurate mapping of the gravitational potential environment of galaxies from current and future generation galaxy redshift surveys .
[ firstpage ] methods : analytical , statistical cosmology : theory , large - scale structure of universe galaxies : formation |
the experimental demonstration of bose - einstein condensates ( bec ) @xcite has lead to the development of atom lasers by outcoupling atoms from trapped becs by either a radio frequency transition or a raman transition to change the internal state of the atom to one that is either untrapped or anti - trapped @xcite .
atom lasers are coherent matter waves with spectral fluxes many orders of magnitude higher than thermal sources of atoms .
the coherence of these sources will enable an increase in the sensitivity of interferometric measurements @xcite .
although current experiments usually operate in parameter regimes limited by technical noise , the fundamental limit on these measurements will be caused by the shot noise of the atomic field , which will be intrinsic to all interferometers without a non - classical atomic source .
sensitivity is increased in optical interferometry by ` squeezing ' the quantum state of the optical field , where the quantum fluctuations in one quadrature are reduced compared to a coherent state , while the fluctuations in the conjugate quadrature are increased . in the context of atom optics ,
it is interesting to ask whether highly squeezed atom optical sources can be produced .
there is also great interest in the production of entangled atomic beams for quantum information processing and tests of quantum mechanics with massive particles @xcite .
this paper will describe methods of coupling the quantum statistics from optical fields to produce non - classical atomic sources with high efficiency .
generation of squeezed atomic beams has been proposed by either utilising the nonlinear atomic interactions to create correlated pairs of atoms via either molecular down conversion or spin exchange collisions @xcite , or by transferring the quantum state of a squeezed optical field to the atomic beam @xcite . in the first case
, it was shown that collisions between two condensate atoms in the @xmath0 state can produce one atom in the @xmath1 and one atom in the @xmath2 , with sufficient kinetic energy to escape the trap @xcite .
it was shown that this scheme produced pairs of atoms entangled in atomic spin . in the second scheme a bec of molecules composed of bosonic atoms
is disassociated to produce twin atomic beams , analogous to optical down conversion @xcite .
it was shown that the beams were entangled in the sense that phase and amplitude measurements on one beam could infer phase and amplitude measurements of the other beam better than the heisenberg limit .
although each atomic pair is perfectly correlated in direction in each of these schemes , there is very little control of direction of each pair , so the spectral flux would be limited .
the generation of nonclassical light is well established experimentally @xcite .
this suggests that a nonclassical atom laser output could be generated by transferring a the quantum state of an optical mode to an atomic beam .
_ showed that a quantized probe field could be partially transferred to momentum ` side modes ' of a condensate consisting of three - level atoms in the presence of a strong pump field @xcite . jing _
et al . _ performed a single mode analysis of the atom laser outcoupling process for a two - level atom interacting with a quantized light field , and showed that the squeezing in light field would oscillate between the light field and the atomic field at the rabi frequency @xcite . as this
was a single mode analysis , the interaction with the atoms as they left the outcoupling region was not taken into account .
@xcite showed that raman adiabatic transfer can be used to transfer the quantum statistics of a propagating light field to a continuously propagating beam of atoms by creating a polariton with a spatially dependent mixing angle , such that the output contained the state of the probe beam . in this paper
, we model the dynamics of an atom laser produced by outcoupling three - level atoms from a bec via a raman transition , and investigate the transfer of quantum statistics from one of the optical modes to the atomic field . ideal transfer will occur when the time taken for each atom to leave the outcoupling region is a quarter of a rabi period .
the finite momentum spread of a trapped condensate means that there will be a broadening of the time taken to leave the outcoupling region , and hence ideal transfer will not be possible . to determine the effectiveness of the quantum state transfer , we require a multimode model that takes into account back coupling and the finite momentum spread of the condensate . in section
ii we describe an atom laser beam made by outcoupling from a bec using a non - trivial optical mode , using the simplest possible model that contains the spatial effects in the output mode .
we derive the heisenberg equations of motion for this system under suitable approximations .
section iii introduces the method used to solve these equations and investigates some properties of the outcoupled atoms , showing that complicated spatial behaviour occurs in the output even when the optical and bec fields are described by a single mode . in section
iv we investigate continuous outcoupling with two mode squeezing , and show that it can be used to generate continuous variable entanglement in twin atom laser beams propagating in different directions .
when an atomic and an optical field are coupled , and they can both be described by a single mode , then complete state transfer must occur between them in a rabi - like cycle .
when producing an atom laser beam in this manner , however , the single mode approximation can not be made for the output field , even though it may be applicable to the optical and bec fields . in this section
we develop such a model , and derive the heisenberg equations of motion for the output field and the optical field operators .
we model an atom laser in one dimension as a bec of three - level atoms coupled to free space via a raman transition , as shown in figure [ fig : levels ] .
state @xmath3 represents the internal state of the trapped condensate , @xmath4 the excited state , and @xmath5 the untrapped atomic mode .
@xmath6 is the annihilation operator for the probe optical mode ( transition @xmath7 ) , and @xmath8 is the annihilation operator for the pump optical mode ( transition @xmath9 ) . the pump field is assumed to be a large coherent state , much stronger than the probe field , so it is approximated well by a classical field @xmath10 .
internal energy levels of a three level atom . a condensate of state @xmath3 atoms confined in a trapping potential are coupled to free space via a raman transition affected by a probe beam ( annihilation operator @xmath6 ) which is detuned from the excited state ( @xmath4 ) by an amount @xmath11 , and a pump field ( annihilation operator @xmath8 ) which is detuned from the excited state by an amount @xmath12 . ]
the hamiltonian for the system ( in the rotating wave approximation ) is : @xmath13 where @xmath14 is the k - space annihilation operator the condensate mode ( internal state @xmath3 ) , @xmath15 is the annihilation operator for atoms in the excited atomic state ( @xmath4 ) , and @xmath16 is the annihilation operator for the untrapped free propagating mode ( @xmath5 ) . the annihilation operators obey the usual bosonic commutation relations : @xmath17 = [ \hat{\psi}^{\dag}_i(k),\quad \hat{\psi}^{\dag}_j(k ' ) ] = 0 , \\ \nonumber [ \hat{\psi}_i(k),\quad \hat{\psi}^{\dag}_j(k')]=\delta_{ij}\delta(k - k ' ) \end{aligned}\ ] ] @xmath18 is the single particle hamiltonian for the trapped atoms , @xmath19 is the mass of the atoms , @xmath20 is the rabi frequency for the pump transition , @xmath21 is the coupling strength between the atom and the probe field , @xmath22 is the internal energy of the excited state @xmath4 atoms , and @xmath23 and @xmath24 are the momentum kicks due to the pump and probe light fields respectively . for simplicity
we have assumed a laser geometry where the pump and probe fields are counter propagating , to give the maximum possible momentum kick to the untrapped atoms .
the equations of motion for the heisenberg operators are : @xmath25 where @xmath26 and @xmath27 , and @xmath28 is the two - photon detuning
. the population of state @xmath4 will be much less than the other levels when the detunings ( @xmath11 , @xmath12 ) are much larger than the other terms in the system ( including the kinetic energy of the excited state atoms ) .
furthermore , most of the dynamics will occur on time - scales less than @xmath29 , so in this regime we can set @xmath30 . if the condensate has a large number of atoms and is approximately in a coherent state , we can write @xmath31 , where @xmath32 is the condensate wavefunction ( which we will assume is in the ground state of the harmonic oscillator , with @xmath33 the trapping frequency ) and @xmath34 is the condensate number . we have ignored the atom - atom interactions in our model , which is valid only if the condensate is dilute .
strong atom - atom interactions would have the effect of introducing complicated evolution to the quantum state of the condensate mode .
inclusion of these effects is not possible with our method , and a more complicated technique such as a phase space method would be required @xcite .
the approximation of ignoring the back action on the condensate is only valid if we are in the regime where the outcoupling is weak , ie . the number of photons in the probe field is is much less than the number of atoms in the condensate . in an experiment ,
measuring the quantum noise on the atom laser beam would require small classical noise on the beam , and in practice this is easier to achieve with weak outcoupling . with these approximations our equations of motion for the free propagating atoms and
the probe field become @xmath35 with @xmath36 , @xmath37 , @xmath38 ) , and @xmath39 . in the next section we will discuss the solution to these equations and the properties of the outcoupled atoms .
the solution to equations ( [ psidot ] ) and ( [ adot ] ) is @xmath40 where @xmath41 and @xmath42 are the schrdinger picture operators , and @xmath43 , @xmath44 , @xmath45 , @xmath46 are complex functions satisfying : @xmath47 with initial conditions @xmath48 , @xmath49 , and @xmath50 .
this ansatz has reduced the field operator equations to a set of coupled partial differential equations .
this will only be possible for heisenberg equations of motion that do not have terms with products of operators , but it allows the possibility of an analytic or numerical solutions to the full quantum problem .
we solved equations ( [ semiclassical ] ) numerically using a fourth order runge kutta algorithm using the xmds numerical pacakge @xcite .
we chose parameters realistic to atoms optics experiments with rb@xmath51 atoms .
unless stated otherwise , we have set @xmath52 kg , @xmath53 rad @xmath54 , @xmath55 m@xmath56 , which corresponds to twice the wave number of the @xmath57s@xmath58,f@xmath59 @xmath60 @xmath57p@xmath61,f@xmath62 transition in rb@xmath51 .
@xmath32 was chosen to be the ( normalized ) ground state momentum space wave function of a condensate ( ignoring interactions ) in a harmonic trap , and we set @xmath63 with @xmath64 rad s@xmath56 .
the results are reasonably insensitive to the absolute magnitude of @xmath65 and @xmath66 , but they are quite sensitive to the relative values . to maintain resonance between the two fields , we set @xmath67 .
these relationships can be obtained with physically realistic parameters and are consistent with the approximations made in this model .
we set @xmath68 rad s@xmath56
. figures ( [ fig : f1 ] ) , ( [ fig : g1 ] ) , ( [ fig : p1 ] ) and ( [ fig : q1 ] ) show the solutions to equations ( [ semiclassical ] ) for the values indicated above .
@xmath69 for the values of parameters indicated in the text .
@xmath70 is the kick acquired due to the raman transition , ie @xmath71 .
the function was discretized for numerical calculation by replacing @xmath72 with @xmath73 , where @xmath74 is the grid spacing .
the dip in the function near the @xmath75 resonance shows how the quantum state of the atoms has been affected by the interaction with the optical fields . ]
@xmath76 as found numerically for the values of parameters indicated in the text .
this shows that atoms are created around @xmath75 with a quantum state related to the initial state of the probe field . ]
@xmath77 as found numerically for the values of parameters indicated in the text . ] @xmath78 as found numerically for the values of parameters indicated in the text . ]
the solution of equations ( [ semiclassical ] ) gives us the solution of eqs.([solutionpsi ] ) and ( [ solutiona ] ) for all possible initial quantum states of the optical field and the free propagating atomic field .
we will assume that the initial state of the field is @xmath79 [ initstate ] , where @xmath80 represents an arbitrary state for the optical mode , and @xmath81 represents a vacuum mode at all points in @xmath82 space for the atomic field .
the expectation value of the density of outcoupled atoms @xmath83 with @xmath84 is @xmath85 it is interesting to note that when the initial state of the untrapped atomic field is the vacuum , then the spatial structure of the density of the untrapped atoms at later times depends only on the the functional form of @xmath86 , which depends on the efficiency of the outcoupling process . figure ( [ fig : xdens ] ) show the density of outcoupled atoms when the expectation value of the initial number of photons is @xmath87 .
@xmath88 for @xmath87 . ]
the number operator is @xmath89 . using our solution ( eq . [ solutionpsi ] ) and
our initial state , the expectation value is @xmath90 .
the variance of the number operator is @xmath91 using our solution ( eq . [ solutionpsi ] ) and our initial state , this becomes @xmath92 with @xmath93 .
we note that as @xmath94 , the variance in the number of outcoupled atoms approaches the variance of the initial optical mode , as the quantum statistics of the outcoupled atoms depends only on the initial quantum state of the optical field and the efficiency of the outcoupling process .
figure ( [ fig : nvar1 ] ) shows the variance of the outcoupled atoms versus time for different states of the optical mode .
the relative variance @xmath95 for the outcoupled atoms versus time for different states of the optical field .
the solid line represents the initial optical mode in a coherent state @xmath96 with @xmath97 , the dotted line represents a squeezed state @xmath98 with @xmath97 , @xmath99 , and the dashed line represents a fock state @xmath100 with @xmath101 . ] a more interesting observable to look at is the flux of the outcoupled atoms , as the spatial structure of the outcoupled beam becomes apparent .
the flux operator is : @xmath102 which , using our solution for @xmath103 becomes @xmath104 with @xmath105 and with @xmath106 . using our initial state @xmath107 ,
the expectation value of the flux operator becomes : @xmath108 this shows that the mean atom flux in the output pulse depends only on the details of the coupling process , and not on the statistics of the outcoupling field .
figure [ fig : flux1 ] shows the flux of outcoupled atoms for @xmath109 .
flux of outcoupled atoms at a point in the atomic beam ( @xmath110 mm ) for @xmath109 . ] to investigate how the quantum statistics are transferred to the atomic beam , we look at the variance in the flux .
@xmath111 the variance in the flux has two terms , one proportional to the variance in the photon number , and the other proportional to the photon number itself . for a fock state
photonic input the first of those terms is zero , and the variance is proportional to the function @xmath112 .
this can be contrasted to the case where the optical field is in a coherent state , and the total variance in the flux is simply proportional to the function @xmath113 .
a reasonable measure of the transfer of the quantum state of the zero - dimensional photon field to the larger space of the output pulse is therefore the function @xmath114 which shows the minimum possible variance in the output flux normalised to the flux variance produced by output with a coherent optical state .
figure ( [ fig : fluxfigy ] ) shows @xmath115 for different values of the coupling constant @xmath116 . even in our simplified model
where we have assumed a single mode for the optical beam and the condensate , the outcoupled atoms still display complicated spatio - temporal dynamics .
weak outcoupling gives a steady flux , but very little suppression of the shot noise because the timing of the output of each atom becomes uncertain , making the number statistics uncertain in the transient period .
when the outcoupling rate is increased , a significant amount of flux squeezing is displayed in a localised pulse .
further increase of the outcoupling rate shows more complicated dynamics , as some of the outcoupled atoms are coupled back into the condensate .
this causes the atoms to come out in a series of pulses , with less flux squeezing than for optimal outcoupling .
an interesting sidenote is that the flux variance produced by the coherent optical state ( the denominator of @xmath115 ) is simply proportional to the flux itself , with the same proportionality constant for all times , and all values of @xmath116 .
@xmath115 at a point in the path of the atomic beam ( @xmath110 mm ) for @xmath117 rad s@xmath56 ( dotted line ) , @xmath118 rad s@xmath56 ( solid line ) , and @xmath119 rad s@xmath56 ( dashed line ) .
coupling weakly produces a long pulse , but the variance in the flux is almost unaffected by the statistics of the optical state .
coupling too strongly causes significant back coupling from the output field to the photonic state . ]
although the variance in the number of the pulse is quite insensitive to the strength of the coupling between the trapped and untrapped fields , this is not true for the variance in the flux . figure ( [ fig : maxsqueezy ] ) shows the maximum suppression of shot noise in @xmath115 for different values of @xmath116 .
we can estimate the outcoupling that will produce the minimum @xmath115 by finding the maximum rabi frequency that will not cause significant back - coupling to the condensate . equating the quarter - period of a rabi oscillation @xmath120 , with the time taken for the kicked atoms to leave the coupling region @xmath121 where @xmath122 is the spatial width of the condensate . from this
we can estimate that optimum outcoupling will occur when @xmath123 rad s@xmath56 for the parameters used in this paper .
this agrees well with the calculated minimum shown in figure ( [ fig : maxsqueezy ] ) .
minimum value of @xmath115 versus @xmath116 .
the shot noise is below the vacuum noise for all values of @xmath116 when using a fock state to outcouple . ]
we have shown that the quantum statistics of an optical mode can be transferred to an atom laser beam to produce a pulse of atoms with better defined number , or to partially suppress the fluctuations in the flux .
furthermore , we have shown that the quantum statistics of the optical mode can be transferred independent of the initial quantum state of the optical mode , which suggests that two - mode optical squeezing could be used to generate spatially separated entangled atomic beams . in the next section
we investigate the possibility of using twin optical beams produced from a non - degenerate optical parametric oscillator to generate two entangled atomic beams propagating in different directions .
continuous wave generation of correlated atom beams requires a more complicated scheme .
we consider a probe field created by a nondegenerate opo producing twin optical beams ( figure ( [ fig : eprscheme ] ) ) .
these modes have the same wavelength , but travel in different directions and hence they have different momenta .
the opo is driven by a classical , non - depletable driving field .
this will produce twin atom laser beams with different momenta .
this differs from the previous case in that it allows continuous outcoupling of the atoms , rather than just a pulse .
the hamiltonian for the system is now twin atom laser beams produced by outcoupling with two - mode squeezed light .
an opo is driven by a classical , non - depletable driving field ( @xmath124 ) .
the @xmath125 process produces two optical modes @xmath126 and @xmath127 , which are used to outcouple the atom laser beams . ]
@xmath128 where @xmath126 and @xmath127 represent the annihilation operators for the twin probe fields produced from the down conversion process , both assumed to affect the @xmath7 transition , @xmath124 is the complex amplitude of the pump field , @xmath129 is the frequency of the pump , and @xmath130 is the nonlinear coefficient of the down conversion medium . @xmath131 and @xmath132 are the magnitudes of the momentum kicks due to absorption from photons in @xmath126 and @xmath127 respectively .
we have assumed that the photons are resonant in an optical resonator with 100% reflective mirrors .
this assumption is valid as the dominant form of loss out of the cavity will be due to atomic absorption . for computational convenience in our one - dimensional model ,
we have chosen @xmath133 , ie .
the resultant momentum kicks that the atoms obtain after being outcoupled are of equal magnitude and opposite direction . by adiabatically eliminating the excited state , and assuming the condensate is a large coherent state as before , we obtain the following equations of motion for the outcoupled atoms and the probe fields : @xmath134 with @xmath36 , @xmath135 , @xmath136 for @xmath137 , and @xmath138 .
the @xmath139 cross coupling term between the two optical modes is due to atoms absorbing a photon from one beam and emitting it into the other beam
. this term will be small due to the large momentum difference between the two modes .
however , the functional form of @xmath139 is due to our assumption that the condensate remains single mode
. cross coupling between the two optical modes will cause momentum ` side bands ' on the condensate mode @xcite , but the effect of this cross coupling will be small if the number of photons in the probe beam is small compared to the number of atoms in the condensate . as the results in this section are calculated in a parameter regime where the chance of an outcoupled atom coupling back into the condensate is low , it is valid to neglect this term in our calculations .
the general solution to equations ( [ opoeom ] ) is @xmath140 where @xmath141 , @xmath142 , @xmath143 , @xmath144 , @xmath145 , @xmath146 are complex functions satisfying differential equations obtained by substituting the solutions ( [ oposol ] ) into ( [ opoeom ] ) ( see appendix ) . from the solution of these equations
, we can calculate any observable of the system .
we solved the equations ( [ opoprop ] ) numerically for @xmath147s@xmath56 , @xmath148 with @xmath149 rad s@xmath56 for @xmath137 .
we set @xmath150 m@xmath56 , and @xmath151 , with all other parameters as before .
if we assume that the two optical modes and the untrapped atomic field are initially in the vacuum state , using equation [ [ oposol ] ] the expectation value of the atomic density @xmath152 is @xmath153 where @xmath154 , @xmath155 and @xmath156 .
figure([fig : opodensity ] ) shows the atomic density versus time .
two atomic beams in opposite directions are produced , with steady flux .
density of outcoupled atoms versus time .
outcoupling using light from a non - degenerate opo produces two atomic beams in opposite directions , with steady flux . ] to check whether there are correlations present in the two atomic beams , the relevant observable is the difference in the flux of the two beams .
if the two beams are completely uncorrelated , then we would expect @xmath157 where @xmath158 and @xmath159 are points that lie in the rightward and leftward propagating beams respectively .
figure ( [ fig : opovariance ] ) shows the variance of the flux difference @xmath160 versus time , and shows that the fluctuations in the difference are approximately eight times less than for uncorrelated atoms .
variance in the flux difference .
the fluctuations are eight times smaller than for uncorrelated atoms . ]
we now investigate whether we can use this process to generate entanglement between the two atomic beams .
we quantify our entanglement using the epr criterion of reid and drummond @xcite , the requirement being that two conjugate variables on one of the beams can be inferred from measurements on the other beam to below the quantum limit .
we define four quadratures : @xmath161 with @xmath162 the commutator of the conjugate quadratures give us the uncertainty relation @xmath163 since @xmath164 .
the beams are entangled under the epr criterion if by making measurements of quadratures of one beam ( eg . @xmath165 and @xmath166 ) , then quadratures of the other beam ( @xmath167 and @xmath168 ) can be inferred to better than this quantum limit .
quantitatively : @xmath169 is the requirement for entanglement , where @xmath170 , @xmath171 and @xmath172 .
we note here that the correlations present are between conjugate quadratures of each beam .
figure ( [ fig : epr1 ] ) shows the product of the inferred variances @xmath173 plotted against time .
as the intensity of the beams increase and become more monochromatic , the product of the inferred variances dip well below the requirement for entanglement .
the initial increase is due to the beams initially not approximating plane waves
. this could be fixed by appropriate choice of @xmath174 , to better match the mode shape of the output atom laser beams .
product of the inferred variances @xmath173 versus time . as the system goes to steady state ,
the requirement for entanglement is satisfied . ] in the long time limit , the product of the inferred variances is on the order of three orders of magnitude below the classical limit , demonstrating that this system produces an almost pure epr correlated state .
our model uses an ideal opo , so the squeezing in the optical modes would grow without bound in the absence of the damping due to the atoms . in practice
, the entanglement on the atomic beams will not exceed the optical entanglement that can be obtained from a real opo .
the limit to the entanglement between the atomic beams in this model is given by the finite momentum width of the condensate .
as the epr state is expected to be very pure , the dominant noise in an experiment may actually be due to some of the effects we have ignored in our model due to their small effect on the dynamics . in particular , there may be a small reduction in fidelity due to effects of the back action of the outcoupling on the condensate wavefunction , which we have ignored in this calculation .
we have modelled the dynamics of an atom laser produced by outcoupling from a bose - einstein condensate with squeezed light .
we modelled the multimode dynamics of the output field and showed that a significant amount of squeezing can be transferred from an optical mode to a propagating atom laser beam .
we also demonstrated that two - mode squeezing can be used to produce twin atom laser beams with continuous variable entanglement in amplitude and phase .
this research was supported by the australian research council centre of excellence for quantum atom optics .
we would like to acknowledge useful discussions with m. k. olsen , a. m. lance and h. a. bachor .
@xmath175 , @xmath176 , @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 , @xmath187 , @xmath188 , @xmath189 , @xmath190 , @xmath191 , @xmath192 , must satisfy : @xmath193 with @xmath194 , and initial conditions @xmath195 , @xmath196 , @xmath197 with all other fields zero . from the solution of these equations
, we can calculate any observable of the system .
99 f. dalfovo and s. giorgini , rev .
phys . * 71 * , 463 ( 1999 ) .
m. r. andrews , _ et al .
_ , science * 275 * , 637 ( 1997 ) .
lett . * 78 * , 582 ( 1997 ) .
j. l. martin _
et al . _ , j. phys .
b 32 , 3065 ( 1999 ) .
i. bloch , t. w. hansch , and t. esslinger , phys .
lett . * 82 * , 3008 ( 1999 ) .
e. w. hagley _
et al . _ ,
science * 283 * 1706 ( 1999 ) .
b. p. anderson and m. a. kasevich , science * 282 * , 1706 ( 1998 ) .
y , le coq _ et al . _ ,
* 87 * , 170403 ( 2001 ) .
giovanni cennini _
et al . _ ,
lett . * 91 * , 240408 ( 2003 ) .
n. p. robins , c. m. savage , j. j. hope , j. e. lye , c. s. fletcher , s. a. haine , and j. d. close , phys .
a. * 69 * , 051602(r ) ( 2004 ) .
t. l. gustavson _
et al . _ ,
* 78 * , 2046 ( 1997 ) .
t. calarco , h.j .
briegel , d.jaksch , j. i. cirac , p. zoller , journal of modern optics * 47 * , 2137 , ( 2000 ) .
duan , a. sorensen , j. i. cirac and p. zoller , phys .
* 85 * , 3991 ( 2000 ) .
h. pu and p. meystre , phys .
lett . * 85 * , 3987 , ( 2000 ) .
p. d. drummond and k. v. kheruntsyan , phys .
a. * 70 * , 033609 , ( 2002 ) ; k. v. kheruntsyan , m. k. olsen and p. d. drummond , e - print cond - mat/0407363 , ( 2004 ) . h. a. bachor and t. c. ralph , _ a guide to expeiments in quantum optics _ , ( wiley - vch verlag gmbh & co. kgaa , weinheim , 2004 ) . m. g. moore , o. zobay and p. meystre , phys . rev
. a. * 60 * , 1491 ( 1999 ) .
hui jing , jin - ling chen , and mo - lin ge , phys .
rev . a. * 63 * 015601 - 1 ( 2000 ) .
michael fleischhauer and shangqing gong , phys .
lett . * 88 * , 070404 - 1 ( 2002 ) .
d. f. walls and g. j. milburn , _ quantum optics _ , ( springer - verlag heidelberg , 1994 ) .
g. gollecutt , p. d. drummond , p. cochrane and j. j. hope ,
extensible multi - dimensional simulator " , documentation and source code available from http://www.xmds.org m. d. reid and p. d. drummond , phys .
lett . * 60 * , 2731 ( 1988 ) . | we examine the properties of an atom laser produced by outcoupling from a bose - einstein condensate with squeezed light .
we model the multimode dynamics of the output field and show that a significant amount of squeezing can be transfered from an optical mode to a propagating atom laser beam .
we use this to demonstrate that two - mode squeezing can be used to produce twin atom laser beams with continuous variable entanglement in amplitude and phase . |
an understanding of the nature of the electron spectral function in the underdoped ` pseudogap ' regime has emerged as one of the central problems in the study of the cuprate superconductors .
a wealth of data has appeared in photoemission studies , some of which @xcite has been interpreted using a model of `` fermi arcs '' across the brillouin zone diagonals ; other studies @xcite have indicated the presence of pocket fermi surfaces in the same region of the brillouin zone .
scanning tunnelling microscopy ( stm ) studies @xcite also indicate a fermi arc of excitations which appears to end abruptly at the magnetic brillouin zone boundary associated with two sublattice nel order .
another issue of interest in experiments has been the angular dependence of the electronic excitation gap energy in the superconducting state .
a ` dichotomy ' has been noted @xcite between the behavior of the gap near the nodal points on the brillouin zone diagonals , and the antinodal points along the principle axes of the brillouin zone .
related studies have also been made of the electronic spectra in the electron - doped cuprates @xcite . here
the spectra seem to be closer to that expected from conventional spin density wave theory , and so constitute an important testing ground of theoretical ideas .
the recent observation of quantum oscillations in a strong magnetic field @xcite has given further impetus to the development of a theory the normal state of the underdoped cuprates .
some of the transport data @xcite has been interpreted in terms of the presence of electron pockets in the hole - doped cuprates , for which there is no apparent evidence in the photoemission studies ; the latter , however , have only been carried out without a magnetic field .
finally , an important motivation for our study comes from numerical studies @xcite of hubbard - like models for the cuprate superconductors .
these studies show a significant regime without any antiferromagnetic order , but with a dichotomy in the normal state electronic spectra between the nodal and anti - nodal regimes . it would clearly be useful to have analytic effective models which can be used to interpret the numerical data , and we shall propose such models here .
the theory we present here builds upon the framework set up in refs . .
these papers described a non - fermi liquid state which was labeled an ` algebraic charge liquid ' ( acl ) ; this state was obtained after the loss of antiferromagnetic order in a metal via an unconventional transition @xcite . as we will review below , in its simplest realization ,
the degrees of freedom of the acl are spinless fermions and a @xmath0 complex boson @xmath1 which interact via an emergent u(1 ) gauge force .
this gauge force has strong effects even in possible deconfined phases , and can lead to the formation of electron - like bound states between the fermions and @xmath1 .
the main purpose of this paper is to present a general discussion of the dynamics of these fermionic bound states , and their influence on the photoemission spectra .
a full description of the pseudogap regime in the hole - doped cuprates will also require @xcite considerations of pairing of these fermionic bound states into bosonic cooper pairs : this we defer to a subsequent paper . in our previous work @xcite , we presented analytic arguments based upon the structure of a continuum theory valid at long wavelengths . here
, we shall present a more general formulation of the theory , which allows computation of the electronic spectrum across the entire brillouin zone
. we shall show how arguments based upon symmetry and gauge invariance allow construction of an effective theory for the electronic spectrum .
the theory will contain a number of coupling constants , whose values will have to be determined by comparing to numerical studies or experiments . also , while the previous work @xcite used strong - coupling perspective , starting from the schwinger boson theory of the antiferromagnet .
it is possible to derive the results presented below also from this strong - coupling approach .
however , we will choose to present our results by departing @xcite from the `` spin - fermion '' model , which was originally developed from the weak coupling expansion .
let us begin by defining the lagrangian , @xmath2 , of the spin - fermion model .
@xcite we consider fermions @xmath3 @xmath4 hopping on the sites of a square lattice .
these are coupled to the fluctuations of the unit vector field @xmath5 ( @xmath6 ) representing the local orientation of the collinear antiferromagnetic nel order .
we will restrict our attention here to antiferromagnetic order at wavevector @xmath7 , although generalizations to other @xmath8 are possible . throughout , we will freely make a gradient expansion of @xmath5 over spatial co - ordinates @xmath9 , focusing on the long wavelength fluctuations of the order parameter . however , the fermion fields @xmath10 have important low energy modes at many locations in the brillouin zone , and so we will not make any gradient expansion on the fermion operators .
we have the imaginary time ( @xmath11 ) lagrangian @xmath12 . \label{esf}\end{aligned}\ ] ] here @xmath13 are arbitrary hopping matrix elements describing the `` large '' fermi surface , @xmath14 is the chemical potential , @xmath15 is the spin - fermion coupling , @xmath16 controls the strength of the antiferomagnetic fluctuations , @xmath17 are the pauli matrices , and the @xmath18 field obeys the local constraint @xmath19
. almost all previous studies of the spin - fermion model @xcite have involved a perturbative treatment in powers of the coupling @xmath15 , along with resummations of this expansion . here
, we will not expand in powers of @xmath15 , treating it is a coupling of order unity : instead our analysis are motivated by expansions either in the number of field components , or by small @xmath16 .
for sufficiently small values of the coupling @xmath16 , the model @xmath2 clearly has an antiferromagnetically ordered spin density wave ( sdw ) ground state with @xmath20 .
we are interested here in the mechanism by which this order is lost as @xmath16 is increased , and a metallic state with no broken symmetries is obtained . in a recent paper @xcite with others , we argued that there were 2 generic possibilities : + ( _ i _ ) in the first case , there was a direct transition at a single critical @xmath21 to a fermi liquid metal with a large fermi surface .
this transition has been examined in previous work @xcite , and is directly expressed in terms of fluctuations of the o(3 ) order parameter @xmath18 ; the @xmath22 crossovers above @xmath21 have also been described @xcite .
+ ( _ ii _ ) the second possibility involved intermediate non - fermi liquid phases before the large fermi surface metal was reached at sufficiently large @xmath16 . in this case , the o(3 ) order parameter was parameterized in terms of the spinor @xmath1 by @xmath23 the spinor field @xmath1 is the natural variable to describe the loss of magnetic order at @xmath21 , and the non - fermi liquid phases above @xmath24 , and replaces the o(3 ) order parameter @xmath18
. our focus in the present paper will be on the second possibility .
part of our motivation comes from transport measurements @xcite , which show an extended regime of non - fermi liquid behavior as @xmath25 in high magnetic fields .
we shall describe the photoemission spectra at non - zero temperatures on both sides of @xmath24 .
our @xmath26 results are candidates for the pseudogap regime of the cuprates , and relate especially to recent experimental results of meng _ et al . _ @xcite . to complement the approach taken in ref .
, here we will motivate our choice of the non - magnetic non - fermi liquid phases by extending the theory @xcite of the loss of nel order in the insulator at half filling .
although motivated in a theory of metals , @xmath2 also contains ( in principle ) a complete description of the insulating states at half - filling .
crucial to the description of insulators,@xcite are ` hedgehog ' point tunneling events ( ` instantons ' ) in which @xmath18 points radially outwards / inwards from a spacetime point .
these hedgehogs carry berry phases : in @xmath2 the berry phases are expected to appear from the determinant of the gapped fermions integrated out in a hedgehog field for @xmath18 .
all previous treatments of the spin - fermion model have neglected the hedgehog berry phases , and this may well be appropriate under suitable circumstances in certain superconducting states.@xcite in this paper , we wish to focus on regimes and phases where the hedgehog tunnelling events are suppressed . in the insulator ,
the hedgehog suppression is a consequence of quantum interference from the hedgehog berry phases , and leads to interesting new ` deconfined ' phases and critical points.@xcite hedgehog suppression is also possible in certain exotic metallic states ( to be described below ) , where fermionic excitations near a fermi surface lead to a divergence in the hedgehog action.@xcite it is useful to begin our analysis by adapting the phase diagram of ref . describing the doping of an insulating deconfined critical point
see fig .
[ figrkk ] . .
the strength of antiferromagnetic fluctuations in the insulator is controlled by the coupling @xmath16 . at @xmath27
, there is a deconfined critical point @xcite , separating a nel state and a valence bond solid @xcite .
the broken symmetries in these two states survive in metallic fermi liquid state at non - zero @xmath28 . more importantly
, the deconfined critical point broadens into a non - fermi liquid phase , the algebraic charge liquid ( acl ) , which is the focus of attention in the present paper . the conventional large fermi surface fermi liquid
is not shown in the phase diagram above .
it appears at larger @xmath28 , and its phase transitions to the acl and the sdw states were discussed in earlier work @xcite.,width=480 ] in the insulator , at @xmath27 , the transition is between a nel state and a valence bond solid.@xcite this transition is described @xcite by a deconfined critical point , with the order parameter @xmath18 replaced by a complex ` relativistic ' boson @xmath1 which carries charges under an emergent u(1 ) gauge field @xmath29 : this mapping will be reviewed in section [ sec : acl ] . upon moving away from the insulator ,
the fermions @xmath30 are replaced by fermions @xmath31 fermions which do not carry spin , but do carry charge @xmath32 of the u(1 ) gauge field @xmath29 .
we will present the lagrangian of @xmath1 , @xmath31 , @xmath29 which described the phase diagram in fig .
[ figrkk ] in section [ sec : acl ] . in the sdw phase of fig .
[ figrkk ] , the @xmath1 , @xmath31 , @xmath29 theory is formally identical to the spin - fermion model of @xmath33 , @xmath18 .
this is because hedgehogs are strongly suppressed when there is magnetic order .
we explore this connection in appendix [ smallg ] , and in section [ sec : rc ] to obtain the spectral functions of the electrons in the ` renormalized classical ' ( rc ) regime @xcite . the transition out of the sdw phase in fig .
[ figrkk ] is into a fractionalized metal phase , which was dubbed@xcite an algebraic charge liquid ( acl ) .
this transition is in the o(4 ) universality class.@xcite to leading order , the elementary excitations of the acl are simply the constituent fractionalized fields @xmath1 , @xmath31 , @xmath29 , and their interactions have been discussed in previous work.@xcite here , our primary results will be on the fermionic spectrum over a large intermediate length scale near the boundary between the acl and the confining metallic states shown in fig .
[ figrkk ] .
a key characteristic of the deconfined critical point is parametrically large separation between the spin correlation length , @xmath34 , and the confinement scale , @xmath35 , at which hedgehogs proliferate . in the doped system , as argued in ref .
, the fractionalized excitations are already bound into gauge neutral excitations at the scale @xmath34 . among these bound states
are electron - like fermions which carry charge @xmath36 and spin @xmath37 , and so can be directly detected in photoemission experiments .
we will present a phenomenological effective hamiltonian for these excitations in section [ sec : bound ] .
the basic idea @xcite of the mapping to the theory of the acl is to transform to a new set of fermions , @xmath38 with @xmath39 , with their spin components @xmath40 polarized along the direction of the local sdw order .
we perform this rotation with the spacetime dependent su(2 ) matrix @xmath41 so that @xcite @xmath42 we choose @xmath43 so that spin - fermion coupling is only along @xmath44 , and so @xmath45 this relationship is equivalent to @xmath46 now , we parameterize @xmath47 with @xmath48
. we can verify that eq .
( [ rz ] ) yields the usual relation in eq .
( [ nz ] ) .
a crucial feature of the resulting hamiltonian for the @xmath49 and @xmath50 is that it is invariant under a local u(1 ) gauge transformation .
this follows from the invariance of eqs .
( [ cab ] ) and ( [ nz ] ) under the transformation @xmath51 where @xmath52 has an arbitrary dependence on space and time .
note that the @xmath49 carry opposite charges @xmath53 under the u(1 ) gauge transformation ( which is unrelated to the gauge invariance associated with the physical electromagnetic force ) . associated with this
u(1 ) gauge invariance , we will introduce an internal dynamical gauge field @xmath29 in constructing the effective theory .
ref . argued that describing the transition to the large fermi surface fermi liquid required the inclusion of additional degrees of freedom so that the theory had a su(2 ) gauge invariance .
we will not consider this extension here .
however , we expect that simple extensions of the results in section [ sec : bound ] apply also to the su(2 ) acl phases found in ref . .
2 .transformations of the lattice fields under square lattice symmetry operations .
@xmath66 : translation by one lattice spacing along the @xmath28 direction ; @xmath67 : 90@xmath68 rotation about a dual lattice site on the plaquette center ( @xmath69 ) ; @xmath70 : reflection about the dual lattice @xmath71 axis ( @xmath72 ) ; @xmath73 : time - reversal , defined as a symmetry ( similar to parity ) of the imaginary time path integral .
note that such a @xmath73 operation is not anti - linear .
the transformations of the hermitian conjugates are the conjugates of the above , except for time - reversal of fermions @xcite . for the latter , @xmath74 and @xmath75
are treated as independent grassman numbers and @xmath76 ; similarly for @xmath77 , @xmath78 . [ cols="^,^,^,^,^",options="header " , ] the analysis in the following section [ sec : bound ] will be based largely on symmetry , and so it is useful to recall @xcite now how the fields introduced so far transform under symmetry operations .
these are summarized in table [ table0 ] .
as discussed in section [ sec : intro ] , we want to work in the regime where the photon creates bound states between the @xmath79 fermions and the @xmath1 spinons , but the monopole induced confinement has not yet occurred .
thus we are in a fluctuating sdw state with a spin correlation length @xmath34 , but we are interested in phenomena at a scale larger than @xmath34
. however , the confinement of the photons occurs at a scale @xmath35 , and so we will restrict ourselves to the @xmath80 .
we also note the complementary considerations in the work of wen and lee @xcite : they considered a spin liquid model with fermionic spinons and bosonic holons ( in contrast to our bosonic spinons and spinless fermion charge carriers ) , and described spinon - holon bound states in the electron spectral functions .
also , essler and tsvelik @xcite described a model of weakly coupled chains , and considered the bound states of the spinons and holons of the one - dimensional spin liquid on the chains . in our approach , we do not appeal to these spin liquid states , but deal instead with states motivated by the fluctuations in the observed spin density wave order .
the regime @xmath80 was treated in section iv of ref . .
see also the subsection on holon - spinon binding in the appendix of ref . .
here we shall provide a more general treatment , which should also allow for a computation of spectral functions .
let the bound state between the @xmath81 fermions and the @xmath1 be @xmath77 .
the bound state should have the full symmetry of the square lattice , and so we can define a local operator @xmath82 , which creates this bound state centered at the lattice site @xmath83 . as was emphasized in the initial analysis@xcite , there is a second independent bound state , and a consequent doubling of the fermion species .
this is the bound state between the @xmath84 and @xmath85 , which we denote by the local operator @xmath86 . in the ordered nel state
, the sublattice location of a fermion also fixes its spin .
however , when we move to length scales larger than @xmath34 , this is no longer the case because the spin direction of the background nel state has been averaged over .
thus we can view @xmath77 and @xmath78 as fermions that reside preferentially ( but not exclusively ) on the two sublattices , and they separately have an additional degeneracy associated with carrying @xmath0 . more formally , all we will really need are the properties of @xmath87 and @xmath88 under the square lattice symmetry operations : these are summarized in table [ table0 ] , and will form the bases of our analysis below .
the bare electron operator will have a non - zero overlap with both the @xmath77 and @xmath78 fermions .
this will be non - local over the scale @xmath34 .
we approximate this connection from eq .
( [ cab ] ) as @xmath89 where @xmath90 is some quasiparticle renormalization factor depending upon the fermion - spinon bound state wavefunction . in general
, @xmath90 should be non - local over a scale @xmath34 , but have limited ourselves for simplicity to a momentum independent wavefunction renormalization .
note that eq .
( [ cfg ] ) and the symmetry transformations in table [ table0 ] ensure that @xmath30 is invariant under all operations of the square lattice symmetry .
the possible non - local terms in eq .
( [ cfg ] ) can be deduced by the requirements of symmetry .
we now need an effective hamiltonian for the @xmath82 and @xmath91 .
formally , any hamiltonian which is invariant under the symmetry transformations of table [ table0 ] is acceptable ; however , we use simple physical requirements to restrict the large class of possibilities . for the diagonal terms which do not mix the @xmath92 and @xmath93 , we assume ( for simplicity ) that they just inherit the terms for @xmath81 and @xmath84 in @xmath61 in eq .
( [ lpsi ] ) . the mixing between the @xmath92 and @xmath93
is provided by the shraiman - siggia term @xmath65 .
physically this can be understood as the mixing corresponds to hopping between two sublattices ( as @xmath92 and @xmath93 reside preferentially on different sublattices ) , and shraiman - siggia term describes such hopping and associated spin - flipping process@xcite .
these terms are more simply considered in their real space form which are the last two terms in eq .
( [ hop ] ) . combining these terms
, we have the effective hamiltonian @xmath94 where the second line comes from @xmath65 . here
@xmath13 and @xmath95 are renormalized in some unknown manner from the bare hopping matrix elements in @xmath2 .
we can now verify that all the terms above are invariant under the transformations in table [ table0 ] .
the above hamiltonian of bound state can be diagonalized in momentum space .
first , rewrite equation ( [ heff ] ) in momentum space @xmath96 where we parameterize @xmath97 and @xmath98 as @xmath99 here @xmath100 , @xmath101 and @xmath102 are nearest neighbor , next nearest neighbor and next - next - nearest neighbor hopping @xmath13 respectively .
@xmath103 , @xmath104 and @xmath105 are the hopping elements of @xmath106 . here
we only included terms up to third nearest neighbor hopping , which is capable to capture the shape of the fermi surface , but higher order terms can be included in a similar fashion .
@xmath107 is the matrix element of the on - site mixing term @xmath108 , which is also allowed by symmetry . to diagonalize hamiltonian ( [ eq : h - bs ] ) , we change basis to @xmath109 note that @xmath110 has momentum @xmath111 according to the transformation under translation symmetry listed in table [ table0 ] . in the new basis ,
the hamiltonian becomes @xmath112 \label{eq : h - bs - cd}\end{aligned}\ ] ] the spectrum of electron operator @xmath113 can be obtained by diagonalizing the hamiltonian of @xmath114 and @xmath115 fermions , as @xmath113 is related to @xmath114 operator according to equation ( [ cfg ] ) @xmath116 . in the hamiltonian ( [ eq : h - bs - cd ] ) , @xmath114 and @xmath115 fermions have dispersions @xmath117 and @xmath118 respectively , and they are mixed through the @xmath15 term . with the mixing , gaps will open where the fermi surfaces of @xmath114 and @xmath115 fermions intersect and the large fermi surfaces become fermi pockets . in the case of @xmath119 ,
the @xmath115 fermi surface is the same as the @xmath114 fermi surface shifted by @xmath120 ; so the pockets are symmetric under reflection with respect to the magnetic brillouin zone boundary , and therefore centered at @xmath121 . however , with a non - vanishing @xmath98 , the dispersions of @xmath114 and @xmath115 are different , so the pockets are no longer symmetric about the magnetic brillouin zone boundary , and are not necessarily centered at @xmath122 . .
subplot ( a ) shows the fermi surface of @xmath123 and @xmath124 fermions as if there is no mixing term @xmath15 in the hamiltonian ( [ eq : h - bs - cd ] ) .
the red line is @xmath123 and the blue one is @xmath124 .
subplot ( c ) shows pockets like fermi surface of the quasiparticles described by the hamiltonian ( [ eq : h - bs - cd ] ) with @xmath125 . subplot ( b )
shows the same fermi surface with the color representing the weight of electron operator in the quasiparticle excitation .
subplot ( d ) shows the same information as in ( b ) , but by plotting the electron spectral weight at @xmath126 as a function of momentum ; in this plot we manually put in a finite life - time of electron @xmath127 just for visualization purpose .
the dashed line in ( a ) and ( c ) is the boundary of magnetic brillouin zone in the ordered state . ] but with @xmath128 ) and @xmath125 . ] but with @xmath129 ) and @xmath125 . ] but with @xmath130 and @xmath131 . ] to show the qualitative effects of @xmath132 on the shape of fermi pockets , we draw the pockets and electron spectrum functions of some representative choice of @xmath133 in figs . [
fig : poplot]-[fig : poplot4 ] .
the dispersion @xmath134 is chosen with some phenomenological parameters @xmath135 and @xmath136 , and the sdw gap @xmath15 is @xmath137 in figs .
[ fig : poplot]-[fig : poplot3 ] and @xmath138 in fig .
[ fig : poplot4 ] . in fig .
[ fig : poplot ] , a negative @xmath107 shifts the hole pockets outwards , and makes the shape of the hole pockets asymmetric .
the inner side becomes more curved and the outer side more flat . in fig .
[ fig : poplot2 ] , a negative @xmath103 does not shift the position of the hole pockets significantly , but also makes the shape of the hole pockets asymmetric in a similar way as in fig .
[ fig : poplot ] . combining the effect of these two parameters
, we can move the hole pockets inward and make their inner side more curved than the outer side with a position @xmath107 and a negative @xmath103 , as shown in fig .
[ fig : poplot3 ] . along the fermi pockets ,
the fermionic quasiparticles are a mixture of @xmath123 and @xmath139 fermions , while experiments only probe electron spectrum weight .
the weight of electron operator in the quasiparticle is calculated through diagonalizing equation ( [ eq : h - bs - cd ] ) and are plotted in fig .
[ fig : poplot]-[fig : poplot3 ] . in fig .
[ fig : poplot4 ] , we show a plot with larger @xmath15 ( @xmath131 ) so that the anti - nodal electron pocket is completely gapped . in all the cases ,
the inner half of the hole pockets have higher electron quasiparticle weight , since the inner part is primarily made of @xmath123 fermion and the outer part is primarily made of @xmath124 fermion , and electron operator is proportional to @xmath123 .
this bound state theory can be compared to the yang , rice , and zhang ( yrz ) model of hole pockets@xcite . in their theory a phenomenological green s function for the electron in the underdoped state is proposed , based upon `` spin liquid '' physics , and the spectral function derived from the green s function has hole pockets inside the diamond brillouin zone . from our hamiltonian ( [ eq : h - bs - cd ] ) , the green s function of the electron is @xmath140 compare this to the green s function in the yrz model @xcite : @xmath141}.\ ] ] the two results are quite similar . in the present form , instead of the @xmath142-wave gap @xmath143 , our @xmath15 does not have a momentum dependence .
however , this distinction is an artifact of the simple choices made in our form of @xmath144 : we can clearly include more non - local mixing between the @xmath92 and @xmath93 fermions . the more important distinction between our model and yrz lies in the physical input in the fermion spectrum .
the yrz model relies on ` pairing correlations ' implicit in some underlying spin liquid state .
in contrast , we do not assume any pairing , but fluctuating local antiferromagnetic order .
we will describe the influence of pairing on our spectral functions in a future paper .
we note earlier arguments @xcite that the pairing amplitudes are especially strong on the electron pockets , and this may be a contributing factor to removing the electron - like fermi surfaces in figs . [
fig : poplot2 ] and [ fig : poplot3 ] .
recent arpes experiments @xcite reveal that there are hole pockets in the underdoped regime , and the center of the pockets are inside the first magnetic brillouin zone .
in general the shape of hole pockets seen in the experiments can be fit to our model . as shown in fig .
[ fig : poplot]-[fig : poplot4 ] , the weight of electron operator on the outer part of the pockets is tiny and may be hard to see in experiments .
we close this section by remarking on the status of luttinger s theorem in our theory of the electron spectral function .
these issues were discussed in refs . , where we argued that the theorem applied to the sum of the @xmath77 , @xmath78 and the @xmath74 fermi surfaces .
here we will focus on the electron - like @xmath77 , @xmath78 , and will drop the @xmath145 contributions to the present discussion ; the latter will amount to shift in the effective doping level @xmath28 . under this assumption
the total number of @xmath77 and @xmath78 per site is @xmath146 , and so is the number of @xmath147 and @xmath148 , since the canonical transformation in equation ( [ eq : cd ] ) preserves particle number . on the other hand ,
our theory was applied to the doped holes or electrons in the background of a fluctuating antiferromagnet , and each such charge carrier must occupy one state within the fermi surface . counting hole ( electron)-like fermi surfaces as negative ( positive ) , then for a doped antiferromagnet with hole density @xmath28 , the fermi surfaces in figs .
[ fig : poplot]-[fig : poplot4 ] should enclose a total area of @xmath149 ( the last factor of 2 is from spin degeneracy ) . in our present effective hamiltonian model ,
we have found it more efficient to treat not just the fermionic excitations near the electron and hole pockets , but across the entire brillouin zone .
we found that this method was very convenient in treating the contraints imposed by symmetry , without prejudicing the final locations of the fermi surfaces .
such an extension should not be accompanied by any fundamental change in the many body quantum state , and hence can not modify the statement of luttinger s theorem . because the luttinger constraint only controls the electron density modulo 2 per unit cell ,
we therefore conclude that our lattice effective model obeys @xmath150 note that this differs from the value in the conventional fermi liquid phase , in which case the total electron density is @xmath151 .
this difference is acceptable here because we are discussing a phase with topological order , which has an emergent u(1 ) gauge excitation @xcite . in the discussion of yrz
model@xcite a different form of luttinger s theorem is used : there the total number of particles given by green s function ( [ eq : yrz ] ) is given by the total area enclosed by the contours where the green s function changes sign .
this includes not only the fermi pockets on which the green s function diverges , but also the contour where the green s function vanishes . for our green s function of
@xmath114 bound states in equation ( [ eq : gc ] ) this extra luttinger surface of zeros is exactly the original fermi surface of @xmath115 bound states without the mixing between @xmath114 and @xmath115 .
therefore the total area enclosed by both fermi pockets and surface of zeros equal to total number of @xmath114 and @xmath115 states minus the number of @xmath115 states , and the result is the total number of @xmath114 states .
so this form of luttinger s theorem still holds true in our model , although it is not relevant to the physical doping , which is related to the area of fermi pockets only .
apart from the electron - like bound states described in the previous section , the other low - lying excitations in the acl are the gapless u(1 ) photons . here
we discuss how these two low - lying sectors of the theory couple to each other .
our first task is search for terms coupling the @xmath77 , @xmath78 to the photons , while being invariant under all the transformations in table [ table0 ] .
it is useful to do this on the lattice , as many of the operations involve details of the lattice symmetry .
the @xmath77 , @xmath78 are gauge - invariant , and so will couple only to the field strengths : on the lattice , we define these as @xmath152 thus @xmath153 resides at the center of each square lattice plaquette , while @xmath154 reside on the links .
after some searching , we found a single term which is linear in the field strengths which fulfills the needed criteria : @xmath155\ ] ] this coupling contributes to the self - energy correction of the bound states . at the lowest order ,
the self - energy correction is represented by the feynman diagram in fig .
[ fig : self - energy ] , or @xmath93 fermion propagator and the wiggly line represents electric field propagator .
the interaction vertex is given by equation ( [ eq : gamma ] ) .
this diagram is evaluated in equation ( [ eq : self - energy ] ) . ]
the action of u(1 ) gauge field has contributions from both the @xmath145 fermions and @xmath1 field . assuming that there exist fermi surfaces of the @xmath145 fermions , integrating them out screens the fluctuation of the @xmath156 component , and gives the following terms for the action of the transverse components of @xmath157 @xmath158 here @xmath159 is the fermi momentum of the @xmath145 fermi surface .
we consider two possibilities of the @xmath1 fields . at the magnetic ordering critical point , integrating out @xmath1 field gives rise to the following action @xmath160 on the other hand , if the @xmath1 field is gapped as in the acl , integrating it out gives rise to terms proportional to @xmath161 , which are higher order than those in eq ( [ eq : saf ] ) . in summary , the u(1 )
gauge field action can be written as @xmath162 where the polarizition function @xmath163 is given by @xmath164 for imaginary frequencies
@xmath165 . using the action in equation and we can evaluate the feynman diagram in fig .
[ fig : self - energy ] .
@xmath166 % \end{split}\ ] ] here we consider a simple one - band model for the bound state with a quadratic dispersion @xmath167 .
the polarization factor at the end can be simplified as @xmath168 = \frac{4p^2q^2(1-\cos^2\theta)}{p^2+q^2 - 2pq\cos\theta}\ ] ] where @xmath169 is the angle between @xmath170 and @xmath171 .
we are interested in the life - time of quasiparticles near the fermi surface , so we consider the imaginary part of self - energy at the quasiparticle pole @xmath172 . without losing generality , we consider the case of @xmath173 . at zero temperature , energy of the intermediate fermion state
must be greater than zero but smaller than @xmath165 , so its momentum must be inside the shell of @xmath174 .
the imaginary part of self - energy is @xmath175 near the fermi surface we can linearize the dispersion relation @xmath176 , and @xmath177 .
we change the integrated variable from @xmath178 to @xmath179 , @xmath180 . in the limit of @xmath181 , or @xmath182 , the integral can be simplified @xmath183 in both of the cases considered in eqs . and
, the gauge field spectrum function @xmath184 when @xmath185 , and becomes less singular at @xmath186 . as a result
the integral in eq . has the following behavior @xmath187 thus the bound state fermion has a fermi liquid - like damping .
this conclusion is different from essler and tsvelik s work on a model of weakly coupled chains @xcite .
their pocket states arose from binding between the one - dimensional holons and spinons , and the coupling to the one - dimensional spin fluctuations led to a self - energy proportional to @xmath188 .
our model is genuinely two - dimensional , and has an emergent u(1 ) gauge field who presence is also the key to the violation of the luttinger theorem in eq .
( [ luttth ] ) .
they explain the violation of the luttinger theorem by zeros of the green s function @xcite ; the connection of these zeros to our work was discussed below eq .
( [ luttth ] ) .
we will now consider the regime at small non - zero @xmath189 above the antiferromagnetically ordered state present at @xmath190 , the renormalized classical ( rc ) regime . here , we expect the o(3 ) vector formulation of @xmath2 in eq .
( [ esf ] ) to be equivalent to the spinor / u(1 ) gauge theory formulation of @xmath191 in eq .
( [ lacl ] ) .
this is explored at @xmath192 in appendix [ smallg ] , where we show that the small @xmath16 expansions of the two theories match with each other .
thus , while there are differences in the nature of the non - magnetic phases at @xmath193 in the two cases ( as reviewed in section [ sec : intro ] ) , the @xmath190 phase and its low @xmath189 rc regimes are the same .
the rc regime was studied using the o(3 ) formulation by vilk and tremblay @xcite , and using a u(1 ) gauge theory similar to @xmath191 by borejsza and dupuis @xcite . here
we shall also use @xmath191 , and expand upon these earlier results .
we compute the electron spectral function as a convolution of the @xmath1 propagator with the free @xmath79 propagator containing pocket fermi surfaces obtained from @xmath61 . for the @xmath1 propogator we use the simple
damped form motivated by studies in the @xmath194 expansion.@xcite @xmath195 where the dimensionless constant @xmath196 determines the damping constant , and the `` mass '' @xmath197 is determined by the `` large @xmath198 '' equation @xmath199\ ] ] where @xmath200 is the energy scale which determines the deviation from the quantum critical point .
the damping in eq .
( [ gz ] ) is a simple interpolation form which is constant with the expected behavior at the quantum critical point , and with the rc behavior of eq .
( 7.28 ) in ref .
actually , we have neglected the power - law pre - factor of the exponential in ( 7.28 ) .
determining the damping from numerical studies motivated by the @xmath201 expansion @xcite we expect @xmath202 . in fig .
[ fig : conv ] and [ fig : conv2 ] we plot electron spectrum function obtained as a convolution of the @xmath1 and fermion spectrum functions at finite temperature . here
fermion dispersion relation is the same as the one we used in the plots in section [ sec : bound ] .
in contrast to fig . [
fig : poplot]-[fig : poplot4 ] , the pockets are symmetric with respect to the magnetic brillouin zone boundary , and hole pockets are centered at @xmath203 .
again , the inner half of the pockets has higher spectrum weight , as the quasiparticle on outer half of the pockets is primarily made of fermion quasiparticles at @xmath204 .
the broadening of the spectrum comes from two factors : first , the @xmath205 boson has a finite life - time @xmath206 ; second , the convolution is done at finite temperature , so the energy of fermion and boson excitation does not have to be exactly zero , but can vary by energies of order @xmath189 .
the second factor dominantes in the rc regime because @xmath207 , and @xmath196 is of order 1 .
so the electron spectrum is basically the pocket fermion dispersion broadened by a linewidth of order @xmath189 : see fig .
[ fig : conv ] and fig . [ fig : conv2 ] . and the @xmath74 fermions .
in the plot we chose the following parameters : @xmath208 , @xmath209 and the sdw gap @xmath210 .
the white dashed line shows the boundary of the diamond magnetic brillouin zone of the commensurate @xmath120 afm order.,width=480 ] and the @xmath74 fermions . in the plot we chose the following parameters : @xmath208 , @xmath209 and the sdw gap @xmath131 .
the white dashed line shows the boundary of the diamond magnetic brillouin zone of the commensurate @xmath120 afm order.,width=480 ] earlier work @xcite had also obtained a linewidth of order @xmath189 in the rc regime by very different methods . as we noted above
, the present fermi surface locations are symmetric with respect to the magnetic brillouin zone boundary , unlike those in figs .
[ fig : poplot]-[fig : poplot4 ] .
however , as shown in ref . ,
the shraiman - siggia term in eq .
( [ lss ] lifts this symmetry
. we can compute perturbative corrections to leading order in @xmath15 , and these will contribute spectral weight of width @xmath189 which is asymmetric about the magnetic brillouin zone boundary .
we conclude by reviewing the different routes @xcite by which the spin fermion model @xmath2 in eq .
( [ esf ] ) can lose antiferromagnetic order , and their implications for the photoemission spectrum at finite temperatures .
the first , more conventional , route is that there is a single direct transition at @xmath21 to a fermi liquid with a large fermi surface .
then in the @xmath189-@xmath16 plane , we have the conventional @xcite quantum disordered ( qd ) , quantum critical ( qc ) , and renormalized classical ( rc ) regions .
in qd region at small @xmath189 for @xmath211 , the electron spectrum will show quasiparticle peaks with a fermi liquid linewidth @xmath212 . in the qc region near @xmath213 ,
the spectrum will again show weight along the large fermi surface , but with large anomalous linewidths near the ` hot spots ' : these are points along the large fermi surface connected by the ordering wavevector @xmath8 . finally , in the rc region at small @xmath189 for @xmath190 , we have the behavior described in section [ sec : rc ] : the spectrum has ` small ' fermi pockets which are centered at the antiferromagnetic brillouin zone boundary , and the quasiparticle peaks have a width @xmath214 ; examples of such spectra were shown in figs .
[ fig : conv ] , [ fig : conv2 ] .
this paper mainly considered a more exotic route @xcite towards loss of antiferromagnetic order in the spin - fermion model .
this route is possible if topological `` hedgehog '' tunnelling events are suppressed as the transition , as they are at magnetic - disordering transitions in the insulator ( see fig .
[ figrkk ] ) .
then the transition at @xmath21 is to a non - magnetic non - fermi liquid acl , with the spinless fermions @xmath145 and complex bosonic spinons @xmath1 as elementary excitations , interacting ( in the simplest case ) via an emergent u(1 ) gauge force .
again , around this critical point at @xmath21 , we can define the corresponding rc , qc , and qd regions .
the electron spectrum in the rc region is just as in the first case above , as discussed in section [ sec : rc ] and appendix [ smallg ] .
the novel spectrum in the qd region was the main focus of section [ sec : bound ] .
one contribution to the spectral weight comes from the convolution of the deconfined spinons and the @xmath74 .
this leads to incoherent spectral weight in ` arc'-like regions which were described earlier in ref . .
however , it has been argued @xcite that the spinons and @xmath145 form electron - like bound states , and these were described in more detail in section [ sec : bound ] .
we found that the bound states lead to pocket fermi surfaces , as shown in figs [ fig : poplot]-[fig : poplot4 ] .
an important feature of these spectra are that the pockets are not centered at the point @xmath203 on the antiferromagnetic brillouin zone boundary .
in fact , this zone boundary plays no special role , and there are no symmetry relations on the quasiparticle dispersions across it .
we note that recent photoemission observations @xcite show features related to this qd region .
we did not address the qc region here , but it should mainly have the incoherent arc spectra , as described in ref . .
we also noted a similarity between our acl qd results , and the yrz model @xcite in section [ sec : bound ] .
the assumptions of yrz are very different from ours , as they depart from a paired fermionic spinons in a spin - liquid state .
the fermi surfaces in the yrz model to not obey the conventional luttinger theorem , as is the case in our model .
however , such a violation must be accompanied by gauge forces reflecting the topological order in such a state @xcite : these gauge fields do not appear in their formulation . also , we dealt mainly with the influence of local antiferromagnetic order on the electron spectrum .
we have not included pairing effects in our computations yet , or the transition to superconductivity ; these are issues we hope to address in forthcoming work .
we thank j. carbotte , m. metlitski , and a .- m .
tremblay for useful discussions .
this research was supported by the national science foundation under grant dmr-0757145 , by the fqxi foundation , and by a muri grant from afosr .
this appendix will compare the properties of the lagrangian @xmath2 defined in eq .
( [ esf ] ) , with the ` fractionalized ' lagrangian @xmath191 in eq .
( [ lacl ] ) .
we will work at @xmath192 in the limit of small @xmath16 , where both models have long range sdw order , and are expected to be essentially identical .
we will compare the two models here by computing the on - shell matrix element for the decay of a spin - wave in the sdw state into a fermionic particle hole pair .
we will work in the `` diamond '' brillouin zone associated with antiferromagnetic ordering with wavevector @xmath215 . in this zone
we define @xmath216 all expressions below are implicitly in this diamond brillouin zone .
we perform the small @xmath16 expansion for the order parameter by the following parameterization in terms of the spin - wave field @xmath217 : @xmath218 then the lagrangian for @xmath18 is @xmath219 + \frac{g^2}{8 } \frac{|\phi|^2}{(2- g |\phi|^2 ) } \left ( \phi^\ast \partial_\mu \phi + \phi \ , \partial_\mu \phi^\ast \right)^2 \label{lnp}\ ] ] and can be analyzed as usual in an expansion in @xmath16 . for the fermion sector , we diagonalize the @xmath220 at @xmath221 by introducing fermion operators @xmath222 , and ( we replace the electron index @xmath223 by @xmath53 ) @xmath224 where @xmath225 , @xmath226 are real and obey @xmath227 .
we choose @xmath228 , @xmath229 , and then @xmath230 the fermion lagriangian at @xmath221 is @xmath231 where @xmath232\ ] ] finally , the non - linear couplings between the spin waves and the fermions is given by @xmath233_{\bf q } \bigl [ \nonumber \\ & ~&~~~~~~~~~~~~~~~~(u_{{\bf k}+{\bf q } } v_{{\bf k } } - u_{\bf k } v_{{\bf k } + { \bf q } } ) ( \gamma_{1-}^\dagger ( { \bf k } + { \bf q } ) \gamma_{1 + } ( { \bf k } ) - \gamma_{2-}^\dagger ( { \bf k } + { \bf q } ) \gamma_{2 + } ( { \bf k } ) ) \nonumber \\ & ~&~~~~~~~~~~~~+ ( u_{{\bf k}+{\bf q } } u_{{\bf k } } + v_{\bf k } v_{{\bf k } + { \bf q } } ) ( \gamma_{1-}^\dagger ( { \bf k } + { \bf q } ) \gamma_{2 + } ( { \bf k } ) + \gamma_{2-}^\dagger ( { \bf k } + { \bf q } ) \gamma_{1 + } ( { \bf k } ) ) \bigr ] + \mbox{h.c .
} \nonumber \\ & ~ & + g \lambda \sum_{{\bf k } , { \bf q } } \sum_p \left [ |\phi|^2 \right]_{\bf q } \bigl [ ( u_{{\bf k}+{\bf q } } v_{{\bf k } } + u_{\bf k } v_{{\bf k } + { \bf q } } ) ( \gamma_{1p}^\dagger ( { \bf k } + { \bf q } ) \gamma_{1p } ( { \bf k } ) - \gamma_{2p}^\dagger ( { \bf k } + { \bf q } ) \gamma_{2p } ( { \bf k } ) ) \nonumber \\ & ~&~~~~~~~~~~~~+ p ( u_{{\bf k}+{\bf q } } u_{{\bf k } } - v_{\bf k } v_{{\bf k } + { \bf q } } ) ( \gamma_{1p}^\dagger ( { \bf k } + { \bf q } ) \gamma_{2p } ( { \bf k } ) + \gamma_{2p}^\dagger ( { \bf k } + { \bf q } ) \gamma_{1p } ( { \bf k } ) ) \bigr]\end{aligned}\ ] ] now we can obtain the self - energy of the @xmath217 spin - wave excitation to order @xmath16 : @xmath234 \nonumber \\ & & ~~~~~~~~~~~~~+ 2 g \lambda \sum_{{\bf k } } 2 u_{\bf k } v_{\bf k } ( f(e_1({\bf k } ) ) - f(e_2 ( { \bf k } ) ) ) \end{aligned}\ ] ] as expected , we have @xmath235
. we can also estimate the on - shell decay rate at @xmath192 : @xmath236 , as long as @xmath237 .
we can also use @xmath238 to compute the fermion spectral density at order @xmath16 . for the case of a single particle in the insulator
, it seems to me that the result has the same form as ( 2.47 ) in ref . .
for small @xmath16 , we parametrize the @xmath1 field of the cp@xmath55 model as @xmath239 then the lagrangian for @xmath1 is @xmath240 if we integrate over @xmath29 in the gauge @xmath241 , we reproduce the action in eq .
( [ lnp ] ) .
the integral over @xmath29 also produces a coupling between @xmath242 and the u(1 ) current over the @xmath243 fermions .
note that this is a quartic coupling , a bilinear in @xmath217 coupling to a bilinear in @xmath243 . in our discussion in section [ app : sf ] , the dominant coupling between the fermions and the spin - waves was the term in @xmath238 which was _ linear _ in @xmath217 .
such terms arise here exclusively from @xmath65 .
we evaluated the on - shell matrix elements from such terms for the decay of a @xmath217 spin wave into a particle hole pair of the @xmath244 ( and also to the @xmath245 ) . from the first term in @xmath65 we have the matrix element @xmath246 for the time - derivative term in @xmath65
we replace the frequency by its on - shell value @xmath247 to obtain the matrix element @xmath248 the sum of ( [ m1 ] ) and ( [ m2 ] ) is equal to the matrix element obtained in section [ app : sf ] , which is @xmath249 a. kanigel , m. r. norman , m. randeria , u. chatterjee , s. suoma , a. kaminski , h. m. fretwell , s. rosenkranz , m. shi , t. sato , t. takahashi , z. z. li , h. raffy , k. kadowaki , d. hinks , l. ozyuzer , and j. c. campuzano , nature physics * 2 * , 447 ( 2006 ) .
j. chang , y. sassa , s. guerrero , m. mnsson , m. shi , s. pailh s , a. bendounan , r. mottl , t. claesson , o. tjernberg , l. patthey , m. ido , m. oda , n. momono , c. mudry and j. mesot , new j. physics * 10 * , 103016 ( 2008 ) .
jianqiao meng , guodong liu , wentao zhang , lin zhao , haiyun liu , xiaowen jia , daixiang mu , shanyu liu , xiaoli dong , wei lu , guiling wang , yong zhou , yong zhu , xiaoyang wang , zuyan xu , chuangtian chen , and x. j. zhou , nature * 462 * , 335 ( 2009 ) .
n. p. armitage , d. h. lu , c. kim , a. damascelli , k. m. shen , f. ronning , d. l. feng , p. bogdanov , x. j. zhou , w. l. yang , z. hussain , p. k. mang , n. kaneko , m. greven , y. onose , y. taguchi , y. tokura , and z .- x .
shen , phys .
b * 68 * , 064517 ( 2003 )
. a. f. bangura , j. d. fletcher , a. carrington , j. levallois , m. nardone , b. vignolle , p. j. heard , n. doiron - leyraud , d. leboeuf , l. taillefer , s. adachi , c. proust , and n. e. hussey , phys .
100 * , 047004 ( 2008 ) . c. jaudet , d. vignolles , a. audouard , j. levallois , d. leboeuf , n. doiron - leyraud , b. vignolle , m. nardone , a. zitouni , ruixing liang , d. a. bonn , w. n. hardy , l. taillefer , and c. proust , phys . rev
. lett . * 100 * , 187005 ( 2008 ) .
d. leboeuf , n. doiron - leyraud , j. levallois , r. daou , j .- b .
bonnemaison , n. e. hussey , l. balicas , b. j. ramshaw , r. liang , d. a. bonn , w. n. hardy , s. adachi , c. proust , and l. taillefer , nature * 450 * , 533 ( 2007 ) .
j. chang , r. daou , cyril proust , david leboeuf , nicolas doiron - leyraud , francis laliberte , b. pingault , b. j. ramshaw , ruixing liang , d. a. bonn , w. n. hardy , h. takagi , a. antunes , i. sheikin , k. behnia , and louis taillefer , arxiv:0907.5039 ( 2009 ) .
t. senthil , a. vishwanath , l. balents , s. sachdev , and m. p. a. fisher , science * 303 * , 1490 ( 2004 ) ; t. senthil , l. balents , s. sachdev , a. vishwanath , and m. p. a.
fisher , phys .
b * 70 * , 144407 ( 2004 ) . | we describe fluctuating two - dimensional metallic antiferromagnets by transforming to a rotating reference frame in which the electron spin polarization is measured by its projections along the local antiferromagnetic order .
this leads to a gauge - theoretic description of an ` algebraic charge liquid ' involving spinless fermions and a spin @xmath0 complex scalar .
we propose a phenomenological effective lattice hamiltonian which describes the binding of these particles into gauge - neutral , electron - like excitations , and describe its implications for the electron spectral function across the entire brillouin zone .
we discuss connections of our results to photoemission experiments in the pseudogap regime of the cuprate superconductors . |
x - ray diffraction is a powerful method for investigating structures of molecules .
x - ray crystallography has become the standard tool for identifying the structure of large molecules and proteins @xcite .
recent developments in x - ray sources have opened new opportunities @xcite for imaging membrane proteins and other macromolecules that can not be crystallized @xcite .
one possibility is single - molecule imaging , where one molecule at a time is probed by an intense x - ray pulse @xcite , which subjects the molecule to severe damage @xcite .
a series of diffraction patterns has to be collected and classified according to the molecular orientation to get full structural information from randomly oriented molecules @xcite .
new iterative phase - retrieval algorithms for noncrystalline specimens have been developed to invert the diffraction data @xcite .
alternative approaches with laser - aligned molecules have been proposed , where an ensemble of molecules , rather than a single molecule , is exposed to x - ray pulses @xcite . in this approach the radiation dose for each molecule
is strongly reduced and stays well below the critical value for damage @xcite .
three - dimensional information on the single - molecule structure can be gained from a well - aligned ensemble by capturing many two - dimensional diffraction patterns .
the ability to accumulate the signal over a large number of x - ray pulses reduces the radiation dose further .
notice that alignment , rather than orientation , of the molecules is expected to be sufficient for the reconstruction of the molecular structure @xcite .
nevertheless , various methods have been proposed to orient molecules @xcite by the linear stark effect @xcite , the ac stark effect @xcite , or two - color fields @xcite .
much theoretical and experimental research has been done on laser - induced alignment @xcite . with nonresonant laser frequencies
, an induced dipole moment can be created that couples back to the laser electric field and forces the molecule to be aligned .
earlier work has focused on one - dimensional alignment of the most polarizable axis by using linearly polarized light @xcite .
the demonstration of three - dimensional alignment @xcite , which requires an asymmetric - top molecule and an elliptically polarized pulse ( or two linearly polarized pulses ) opens the door for probing the three - dimensional structure .
the coulomb explosion technique has been exploited to detect three - dimensional alignment @xcite , where rotational temperatures down to 1 k have been accomplished @xcite .
depending on the duration of the laser pulse , @xmath1 , relative to the rotational period of the molecule , @xmath2 , the alignment dynamics can be classified into three distinct regimes . in the limits of adiabatic ( @xmath3 ) and impulsive ( @xmath4 ) alignment ,
the dynamics can be described analytically @xcite .
impulsive alignment reveals the quantum mechanical nature of this process by showing alignment revivals after the laser pulse is over @xcite . in the adiabatic limit ,
the alignment dynamics follow the laser pulse shape @xcite .
no analytic solution exists in the intermediate regime ( @xmath5 ) , and the time propagation of the molecular ensemble in the presence of the laser pulse has to be performed numerically @xcite . especially the quantum mechanical description of three - dimensional alignment @xcite has proved to be numerically expensive @xcite .
a general theory of x - ray diffraction from laser - aligned symmetric - top molecules was developed in ref .
applications to adiabatically aligned molecules may be found in refs .
reference @xcite describes calculations on x - ray scattering from impulsively aligned molecules , exploiting the alignment revivals for probing field - free molecular structure .
it has been shown that , for a symmetric - top molecule containing a single heavy scatterer , a holographic algorithm can successfully reconstruct the molecular structure from an x - ray scattering pattern @xcite . in this article , we discuss x - ray scattering from an ensemble of rigid , asymmetric - top molecules aligned three - dimensionally by elliptically polarized light at finite rotational temperature .
we point out the symmetries in the quantum mechanical theory of three - dimensionally aligned molecules that can be used to significantly reduce the numerical time propagation .
we restrict our analysis to electronic and vibrational ground - state configurations and neglect deformations .
our approach allows us to investigate x - ray diffraction from molecules in all three alignment regimes ( adiabatic , impulsive , and intermediate ) . in order to probe molecular structure of gas - phase molecules by x - ray scattering
, the degree of alignment must be rather high for sufficient resolution .
this favors the adiabatic alignment approach @xcite with low - temperature molecules @xcite . in sec .
[ sec.theory ] , we present the theory and inherent symmetries of the hamiltonian underlying x - ray diffraction from laser - aligned molecules .
section [ sec.methods ] focuses on the numerical implementation and computational efficiency , and basic ideas are presented of the phase - retrieval algorithm that is used for structure reconstruction . in sec .
[ sec.results ] , the three - dimensional alignment dynamics and their impact on the reconstruction are demonstrated using the example of the organic molecule naphthalene .
we conclude with a discussion of the feasibility and limitations of x - ray diffraction from laser - aligned gas - phase molecules .
atomic units @xcite are employed throughout , unless otherwise noted .
the hamiltonian for x - ray diffraction from laser - aligned molecules is @xcite @xmath6 where @xmath7 describes the field - free rotational motion of a molecule , @xmath8 is the hamiltonian of the free quantized x - ray fields , and @xmath9 and @xmath10 describe the interactions of a molecule with the laser and x - ray field , respectively .
the electronic and vibrational states of the molecule are omitted , since we assume that the molecule remains in its electronic and vibrational ground state throughout our discussion .
the x - ray field and its interaction with the molecule is described in a quantized manner .
the laser field is formulated as a classical field .
the eigenstates of the noninteracting system , i.e. , @xmath11 , are @xmath12 with @xmath13 where @xmath14 are the rotational eigenstates of an asymmetric - top molecule @xcite and @xmath15 are the x - ray fock states @xcite .
the density matrix of the whole system is @xmath16 where the gas - phase molecules @xcite are described by a canonical ensemble @xcite , @xmath17 is the statistical weight , and @xmath18 denotes the initial distribution of all the occupied field modes @xcite . with including both interactions , @xmath9 and @xmath10 , the states of eq .
( [ eq.eigenstates_overall ] ) are no longer eigenstates of the system .
however , each state @xmath19 can be written as @xmath20^{j_1\tau_1m_1;\{n_1\}}_{j_0\tau_0 m_0;\{n_0\ } } } { \left|j_1\tau_1m_1;\{n_1\}\right>},\end{aligned}\ ] ] where the expansion coefficients @xmath21^{j_1\tau_1m_1;\{n_1\}}_{j_0\tau_0 m_0;\{n_0\}}}$ ] statisfy the initial condition @xmath22^{j_1\tau_1m_1;\{n_1\}}_{j_0\tau_0 m_0;\{n_0\ } } } & = & \delta_{j_0j_1}\delta_{\tau_0\tau_1}\delta_{m_0m_1}\delta_{\{n_0\},\{n_1\}}. \end{aligned}\ ] ] the equation of motion for the expansion coefficients in the interaction picture ( subscript i ) reads @xmath23^{j_2\tau_2m_2;\{n_2\}}_{j_0\tau_0 m_0;\{n_0\ } } } & = & \hskip-2ex \sum_{j_1\tau_1m_1,\{n_1\}}\hskip-2ex { \left
< j_2\tau_2m_2;\{n_2\}\right| } \hat h_{l}(t ) + \hat h_{x } { \left|j_1\tau_1m_1;\{n_1\}\right > } \\&&\times\nonumber e^{i(e^\text{rot}_{j_2\tau_2}-e^\text{rot}_{j_1\tau_1}+e^x_{\{n_2\}}-e^x_{\{n_1\}})t } \ { \left[c_i(t)\right]^{j_1\tau_1m_1;\{n_1\}}_{j_0\tau_0 m_0;\{n_0\}}}.\end{aligned}\ ] ] in the following , we assume we found @xmath24^{j_1\tau_1m_1}_{j_0\tau_0 m_0}}$ ] , the solution for the laser - only problem , i.e. , @xmath25 . note that in the laser - only hamiltonian no x - ray field is involved and we will drop the x - ray field indices in its solution .
the interaction between the laser - aligned molecules and the x - ray field is taken into account by first - order perturbation theory .
the solution of eq .
( [ eq.eom1 ] ) becomes @xmath26^{j_4\tau_4m_4;\{n_4\}}_{j_0\tau_0 m_0;\{n_0\ } } } & = & -i\hskip-3ex\sum_{{\genfrac{}{}{0pt}{}{j_1\tau_1m_1 , j_2\tau_2m_2}{j_3\tau_3m_3 } } } \hskip-2ex { \left[c_{l , i}(t)\right]^{j_4\tau_4m_4}_{j_3\tau_3 m_3 } } \int_{-\infty}^t\!\!dt'\ e^{i(e^\text{rot}_{j_2\tau_2}-e^\text{rot}_{j_1\tau_1}+e^x_{\{n_4\}}-e^x_{\{n_0\}})t ' } \\&&\times\nonumber { \left[c^{-1}_{l , i}(t')\right]^{j_3\tau_3m_3}_{j_2\tau_2 m_2 } } { \left
< j_2\tau_2m_2;\{n_4\}\right| } \hat h_{x } { \left|j_1\tau_1m_1;\{n_0\}\right > } { \left[c_{l , i}(t')\right]^{j_1\tau_1m_1}_{j_0\tau_0 m_0 } } .\end{aligned}\ ] ] the expectation values of interest can be calculated by @xmath27 \\\nonumber&= & \label{eq.expval_detail } \hskip-3ex\sum_{{\genfrac{}{}{0pt}{}{j_2\tau_2 m_2,j_1\tau_1m_1,j'\tau'm'}{\{n_2\},\{n_1\},\{n'\},\{n''\}}}}\hskip-4ex w_{j'\tau'}\ \rho^x_{\{n'\},\{n''\ } } { \left < j_2\tau_2 m_2;\{n_2\}\right|}\hat o{\left|j_1\tau_1m_1;\{n_1\}\right > } \\\nonumber&&\times e^{i(e^\text{rot}_{j_2\tau_2}-e^\text{rot}_{j_1\tau_1}+e^x_{\{n_2\}}-e^x_{\{n_1\}})t } { \left[c_i(t)\right]^{j_1\tau_1m_1;\{n_1\}}_{j'\tau ' m';\{n'\ } } } { \left[c^*_i(t)\right]^{j_2\tau_2 m_2;\{n_2\}}_{j'\tau ' m';\{n''\}}},\end{aligned}\ ] ] from which the alignment signals ( cf . sec .
[ ssec.cos2 ] ) and the x - ray scattering probability ( cf .
[ ssec.angu_prob_dist ] ) can be derived .
we followed here the strategy that is laid out in ref .
@xcite for x - ray diffraction from symmetric - top molecules and adapt it to the asymmetric - top case .
the structures of @xmath28 and @xmath29 are the same for symmetric - top and asymmetric - top molecules .
the most dramatic changes for asymmetric - top molecules happen in the laser - only system .
therefore , we focus for the rest of our discussion just on the laser - only system . in the following subsections
the structure of the field - free hamiltonian @xmath7 and the laser molecule interaction matrix @xmath30 will be investigated . for this purpose
it is convenient to stay in the schrdinger picture . in sec .
[ ssec.eom ] we return to the equation of motion of the laser - only system and point out the symmetries of its solutions @xmath24^{j'\tau'm'}_{j\tau m}}$ ] .
the measure of three - dimensional alignment is described in sec .
[ ssec.cos2 ] .
the theory section closes with exploiting the symmetries in the angular density distribution and the diffraction signal . assuming that structural deformation of the molecule may be neglected
, we treat the molecules as rigid rotors .
the corresponding field - free hamiltonian reads @xcite @xmath31,\end{aligned}\ ] ] where @xmath32 are the rotational constants associated with the principal axes of inertia , @xmath33 are the cartesian components of the angular momentum operator in the molecular frame ; and @xmath34 are the spherical basis components . for symmetric - top ( @xmath35 ) and asymmetric - top ( @xmath36 ) rotors , the angular momentum @xmath37 and its projection on the space - fixed @xmath38-axis , @xmath39 , are conserved .
the angular - momentum projection on the molecular @xmath40-axis , @xmath41 , is conserved only for symmetric - top molecules .
a new quantum number @xmath42 , replacing @xmath41 , must be introduced for asymmetric - top rotors diagonalizing the field - free hamiltonian [ cf .
( [ eq.eigenenergy_hrot ] ) ] .
note , the rotational energies @xmath43 are independent of the quantum number @xmath39 .
hence , each energy level @xmath43 is ( @xmath44)-fold degenerate .
it is possible to express the asymmetric - top eigenstates @xmath14 as a superposition of the analytically known symmetric - top eigenstates @xmath45 @xcite , @xmath46}_{k\tau } { \left|jkm\right > } , \\ \label{eq.eigenstate_symm } { \left<\left.\phi,\theta,\chi\right|jkm\right > } & = & \sqrt{\frac{2j+1}{8\pi^2}}d^{*[j]}_{m , k}(\phi,\theta,\chi),\end{aligned}\ ] ] where the coefficients @xmath47}_{k\tau}$ ] are real and @xmath48 are the three euler angles connecting the space - fixed laser frame ( l ) with the principal axes of inertia in the molecular reference frame ( m ) .
@xmath49}_{m , k}(\phi,\theta,\chi)$ ] is the complex conjugate of the wigner d - matrix with angular momentum @xmath37 .
unfortunately , it is not possible to find a general relation between the @xmath42 and @xmath41 classification @xcite .
there are two symmetries incorporated between the symmetric - top ( @xmath41 ) and asymmetric - top ( @xmath42 ) classification .
the first symmetry decouples states with even @xmath41 from states with odd @xmath41 , since @xmath7 is a linear combination of @xmath50 and @xmath51 [ eq . ( [ eq.hrot ] ) ] . the state class with even @xmath41 is labeled @xmath52 , the one with odd @xmath41
is labeled @xmath53 .
asymmetric eigenstates inherit this separation and are superpositions of just even or odd @xmath41 states .
the invariance of @xmath7 under the substitutions @xmath54 and @xmath55 leads to the second symmetry , the wang symmetry @xcite , where asymmetric - top eigenstates decompose in symmetric and antisymmetric linear combinations , @xmath56}_{k\tau}\big[{\left|jkm\right>}+(-1)^\tau { \left|j(-k)m\right>}\big ] , \qquad\end{aligned}\ ] ] where @xmath57 denotes the parity of @xmath42 , @xmath58 thus , overall there are four separate state classes , which are summarized in table [ tab.rotational_class ] .
.[tab.rotational_class ] rotational classes of asymmetric - top rotors . [ cols= " <
, < " , ] three - dimensional alignment of naphthalene is obtained by employing elliptically polarized light .
the @xmath38 and @xmath59 axes of the space - fixed frame are defined by the major and minor axes , respectively , of the ellipse that characterizes the elliptically polarized light .
following ref .
@xcite , we choose the ratio between the field components using @xmath60 which maximizes the three - dimensional alignment of naphthalene .
perfect three - dimensional alignment with respect to the space - fixed frame is achieved , when @xmath61 .
random orientation corresponds to @xmath62 for all angles .
the typical distance between two neighboring carbon atoms in naphthalene is 1.4 . in order to resolve the atomic structure ,
the resolution must be smaller than this value .
imperfect alignment limits the resolution . to build a connection between the resolution and the alignment in terms of @xmath63 ,
it is helpful to view the angle @xmath64 as an opening angle within which the residual motion of the atoms takes place ( cf .
[ fig.naphthalene ] ) .
hence , the length characterizing the smallest resolvable structure is @xmath65 where @xmath66 is the linear dimension of the molecule ( measured from the center of mass of the molecule ) .
the degree of alignment needed for a resolution of 1 for the outermost carbon atom ( @xmath67 ) is @xmath68 . for a given resolution , smaller opening angles @xmath64
are required when the molecules become larger . in the adiabatic regime at
sufficiently low temperature the maximum alignment for a linear rotor ( @xmath69 ) is given by @xmath70^{[2]}_0 $ ] @xcite .
the polarizability @xmath71^{[2]}_0 $ ] is approximately proportional to @xmath72 , and @xmath73 scales approximately as @xmath74 @xcite . the smallest resolvable dimension
@xmath75 is thus proportional to @xmath76 and might be expected to decrease with increasing molecular size .
note , however , that the temperature required to suppress thermal effects ( @xmath77 ) also decreases as a function of the molecular size .
additionally , in order to remain in the adiabatic regime , the laser pulse duration must increase with increasing molecular size . in practice
this means the laser intensity will decrease . for large molecules
it is more realistic to consider the high temperature limit ( @xmath78 ) , where the degree of alignment is a competitive interplay between rotational temperature and coupling strength @xmath79 ; more precisely , @xmath80 @xcite . by using the same scaling arguments , we find @xmath81{ktr}$ ] .
the expected resolution is now reversed and increases with molecular size and rotational temperature .
consequently , larger molecules can be resolved less precisely .
mk ( solid ) and @xmath82 k ( dotted ) .
the pump laser shown in gray has a peak intensity of @xmath83 tw/@xmath0 , a width of @xmath84 ps , and an ellipticity of @xmath85 .
the upper panel pictures the alignment dynamics on the time scale of the pulse duration . the lower panel is a close - up view of the region highlighted in the upper graph . at @xmath86 ps ,
nonadiabatic oscillation enhances the alignment for all axes at @xmath87 mk and is ideal for a 1-ps x - ray probe pulse . ] in fig .
[ fig.cos2_t10mk ] , the alignment dynamics of naphthalene at 10 mk and 1 k , respectively , are shown .
we assumed a gaussian laser pulse with a peak intensity of @xmath83 tw/@xmath0 and a pulse duration of @xmath84 ps ( fwhm ) . at a rotational temperature of @xmath87 mk ,
almost all naphthalene molecules are in the rotational ground state .
the rotational period of naphthalene at such a low temperature , @xmath88 ps , is of the order of @xmath84 ps , suggesting that the alignment dynamics are quasi - adiabatic @xcite .
this is consistent with the observation that the overall alignment follows the laser pulse shape and , in addition , clear nonadiabatic features ( fast oscillations ) are visible .
one can make use of the nonadiabatic behavior by probing the molecules with an x - ray pulse of 1 ps duration , which is fast enough to resolve the nonadiabatic oscillations .
for instance , at @xmath86 ps the alignment is transiently enhanced in all three dimensions ( cf .
[ fig.cos2_t10mk ] ) . at @xmath82
k , the rotational period of naphthalene is small in comparison to @xmath1 , so the alignment dynamics are adiabatic and fast oscillations are significantly suppressed . as a consequence of the increased thermal motion , the maximum degree of alignment at @xmath82 k is clearly reduced .
the diffraction signal @xmath89 [ cf . eq .
( [ eq.diff_signal ] ) ] collected over the x - ray pulse duration reflects a pulse - averaged , effective alignment @xmath90 where @xmath91 is the normalized x - ray flux .
we assume a gaussian temporal envelope for the x - ray pulse , with a full width at half maximum of @xmath92 . in fig .
[ fig.cos2_eff ] , @xmath93 is shown as a function of @xmath92 .
we may conclude from fig .
[ fig.cos2_eff ] that the effective alignment of the body - fixed axes decreases rapidly when the x - ray pulse duration is longer than the laser pulse duration .
an enhancement in the effective alignment is visible for @xmath94 ps at @xmath87 mk , where nonadiabatic oscillations are not suppressed and can be resolved by the x - ray pulse . , for @xmath87 mk ( solid ) and
@xmath82 k ( dotted ) .
the pump laser has a peak intensity of @xmath83 tw/@xmath0 , a width of @xmath84 ps , and an ellipticity of @xmath85 .
the x - ray pulses are centered at @xmath95 ps . ] when @xmath96 , the influence of @xmath92 is rather weak and the degree of alignment is mainly affected by the rotational temperature , @xmath97 .
figure [ fig.cos2_temp ] shows the effective alignment of the molecular axes as a function of @xmath97 for @xmath98 ps and @xmath99 ps . below 0.25 k , the best aligned axis is the body - fixed @xmath100 axis in the space - fixed @xmath101 direction for @xmath102 ps , despite the fact that no laser field is applied in this direction .
the strong alignment of the @xmath100 axis translates to well - aligned naphthalene molecules in the polarization plane of the laser ( @xmath103 plane ) . within the plane ,
the body - fixed @xmath40 axis is always more strongly aligned in the @xmath38 direction than the @xmath104 axis is aligned in the @xmath59 direction .
( recall that @xmath105 and @xmath106 . ) for higher temperatures , the body - fixed @xmath40 axis is the most strongly aligned axis , since the alignment of the @xmath100 axis is affected by the more rapidly decreasing alignment of the @xmath104 axis .
ps ( solid ) and @xmath99 ps ( dotted ) .
the pump laser has a peak intensity of @xmath83 tw/@xmath0 , a width of @xmath84 ps , and an ellipticity of @xmath85 .
the x - ray pulses are centered at @xmath95 ps . ] as may be seen in fig .
[ fig.cos2_eff ] , the effective alignment @xmath107 of all molecular axes is at least 0.96 for 10 mk naphthalene molecules probed by x - ray pulses shorter than @xmath108 ps .
hence , the smallest resolvable dimension limited by residual pendular motion is @xmath109 .
only the rotational ground state is initially occupied , and the minimum intensity required to accomplish 1 resolution is @xmath83 tw/@xmath0 .
dissociation and ionization play only a minor role at this intensity , but become important at higher intensities @xcite . with an effective alignment of @xmath110 at 1 k and @xmath99 ps
, the smallest resolvable dimension @xmath75 is 1.73 , which is larger than the distance between neighboring carbon atoms . in fig .
[ fig.diffpic_front ] , diffraction patterns @xmath89 of three - dimensionally aligned naphthalene are shown for perfect alignment as well as for imperfect alignment at different temperatures and x - ray pulse widths .
the laser parameters are the same as for fig .
[ fig.cos2_t10mk ] .
the laser polarization plane is assumed to be the @xmath103 plane so that the molecules are aligned as illustrated in fig .
[ fig.naphthalene ] .
the diffraction patterns in fig .
[ fig.diffpic_front ] are two - dimensional planar slices through the three - dimensional @xmath111 space , taken at @xmath112 .
( due to the curvature of the ewald sphere @xcite , experimental scattering patterns do not correspond to exactly planar slices through @xmath111 space , but this is of no consequence here . )
the signal strength falls rapidly for high @xmath113 .
therefore , to highlight the structure at high momentum transfers , the scattering intensities are shown on a logarithmic scale . the effective alignment in fig .
[ fig.diffpic_front ] decreases clockwise , from @xmath114 for perfect alignment to @xmath110 for naphthalene molecules at 1 k probed by a 100-ps x - ray pulse . by increasing the x - ray pulse width from 1 to 100 ps , the effective alignment at @xmath87 mk decreases from @xmath115 to @xmath116 [ cf .
[ fig.diffpic_front](b ) and ( d ) ] , and is significantly smaller than the impact of temperature rise from 10 mk to 1 k [ cf .
[ fig.diffpic_front](c ) and ( d ) ] . of naphthalene .
( a ) perfect alignment . ( b )
@xmath87 mk , @xmath98 ps , x - ray pulse centered at @xmath86 ps ( cf .
[ fig.cos2_t10mk ] ) .
( c ) @xmath82 k , @xmath117 ps , x - ray pulse centered at @xmath95 ps .
( d ) @xmath87 mk , @xmath99 ps , x - ray pulse centered at @xmath95 ps .
for the alignment pulses in ( b ) , ( c ) , and ( d ) , we assumed @xmath83 tw/@xmath0 and @xmath84 ps . ]
although the basic features of the x - ray scattering pattern for perfectly aligned naphthalene ( fig .
[ fig.diffpic_front]a ) are preserved in the scattering patterns for laser - aligned naphthalene [ figs .
[ fig.diffpic_front](b ) , ( c ) , and ( d ) ] , the contours are washed out and the contrast between maxima and minima is less pronounced with decreasing alignment .
incoherent averaging for laser - aligned naphthalene renders the diffraction patterns more cylindrically symmetric with respect to the @xmath101 direction , which limits the accessible structural information particularly at high momentum transfer .
side views of naphthalene in @xmath111 space are shown in fig .
[ fig.diffpic_side ] , where fig .
[ fig.diffpic_side](a ) depicts the long side ( @xmath118 ) and fig .
[ fig.diffpic_side](b ) the short side ( @xmath119 ) of naphthalene .
both display strong similarities to multi - slit diffraction patterns , consistent with the planar structure of naphthalene being well aligned in the @xmath103 plane . .
( b ) diffraction signal of naphthalene for @xmath119 .
other parameters are the same as for fig .
[ fig.diffpic_front](b ) . ]
the achievable real - space resolution depends on two aspects . on the one hand ,
the smallest resolvable dimension @xmath75 is a function of the degree of alignment [ eq .
( [ eq.align_res ] ) ] . on the other hand ,
the pixel size of the real - space structure reconstruction is determined by the maximum momentum transfer @xmath120 in the diffraction signal @xmath89 . here , @xmath121
@xmath122 , corresponding to a pixel size of 0.5 . the approximate range in momentum space within which the assumption of coherent scattering holds may be defined by @xmath123 . in fig .
[ fig.diffpic_highq ] , the diffraction signals of fig .
[ fig.diffpic_front ] are shown up to @xmath124 @xmath122 .
the respective ranges defined by @xmath125 are highlighted .
it is not possible to increase the resolution of the real - space structure reconstruction by choosing @xmath120 much greater than @xmath125 .
in fact , numerical tests have indicated that data beyond @xmath125 can lead to poor convergent structures .
, except @xmath124 @xmath122 .
the momentum transfer @xmath125 ( see the text ) is highlighted by a solid green circle in ( b)(d ) . ] for 12-kev x - ray photons , a real - space pixel size of 0.5 requires the detection of photons scattered up to @xmath126 with respect to the x - ray propagation axis . the azimuthally averaged differential x - ray scattering cross section per naphthalene molecule for @xmath127
@xmath122 is @xmath128 .
since the largest distance between carbon atoms in naphthalene is @xmath129 , the area of a pixel in momentum space must not exceed @xmath130 @xmath131 . at a scattering angle of @xmath126 ,
this corresponds to a solid angle @xmath132 . for a molecular beam width of 1 mm and an x - ray focus area of @xmath133 @xmath134m@xmath135
, it has been demonstrated that the number of molecules in the interaction volume can be as large as @xmath136 @xcite .
hence , requiring a minimum of five scattered photons per pixel @xcite , the estimated acquisition time for one diffraction pattern is around 70 seconds with an x - ray fluence of @xmath137 photons / pulse / mm@xmath135 and a repetition rate of 1 khz .
we applied the phase - retrieval algorithm described in sec .
[ ssec.phase-retrieval ] to the naphthalene diffraction signals shown in fig .
[ fig.diffpic_front ] . the maximum momentum transfer of @xmath138@xmath122 in fig .
[ fig.diffpic_front ] corresponds to a pixel size of 0.5 , which is of the order of the structure we want to resolve in naphthalene . as a consequence ,
each pixel encodes a lot of structure information .
calculations have shown that the reconstruction routine becomes sensitive to the set of initial random phases .
therefore , we follow the spirit of ref .
@xcite and average over reconstructions obtained for 100 different sets of initial phases to define a quality criterion , which we apply subsequently to all 100 reconstructions to select the most meaningful results and average over these selected reconstructions .
our quality criterion is defined as follows : ( 1 ) all pixels of the unfiltered averaged result that have at least 20% of the maximum electron density are selected to build a density core region .
( 2 ) the quality criterion uses the density core region and selects only the reconstructions that have at least 50% of their total electron density within this density core region .
the filtered averages for the x - ray scattering patterns of fig .
[ fig.diffpic_front ] are displayed in fig .
[ fig.reconstruct_front ] . in our calculations
, we employed a grid spacing in @xmath111 space of 0.16 @xmath122 , corresponding to a maximum object size 8 times larger than the size of naphthalene . .
the pixel size is 0.5 .
( a ) perfect alignment .
( b ) @xmath87 mk , @xmath98 ps , x - ray pulse centered at @xmath86 ps .
( c ) @xmath82 k , @xmath99 ps , x - ray pulse centered at @xmath95 ps .
( d ) @xmath87 mk , @xmath99 ps , x - ray pulse centered at @xmath95 ps . ]
figure [ fig.reconstruct_front ] illustrates that with better effective alignment in the diffraction pattern more structural information can be reconstructed . in the perfect alignment case ( cf .
[ fig.reconstruct_front](a ) , the positions of the atoms can be resolved . a comparison with the perfect alignment reconstruction in fig .
[ fig.diff_res_perfect](b ) , where the pixel size is 0.125 and no averaging over several initial random phases was performed , illustrates that the reconstruction of the carbon positions is more difficult when the pixel size is comparable with the size of the atoms .
incoherent averaging for imperfectly aligned molecules affects the effective resolution further , since the assumption of coherent scattering limits the usable diffraction signal to @xmath139 . as a consequence ,
@xmath120 should be of the order of @xmath125 and , therefore , the pixel size in the reconstruction is limited by the degree of effective alignment .
all results displayed in fig .
[ fig.reconstruct_front ] recover the two - ring structure of naphthalene , indicating an effective resolution below 2 .
especially the two carbon atoms connecting both carbon rings are distinctly visible .
figure [ fig.reconstruct_front](b ) and [ fig.reconstruct_front](d ) show the reconstruction for naphthalene at 10 mk with @xmath140ps and @xmath141ps , respectively .
both have a similar effective alignment and reconstruction . in both cases ,
the electron density peaks on the two - ring structure indicate the positions of all carbon atoms and are consistent with the theoretically expected effective resolutions , which are below the typical distance between neighboring carbon atoms ( 1.4 ) . the predicted effective resolution at 1 k ( @xmath142
) is sufficient to resolve the overall structure of naphthalene , which can be seen in fig .
[ fig.reconstruct_front](c ) . for the central region of naphthalene ,
the effective resolution improves and makes it possible to identify the two central carbon atoms at 1 k.
we have theoretically studied the theory of x - ray diffraction from asymmetric - top molecules that are three - dimensionally aligned by elliptically polarized light and have discussed a phase - retrieval method in order to reconstruct molecular structure from the x - ray scattering pattern .
the interaction of the gas phase molecules with the laser and x - ray fields was studied in the density matrix formalism .
we have assumed rigid rotor molecules .
symmetries in the time - dependent hamiltonian and in its solutions as well as in the angular density distribution were exploited and used to improve numerical efficiency . in combination with further computational aspects , a significant enhancement in numerical efficiency has been achieved .
a feasibility study of x - ray diffraction from three - dimensionally laser - aligned molecules has been performed using the organic molecule naphthalene .
we have linked the degree of alignment to an effective resolution and have studied the impact of x - ray pulse duration and temperature on the diffraction patterns .
a phase - retrieval reconstruction was performed on diffraction patterns taken for different effective alignments .
the enhancement of incoherent averaging due to imperfect alignment destroys structural information and hinders the recovery of detailed atomic configurations within the molecule .
the reconstruction of naphthalene confirms our discussion that the degree of alignment is a good indicator for the achievable resolution in the reconstruction . to decode structures on an atomic length scale , high degree of alignment in all three molecular axes
must be achieved .
molecules have to be cooled down to a few kelvin or even sub - kelvin temperatures so rotational motion is sufficiently reduced .
experiments have shown that it is feasible to cool molecules to 1 k @xcite . to accomplish rotational temperatures well below 1
k is a great challenge and would open the opportunity to image large gas - phase molecules at atomic resolution .
the problem of incoherent averaging over a finite range of different orientations has to be further addressed , especially in the case where molecular symmetries do not coincide with the symmetries of the diffraction pattern imposed by alignment . with the ability to detect molecular structure , x - ray scattering from gas phase molecules
can be used to study torsion effects and laser - induced deformations that are expected to occur during alignment in the presence of intense laser pulses @xcite . by systematically varying the delay time between pump ( laser ) and probe ( x - ray ) pulse , it is possible to follow molecular motion on an ultrafast time scale .
of particular interest is the study of physical and chemical processes in the presence of intense laser fields , which simultaneously provide the required alignment for imaging of reactions with atomic resolution in space and time .
we thank cassandra hunt for helpful comments on the manuscript .
this work was supported by the office of basic energy sciences , u.s .
department of energy under contract no .
de - ac02 - 06ch11357 .
in the density matrix , statistical weights , @xmath143 , define the relative number of particles in an ensemble that are in a given quantum mechanical state for a well - defined temperature @xmath97 .
the nuclear spin statistical weights @xmath144 are based on symmetry arguments and represent the number of the allowed nuclear spin states for a given rotational state @xmath14 .
the overall symmetry @xmath145 of the total molecular wave function is independent of rotational or nuclear spin states and determined by the spin statistic theorem @xcite .
however , @xmath145 is also a direct product of symmetries of the different quantum states @xcite , @xmath146 where @xmath147 is the rotational state symmetry and @xmath148 is the nuclear spin state symmetry .
molecules are in their electronic and vibrational ground states and their contributions can be omitted in eq .
( [ eq.mol_symm ] ) .
the representation of symmetries depends on the symmetry group of the molecule .
the molecular symmetry group of asymmetric - top molecules without an inversion center is @xmath149 or lower @xcite .
the rotational symmetry group of an asymmetric - top rotor is always @xmath150 ( isomorphic to @xmath149 ) @xcite .
if there is no common symmetry class between @xmath147 and @xmath148 , @xmath151 .
in other words , molecular symmetries come only into play when molecular rotations are identical to particle exchanges .
all symmetry representations are written as linear combinations of irreducible representations ( irrep ) of @xmath150 in table [ tab.character_table ] .
in addition the character table for the nuclear spin and rotational states are shown .
the four molecular rotations and particle exchanges defining @xmath150 are : @xmath152 .
cccccp1excc & & & + & @xmath52 & @xmath153 & @xmath154 & @xmath155 & & @xmath37 even & @xmath37 odd ' '' '' + @xmath145 & 1 & 1 & 1 & 1 & & ' '' ''
+ @xmath156 & 1 & 1 & @xmath157 & @xmath157 & & @xmath158 & @xmath159 + @xmath160 & 1 & 1 & @xmath161 & @xmath161 & & @xmath159 & @xmath158 + @xmath162 & 1 & -1 & @xmath157 & @xmath161 & & @xmath163 & @xmath164 + @xmath165 & 1 & -1 & @xmath161 & @xmath157 & & @xmath164 & @xmath163 + @xmath148 & @xmath166 & @xmath167 & @xmath167 & @xmath167 & & ' '' '' the characters of the overall wave function @xmath168 can be derived from the pauli principle @xcite and correspond to the overall sign changes induced by the molecular rotations .
first , we assume all carbon atoms in naphthalene are @xmath169c with nuclear spin 0 . by remembering only odd permutations of half - integer particles ( fermions ) change the sign of the overall wave function @xcite
, we find all group operation leave the overall sign for naphthalene unchanged , since all hydrogen permutations are even .
hence , @xmath170=1,\
\forall p\in v$ ] and @xmath171 .
the rotational symmetry classes , in which each rotational state can be classified ( cf . sec .
[ ssec.hrot ] ) , coincide with the irreps of @xmath150 @xcite .
however , whether or not @xmath37 is even or odd defines which irrep corresponds to which symmetry class ( cf . table [
tab.character_table ] ) .
the sum of characters over all possible nuclear spin states @xmath172 $ ] for each operation @xmath173 can be derived by the formula @xcite @xmath174 & = & \prod_{g_p}\ , ( 2s_{g_p}+1)\ \text{sgn}(p_{g_p})^{2s_{g_p}},\end{aligned}\ ] ] where the atom group @xmath175 is defined by the minimum number of identical atoms in the molecule such that after the atom permutation @xmath173 each atom in @xmath175 is replaced by an atom of @xmath175 .
@xmath176 is the sign of the permutation @xmath173 within in the atom group @xmath175 , which is @xmath177 or @xmath178 depending whether the permutation is even or odd in @xmath175 .
the smallest atom group consist of one atom , which does not change position under the operation @xmath173 .
@xmath179 is the nuclear spin of the atoms in @xmath175 . imposing eq .
( [ eq.mol_symm ] ) , nuclear spin and rotational symmetries are linked together .
( for naphthalene : @xmath180 )
. the multiplication of symmetry representations is done by multiplying their characters .
the nuclear spin statistical weights @xmath144 are the number of irreps in @xmath148 such that the product with a given rotational class is @xmath145 . as an example : when @xmath181 and j is even , then @xmath182 and only the product @xmath183 @xcite fulfills eq .
( [ eq.mol_symm ] ) .
since there are 60 @xmath163 irreps in @xmath148 , the nuclear spin statistical weight is 60 . for all rotational states of naphthalene ,
the nuclear spin statistical weights are @xmath184
the squares of the matrix elements of the rotation matrix @xmath186 , i.e. @xmath185 , give a complete picture of the three - dimensional alignment . knowing four @xmath185 is sufficient to describe all , where at least two axes of each reference frame have to be involved in the four @xmath185 .
our choice is @xmath187 , since it is one of the least computationally expensive choices .
the @xmath185-matrices are diagonal in @xmath39 or @xmath41 when @xmath188 or @xmath189 , respectively .
only the matrix @xmath190 is nondiagonal in @xmath41 and @xmath39 .
each @xmath185 can be written in terms of @xmath191}_{mk}(\phi,\theta,\chi)$ ] @xcite , where the matrix elements of the wigner d - functions are given in refs .
the matrix elements of the four matrices that we have chosen are : @xmath192 \bigg ) , \\ { \left < jkm\right|}\cos^2\theta_{za}{\left|j'k'm'\right > } & = & \frac{1}{3 } \delta_{jj'}\delta_{kk'}\delta_{mm ' } - \sqrt{\frac{1}{6}}\sqrt{\frac{2j+1}{2j'+1}}{\left<\left.jm;20\right|j'm'\right > } \\\nonumber & & \times \bigg ( \sqrt{\frac{2}{3}}{\left<\left.j , k;2,0\right|j',k'\right > } - \big[{\left<\left.j , k;2,2\right|j',k'\right>}+{\left<\left.j , k;2,-2\right|j',k'\right>}\big ] \bigg ) , \\
{ \left < jkm\right|}\cos^2\theta_{xa}{\left|j'k'm'\right > } & = & \frac{1}{3 } \delta_{jj'}\delta_{kk'}\delta_{mm ' } + \frac{1}{4}\sqrt{\frac{2j+1}{2j'+1 } } \\\nonumber & & \times \bigg ( \sqrt{\frac{2}{3}}{\left<\left.j , m;2,0\right|j',m'\right > } - \big[{\left<\left.j , m;2,2\right|j',m'\right>}+{\left<\left.j , m;2,-2\right|j',m'\right>}\big ] \bigg ) \\\nonumber & & \times \bigg ( \sqrt{\frac{2}{3}}{\left<\left.j , k;2,0\right|j',k'\right > } - \big[{\left<\left.j , k;2,2\right|j',k'\right>}+{\left<\left.j , k;2,-2\right|j',k'\right>}\big ] \bigg ) . \ ] ] m. r. howells , t. beetz , h. n. chapman , c. cui , j. m. holton , c. j. jacobsen , j. kirz , e. lima , s. marchesini , h. miao , d. sayre , d. a. shapiro , j. c. h. spence , and d. starodub , journal of electron spectroscopy and related phenomena * 170 * , 4 - 12 ( 2009 )
. h. n. chapman , a. barty , s. marchesini , a. noy , s. p. hau - riege , c. cui , m. r. howells , r. rosen , h. he , j. c. h. spence , u. weierstall , t. beetz , c. jacobsen , and d. shapiro , j. opt .
a * 23 * , 1179 ( 2006 ) . | we theoretically and numerically analyze x - ray scattering from asymmetric - top molecules three - dimensionally aligned using elliptically polarized laser light .
a rigid - rotor model is assumed .
the principal axes of the polarizability tensor are assumed to coincide with the principal axes of the moment of inertia tensor .
several symmetries in the hamiltonian are identified and exploited to enhance the efficiency of solving the time - dependent schrdinger equation for each rotational state initially populated in a thermal ensemble . using a phase - retrieval algorithm ,
the feasibility of structure reconstruction from a quasi - adiabatically - aligned sample is illustrated for the organic molecule naphthalene .
the spatial resolution achievable strongly depends on the laser parameters , the initial rotational temperature , and the x - ray pulse duration .
we demonstrate that for a laser peak intensity of 5 tw/@xmath0 , a laser pulse duration of 100 ps , a rotational temperature of 10 mk , and an x - ray pulse duration of 1 ps , the molecular structure may be probed at a resolution of 1 . |
recently , there has been a growing interest in the study of interfaces with non - trivial geometry .
such interfaces arise in a variety of situations including domain walls in random magnets,@xcite fluid invasion in porous media@xcite , spreading on heterogeneous surfaces@xcite , membranes and vesicles in biology,@xcite and epitaxial growth in materials science.@xcite in connection with such problems it is natural to ask whether quantum effects play a significant role . for static properties
it is well established that in nonrandom systems , such as a spin @xmath0 antiferromagnet , with only a nearest neighbor exchange interaction , @xmath10 , there are various regimes . near the critical temperature at @xmath11 ,
thermal fluctuations are dominant . in the ordered phase ,
as long as @xmath12 , quantum effects due to the finiteness of @xmath0 are unimportant . for @xmath13
one is in a quantum regime , where the quantum statistics of spin waves and their interactions leads to dependences on @xmath0 and @xmath14 not present in the classical ( @xmath15 ) limit .
heuristic arguments indicate that quantum effects could influence the nature of domain walls in spin systems . in the picture in which the trajectory of the domain wall is likened to the trajectory of a particle in space as a function of time , quantum effects
cause a smearing out of the trajectory .
for such a system analytic and numerical work is obviously very difficult .
accordingly , we have been led to carry out a program of analytic work for quantum domain walls in one dimension . for this purpose
we consider domain walls at zero temperature @xmath14 in the spin @xmath0 ising model ( with nearest - neighbor exchange coupling @xmath10 ) in a transverse field , @xmath5 , whose hamiltonian is @xmath16 where @xmath17 is a quantum spin @xmath0 operator at site @xmath18 . to study domain walls in this model we introduce boundary conditions in which the spin at one end of the chain is fixed to be `` up '' and that at the other end is fixed to be `` down . ''
this model has some interest in its own right .
for @xmath19 its properties can , in principle , be related to those of the associated free fermion model obtained via the jordan - wigner@xcite transformation .
however , with domain - wall boundary conditions , this relation is not easy to implement .
accordingly , we approach the analysis of the properties of domain walls in this model via perturbation theory for @xmath20 and more generally via numerical solution for the wave functions and energies of the ground state and the low - lying excited states . needless to say , some of the properties of this model in one dimension can not be extrapolated to higher dimensional systems . however , in most cases , one can safely say which features can be so extrapolated and which can not .
we may summarize our results for the one dimensional model as follows .
the magnetization profile has a width of order the length of the chain .
the low - lying excited states comprise a manifold of `` particle '' states , which results when the center of the wall propagates from site to site .
these results are easily understood within perturbation theory in @xmath21 . in the classical limit
the width of the wall is of order the correlation length , i. e. it is of order a lattice constant , as long as one is far from the critical regime at @xmath22 , above which long range order disappears . in the classical limit and for small @xmath21
, we evaluate a barrier energy , analogous to the peierls - nabarro energy,@xcite which prevents the free motion of the domain wall .
when the correlation length becomes very large this barrier energy becomes small and since it is harder to calculate in this limit , we did not attempt such calculations
. it would be interesting to calculate this energy for a quantum system , but in the present case , since the width of the wall is of order @xmath7 , we can say that this barrier energy vanishes in the thermodynamic limit . also , in the large @xmath0 limit , we find that the quantum hopping of the domain wall from one site to the next is actually analogous to a tunneling process , so that the hopping matrix element is not of order @xmath5 , as it is for the low spin case , but is now of order @xmath23 , where @xmath24 .
it would be of some interest to recover this result within a field theory of one space and one time dimension to describe such a quantum effect .
briefly , this paper is organized as follows . in sec .
ii we define the class of models we will analyze in one dimension . in sec .
iii we give numerical results for the magnetization profile for @xmath19 , @xmath25 and @xmath26 and compare these to analytic results we obtain using perturbation theory in the small @xmath21 limit . in sec .
iv we give analytic results for large @xmath0 .
the classical results for @xmath27 are given both in the continuum limit ( i.e. , when the wall is very broad ) and in the discrete limit ( when the wall is very narrow ) .
here we also analyze the quantum system for large @xmath0 . in sec .
v we give various results concerning the nature of the energy spectrum in the presence of a domain wall .
we give numerical and analytic results for the nature of the particle spectrum caused by the matrix element which allows the wall to hop from one site to the next . finally , in sec .
vi we summarize the conclusions to be drawn from our work .
the first one - dimensional model we consider is the spin @xmath0 ising model in a transverse field , with the hamiltonian @xmath28 this formulation of the model has the advantages that a ) the domain wall energy ( for @xmath29 ) is independent of the value of @xmath0 and is equal to @xmath10 , and b ) the mean - field transition temperature @xmath30 ( at which significant correlations begin to develop between neighboring spins ) is of order @xmath10 independent of @xmath0 . to discuss domain walls , the spins @xmath31 and @xmath32
will be fixed by boundary conditions , as discussed below .
thus @xmath7 is the number of `` active '' spins in the chain .
we will often use the notation @xmath33 .
we will also consider the
yz " model in a transverse field , for which the hamiltonian is @xmath34 the most important difference between these two models is that the interaction term ( proportional to @xmath35 ) in @xmath36 allows the spins to tip away from the @xmath4-axis . for @xmath29 and with free - end boundary conditions , the y - z " model with @xmath37 has the continuous @xmath38 symmetry instead of the discrete @xmath39 symmetry of the ising model which results when @xmath40 .
in fact , we recall the ground state phase diagram@xcite of this more general model , shown in fig . 1 for spin @xmath41 , where one sees ( at zero temperature ) the disordered phase ( d ) , the ordered ferromagnetic ising phase ( f ) , and the ordered oscillatory phase ( o ) .
in both ordered phases the spontaneous magnetization @xmath42 is non - vanishing ( where @xmath43 denotes the thermodynamic average at temperature @xmath14 ) .
however , the ( connected ) two - point correlation function , @xmath44 , behaves as follows@xcite for @xmath45 : @xmath46 where the correlation length @xmath47 is a known function of @xmath48 and @xmath35 in each of the three phases , @xmath49 is a constant and @xmath50 .
when going from phase f to phase o , there is a new diverging length @xmath51 , which is not simply related to a gap in the spectrum of @xmath36 ( as is @xmath47 ) , but indicates the wavelength at which the correlation function at large distances oscillates under an exponentially decaying envelope . to discuss the width and energy of a domain wall
, we shall work with `` up - down '' boundary conditions , in which we require @xmath52 we can then study the profile of the wall by evaluating @xmath53 where @xmath54 denotes the ground state . as we discuss in more detail below , in order to obtain a spin profile for a _
quantum _ system , antiperiodic boundary conditions which introduce a boundary coupling @xmath55 _ can not _ be used in a naive way . however , for a classical system such antiperiodic boundary conditions will prove convenient .
the energy of the domain wall , @xmath56 , is defined to be @xmath57 , where @xmath58 is the energy with `` up - down '' or antiperiodic boundary conditions and @xmath59 that with `` up - up '' or periodic boundary conditions .
we begin by presenting numerical results for the profile function @xmath62 for the ferromagnetic phase f. for general spin we are constrained to rather small systems , since the number of quantum states is @xmath63 for @xmath7 sites ( excluding the fixed boundary spins ) .
so why not use translational symmetry to reduce the number of states ?
indeed , that would be possible for the special case of antiperiodic boundary conditions ( with @xmath64 ) mentioned in the previous section .
these boundary conditions are peculiar since @xmath65 commutes with the modified translation operator @xmath66 where @xmath67 is the usual translation operator and @xmath68 changes the sign of @xmath69 .
in addition , @xmath65 can be decomposed ( for any @xmath70 ) into a block diagonal form .
the corresponding states are said to be in the even or odd sector , respectively , depending on their parity under spin reversal .
furthermore , @xmath71 in that case and one should consider instead the quantity @xmath72 where @xmath54 and @xmath73 are the ground states in the even and odd sectors ( with respect to the ising symmetry ) of the model , respectively .
it is well - known that at least for _ periodic _ boundary conditions this defines@xcite a sensible finite - lattice approximation to the order parameter ( see ref . and references therein ) .
however , for @xmath64 , even for @xmath7 finite , there are two _ degenerate _ states @xmath73 , @xmath74 in the odd sector , one of which gives rise to the desired wall profile , while the profile of the other one is essentially flat . on the other hand , for @xmath75 , one does get a wall profile , but its shape depends on the value of @xmath76 .
we avoid these problems by choosing the `` up - down '' boundary conditions defined above in eq .
( [ bc ] ) .
furthermore , since we are interested in situations far from criticality , the convergence of the finite - size data with @xmath77 is usually quite rapid so that the moderate sizes achieved are sufficient for our purposes . the determination of the lowest eigenstates of the hamiltonians @xmath78 and @xmath36 using the lanczos algorithm is fairly standard@xcite . from the ground state @xmath54
we calculate the local magnetization @xmath79 @xmath80 our results pertaining to the ferromagnetic phase f are as follows . in fig .
2 , we display @xmath79 for @xmath81 and @xmath82 and @xmath41 for various values of @xmath7 . with the exception of the smallest @xmath7-values used , we first observe that the data collapse onto a single curve , which implies that the @xmath7 considered are already sufficiently large as compared to the correlation length that they represent the @xmath83 limit .
second , we note that the profile is quite wide and a continuous function of @xmath84 .
we find a similar data collapse for all the situations we are going to consider in the sequel .
next , we show in fig .
3 , that the overall shape of the profile is not a peculiarity of having spin @xmath41 .
rather , we see that the profiles obtained with the parameters @xmath82 and @xmath81 ( fig .
3a ) and @xmath85 ( fig .
3b ) both spins @xmath25 and @xmath86 are very close to the one for @xmath41 .
nevertheless , the finer details of the respective shapes become different as @xmath5 increases . finally , in fig .
4 , we show the effect of varying @xmath48 in the @xmath41 case .
note that the profiles for a transverse field as small as @xmath87 and as large as @xmath81 are superimposed onto each other .
the fact that the width @xmath88 of the domain wall is proportional to the size of the system is a peculiarity of a one dimensional quantum model . in higher dimensions we would expect @xmath88 to remain finite as @xmath83 except possibly at a critical point where the correlation length diverges .
we shall turn to a quantitative explanation of these findings below .
profiles for the oscillatory phase are discussed in subsection c. in order to gain some understanding of the results of the last section , we now present some simple approximate calculations . as it turns out ,
most of the physics of the problem for @xmath48 small is conveniently described in terms of degenerate perturbation theory involving the manifold @xmath89 of the @xmath90 ground states .
this procedure avoids the additional technical complications of the free fermion method for `` up - down '' boundary conditions .
the boundary conditions are that spin @xmath31 is fixed to have @xmath91 and spin @xmath32 has @xmath92 . for @xmath19 the manifold @xmath93 contains the states @xmath94 , for @xmath95 , where @xmath96 denotes the quantum state ( shown in fig .
5 ) in which spins @xmath97 with @xmath98 have their @xmath4-component of spin equal to @xmath99 and those with @xmath100 have their @xmath4-component of spin equal to @xmath101 .
the only nonzero matrix elements of the hamiltonian within the ground manifold are @xmath102 considering only the manifold @xmath93 , one finds the corresponding eigenvectors @xmath103 , and eigenenergies @xmath59 ( for @xmath104 ) to be @xmath105 |n \rangle \ , \ \ \ \ e_p = e_0 + h \cos [ p \pi / ( l+2 ) ] \ , \ ] ] where @xmath106 is a normalization constant .
perturbative corrections to the energy will occur at order @xmath107 and to the wavefunction at order @xmath21 .
we may calculate the profile function within the small @xmath48 approximation : @xmath108 - \sum_{k=1}^n \sin^2 [ kp \pi / ( l+2 ) ] \over \sum_{k=1}^{l+1 } \sin^2 [ kp \pi / ( l+2 ) ] } \nonumber \\ & = & 1 - 2 { \sum_{k=1}^n \sin^2 [ kp \pi / ( l+2 ) ] \over \sum_{k=1}^{l+1 } \sin^2 [ kp \pi / ( l+2 ) ] } \nonumber \\ & = & 1 - 2 { ( 2n+1 ) \sin { p \pi \over l+2 } - \sin { ( 2n+1)p \pi \over l+2 } \over ( 2l+3 ) \sin { p \pi \over l+2 } - \sin { ( 2l+3)p \pi \over l+2 } } \ .\end{aligned}\ ] ] we set @xmath109 and work in the limit of infinite @xmath7 .
thereby we find @xmath110 in fig .
3 , @xmath111 is shown together with the numerically determined ground - state profiles @xmath79 for @xmath41 , @xmath25 and @xmath86 and we find a nice qualitative agreement ( even for moderately large values of @xmath48 ) .
the above calculation can be generalized to larger @xmath0 .
consider a state in which the wall is between lattice sites .
in this state let all spins to the left of the wall have @xmath91 and those to the right have @xmath92 . note that it is possible to change the value of @xmath6 for _ either one _ ( but not simultaneously both ) of the spins adjacent to the wall without changing the unperturbed ( @xmath90 ) energy .
for spin @xmath25 since the nonzero matrix elements of the perturbation due to the transverse field are all the same , the problem is totally equivalent to a spin 1/2 chain of twice the length .
this observation explains the fact that our results for @xmath19 and @xmath25 are indistinguishable .
when @xmath112 , one has to account for the fact that to move the wall through one lattice constant involves matrix elements which depend on the initial and final values of @xmath6 .
this case will be considered later .
the discussion has been so far restricted to the ferromagnetic phase f. the oscillatory phase is distinguished from it by showing a non - monotonic decrease of the correlation function [ see eq . ( [ corrs ] ) ] .
how does this behavior manifest itself on the level of the magnetization profile @xmath79 ? to answer this question , we display in fig .
6a , for @xmath41 , @xmath79 for @xmath113 and @xmath114 and for @xmath7 _ even_. indeed , this profile is distinct form the ones observed previously in the phase f. first , finite - size effects are much larger than in the situations discussed before .
second , @xmath79 displays , at least for @xmath7 finite , a step - like behavior and it looks as if the system was built out of hard objects each occupying two lattice sites . when @xmath48 is increased , these composites soften until they completely melt at the transition towards the f phase .
this picture is modified in interesting ways for @xmath7 _ odd_. in fact , the calculation of @xmath79 requires a little more care in this case . for @xmath7 even , the system has a single well - defined ground state separated by a gap at least of order @xmath115 ( see sec .
v ) from the excited states .
that is not so for @xmath7 odd .
rather , for @xmath90 but @xmath116 , one finds that the ground state is twofold degenerate .
that is a new degeneracy which has nothing to do with the ferromagnetic ordering of the system .
even when @xmath117 , the two lowest states in @xmath89 remain much closer to each other than with the other states , the latter one having gaps of order @xmath118 .
this new ( near ) degeneracy implies that one must reconsider the calculation of the magnetization profile @xmath79 . rather than blindly using eq .
( [ magpro ] ) , we take the _ two _ ground states @xmath54 and @xmath119 to be nearly degenerate .
then we construct the matrix @xmath120 and find its two eigenvalues @xmath121 .
each of those represents a magnetization profile and one of them is shown in fig .
the other profile is obtained by reflection around @xmath122 . the asymmetry in fig .
6b , observed for @xmath7 odd and finite , is easily understood in terms of the composite objects introduced above .
note that within the manifold @xmath93 , the @xmath123 term in the hamiltonian moves the wall through two lattice spaces .
since we are discussing @xmath90 , this is the only kinetic energy of the domain wall .
thus , if we write @xmath124 , where @xmath125 ( @xmath126 ) is the submanifold of state in which there are an even ( odd ) number of up spins , then there are no matrix elements between these two submanifolds .
furthermore , these two submanifolds are related to one another by spatial inversion ( accompanied by @xmath127 ) .
thus , for @xmath7 odd , the problem decomposes into two identical eigenvalue problems ( hence the twofold degeneracy of the energies ) .
the matrix to be diagonalized is exactly the one considered in sec .
iii.b , but now for @xmath128 sites and with @xmath129 .
the two sets of eigenstates are thus @xmath130 ) diagonal .
writing @xmath131 and @xmath132 , respectively , we find for the profile for the @xmath133 set of eigenstates ( indicated by a subscript `` @xmath133 '' ) @xmath134 in particular , this reproduces the steps in @xmath135 observed numerically .
now , taking the continuum limit , it is easy to see that the width of the terraces decreases as @xmath77 and that @xmath136 .
we therefore suggest that the phenomenon observed in fig .
7 is a novel type of finite - size effect . as @xmath77 , the step - like and asymmetric profiles found for @xmath7 finite will converge towards a smooth and symmetric limit function . at least for @xmath137 , this limit function appears to show a simple relationship with the profiles previously found in the f phase .
indeed , our data suggest the following .
the profiles for @xmath7 large as found for @xmath138 are very close to those obtained for @xmath139 . for small @xmath35
our analytic work shows this to be true with @xmath140 . for larger @xmath35 we have the phenomenological result ,
@xmath141 evidence for this is provided in fig .
7 , where some profiles found for @xmath90 and @xmath116 ( data points ) are collapsed with profiles calculated with @xmath142 .
in this section we obtain several analytic results in the large @xmath0 limit .
we first analyze the classical ( @xmath15 ) system .
then we treat the quantum system with @xmath0 large , but not infinite . to treat the classical ( @xmath27 ) system , we write @xmath143 where @xmath144 is a continuous classical variable
. then the energy of the classical system is @xmath145 for the classical calculation without a wall , we will adopt periodic boundary conditions , so that @xmath146 and the last term in eq . ( [ eclass ] ) runs from @xmath147 to @xmath148 . in that case @xmath149 , where , for convenience , we set @xmath150 where @xmath151 is the critical value of @xmath48 above which long - range ising order disappears . for the periodic chain the energy per spin , @xmath152 , is given by @xmath153 in the classical calculations we will use two types of boundary conditions to generate a wall .
the first type of boundary condition is used for the continuum calculation . here
the number of sites is infinite and hence we set @xmath154 .
this continuum calculation is valid when the angle @xmath144 changes slowly with position , as it does for @xmath155 .
the second type of boundary condition is that used in the discrete calculation .
here it is convenient to use antiperiodic boundary conditions in which spins at opposite ends of the chain are coupled antiferromagnetically , with @xmath64 and we require @xmath156 . in any case , the ground state energy is found by minimizing @xmath58 with respect to the @xmath144 s .
apart from end effects this minimization yields the following conditions ( for @xmath157 ) which characterize the exact ground state , @xmath158 for small @xmath159 the angle @xmath144 will change abruptly at the domain wall and we will treat this case in the next subsection . as @xmath159 increases ( up to the critical value @xmath160 ) , @xmath161 will become a smoother function of @xmath18 . in that case
, a continuum approximation for @xmath161 makes sense . to treat the continuum limit for @xmath155 , we set @xmath162 , where @xmath163 , whence eq .
( [ exact ] ) becomes @xmath164 \ .\ ] ] the continuum limit of this equation is @xmath165 \ .\ ] ] the solution to this differential equation is reduced to quadratures : @xmath166 we now discuss the significance of this result .
first we analyze the behavior for large @xmath2 . at large @xmath167 , @xmath168 approaches @xmath169 , where @xmath170 .
thus for large @xmath167 we set @xmath171 $ ] . keeping only the dominant terms in eq .
( [ x ] ) , we obtain @xmath172 this gives @xmath173 where @xmath174 and @xmath47 is the correlation length , given by @xmath175 thus , for @xmath176 , @xmath177 with @xmath178 , @xmath179 , and @xmath180 .
next let us see the behavior of the solution near @xmath181 .
to do that , differentiate the solution above with respect to @xmath2 at @xmath181 : @xmath182 where we used @xmath183 .
this gives @xmath184_{x=0 } = { \sqrt 2 \over a } ( 1-\tilde h ) \ .\ ] ] to estimate the width , @xmath185 , of the wall , we note that over a distance @xmath185 , the angle @xmath144 varies from approximately @xmath186 to approximately @xmath187 .
this reasoning indicates that @xmath188_{x=0 } = 2 \phi_{\rm eq } \ , \ ] ] for @xmath189 , we have @xmath190 , so that @xmath191 thus for @xmath192 , @xmath185 is of order @xmath47 , as one might expect . in this and the next subsection we look at the discrete equations , which are the appropriate ones for small @xmath159 .
since we are dealing with a classical system , it is convenient to use antiperiodic boundary conditions in which spins at opposite ends of the chain are coupled antiferromagnetically , with @xmath64 .
first we consider the case when the center of the wall is at a lattice site . to treat this case
, we consider a chain with an odd number of spins .
number the sites @xmath193 and fix @xmath194 .
then @xmath195 and the energy @xmath196 for this chain of @xmath197 sites is @xmath198 the @xmath161 are determined by eq .
( [ exact ] ) for @xmath199 , by @xmath200 and by @xmath201 to obtain the solutions for the @xmath144 s at small @xmath48 we write @xmath202 for @xmath203 .
then for @xmath204 @xmath205 \ , \ \ \ \tilde h \cot \epsilon_n = { 1 \over 2 } [ \cos \epsilon_{n-1 } + \cos \epsilon_n ] \ .\ ] ] to obtain the energy at order @xmath159 we only need to solve the equations correct to order @xmath159 , in which case @xmath206 then , to find the energy to order @xmath159 , we may set @xmath207 , and @xmath208 , so that @xmath209 the wall energy @xmath210 is given by the large @xmath211 limit of @xmath212 where @xmath152 is the energy per site of the uniform chain .
thus @xmath213 to treat the case when the center of the wall is midway between two lattice sites , we consider a chain of @xmath214 sites , numbered @xmath215 .
we require that @xmath195 and we use the same antiperiodic boundary conditions as in the previous case
. then the energy is @xmath216 as before , the wall energy , @xmath217 is obtained via @xmath218 the @xmath161 are determined by eq .
( [ exact ] ) for @xmath199 , by eq .
( [ phin ] ) , and by @xmath219 to analyze these equations , set @xmath220 .
then , to first order in @xmath221 we have ( for @xmath222 ) @xmath223\ ] ] or @xmath224 we expect an exponentially decaying solution for @xmath221 for @xmath225 .
so set @xmath226 so that @xmath227 from which we get @xmath228 for small @xmath159 , we get @xmath229 . for @xmath176 , we get @xmath230 , in agreement with the continuum result .
these results hold irrespective of the position of the center of the wall .
we now analyze the solution near the wall for small @xmath159 . to do that we first look at eq .
( [ phiref ] ) for @xmath231 when we set @xmath232 : @xmath233 \ .\ ] ] but @xmath234 , so @xmath235 = { 1 \over 4 } \epsilon_1 ^ 2 \ .\ ] ] so @xmath236 from here on one should interpret @xmath237 to mean @xmath238 .
next , look at the equation for @xmath239 : @xmath240 \approx 1 \ .\ ] ] in this way we find that @xmath241 , for @xmath242 .
a more systematic approach shows that we can write the solution for the @xmath35 s as @xmath243 \ , \ \ \ \ \ \epsilon_n = \epsilon_{\rm eq } + a_n(4\tilde h)^{2n-(7/3)}f_n[(4\tilde h)^{2/3 } ] \ , \ \ \ \ \ n > 1 \ , \ ] ] where @xmath244 , @xmath245 , and the functions @xmath246 are analytic functions such that @xmath247 . to calculate @xmath248 , note that , up to order @xmath249 in the energy , we may write @xmath250 ^ 2 + { 1 \over 24 } ( 4\tilde h)^{4/3 } [ f_1(\tilde h^{2/3})]^4 \dots \nonumber \\
\sin \phi_j & = & \cos \epsilon_j = 1 \ , \ \ \ \ \ \ j>1 \ ; \nonumber \\
\phi_1 & = & \sin \epsilon_1 = ( 4\tilde h)^{1/3 } f_1 ( \tilde h^{2/3 } ) + { \rm o}(\tilde h ) \nonumber \\
\phi_j & = & \sin \epsilon_j = { \rm o}(\tilde h ) \ , \ \ \ \ \ \
j>1 \ .\end{aligned}\ ] ] putting these evaluations into eq .
( [ eprpr ] ) , we obtain @xmath251 ^ 2 + { 1 \over 24 } ( 4\tilde h)^{4/3 } [ f_1(\tilde h^{2/3})]^4 + ( n-2 ) \nonumber \\ & & \ \ - \frac{1}{2}\biggl [ 1 - ( 4\tilde h)^{2/3 } [ f_1(\tilde h^{2/3})]^2 + { 1 \over 3 } ( 4 \tilde h)^{4/3 } f_1(\tilde h^{2/3})]^4 \biggr ] + \frac{1}{2 } + 2\tilde h [ ( 4\tilde h)^{1/3 } f_1(\tilde h^{2/3 } ) ] \nonumber \\ & = & n-1 + ( 3/8 ) ( 4\tilde h)^{4/3 } + { \rm o}(\tilde h^2 ) \ .\end{aligned}\ ] ] where we used the definition that @xmath252 .
then the wall energy is @xmath253 note that this result is _ not _ identical to the case when the center of the wall is at a lattice site .
the physics of this result will be discussed in more detail in the next section .
we now consider the quantum chain for large @xmath0 . for a chain of @xmath7 spins of magnitude @xmath0
, there are @xmath254 states which have the same @xmath90 ground state energy and which therefore must be treated within degenerate perturbation theory .
numerical studies show that for @xmath255 , this manifold of states splits into bands which are slightly different depending on whether @xmath256 is even or odd . for @xmath0 half integer
there are @xmath256 bands , each having @xmath7 states except for the middle band which has @xmath257 states . approximately , the centers of these bands are at energies @xmath258 where @xmath259 assumes the values @xmath260 , @xmath261 , @xmath262 , @xmath263 @xmath264 . for integer @xmath0
one has @xmath265 bands , each having @xmath7 states except for the middle band which has @xmath266 states . in this case
the centers of the bands are given by eq .
( [ center ] ) , but @xmath259 assumes the values @xmath260 , @xmath267 , @xmath268 , @xmath263 , @xmath264 . as an example , in fig .
8 we show the density of states for @xmath269 within the manifold @xmath93 . as fig .
8 illustrates , the bands become wider and the gaps between bands become smaller as one approaches zero energy ( about which the levels occur symmetrically ) .
a qualitative explanation of the occurrence of such band gaps is as follows .
if the spin 1/2 case is likened to a hopping model , then the spin @xmath0 case is analogous to a hopping model in which there is a periodically variable hopping matrix element .
this problem is therefore analogous to that of an electron in a weak periodic potential , where one knows@xcite that even for arbitrarily weak potentials one has band gaps .
we will now demonstrate the existence of the band gaps and obtain quantitative information on the band widths for @xmath270 as follows .
we will give a construction for the outermost ( and narrowest ) sub - band , for which @xmath271 in eq .
( [ center ] ) .
consider the submanifold of states , @xmath272 , which contains the states , shown in fig .
5 , @xmath273 ( for @xmath274 .
these states are defined so that all spins ( if any ) having @xmath275 have @xmath276 , all spins ( if any ) having @xmath277 have @xmath278 , and @xmath279
. one may verify that the state @xmath96 is in the subspace @xmath93 of states of minimum energy eigenstates , @xmath280 , of the @xmath90 hamiltonian , i.e. , that @xmath281 .
the states @xmath96 are not orthogonal to one another , but do have an overlap which is small for large @xmath0 and is given by @xmath282 for @xmath283 .
rose@xcite shows that @xmath284 where @xmath285 is a rotation matrix .
therefore , we use states which are orthonormalized to leading order in @xmath286 , @xmath287 where we interpret @xmath288 and @xmath289 to be zero .
we now work only to first order in @xmath286 .
then one notes that unless @xmath290 , @xmath291 has zero overlap with the subspace @xmath93 . to see this , note that @xmath292 .
therefore we write , correct to leading order in @xmath286 , @xmath293 where @xmath294 .
also @xmath295 thus we expect this subspace of states with @xmath296 to form a sub - band centered at energy @xmath297 with a width determined by the hopping matrix element@xcite @xmath298 .
thus the width of this band should be @xmath299 .
numerically , for @xmath300 we found a band width of @xmath301 in excellent agreement with our calculation .
we thus have an analytic description of the outermost two sub - bands corresponding to @xmath302 .
we did not consider an analysis of the remaining inner sub - bands .
note that the results shown in fig .
8 are not for asymptotically large @xmath0 .
we did check that when @xmath0 is large enough , the outermost subband does become symmetric , as one would expect for a one - dimensional density of states corresponding to eqs .
( [ hoff ] ) and ( [ hdiag ] ) . in a sense
, one sees from this calculation that for the wall to move one lattice spacing , it must tunnel through the phase space of spin states indicated by the result of eq .
( [ tunnel ] ) . in this connection
a nice analogy has been suggested by stinchcombe.@xcite consider a rotor in a strong @xmath303 potential .
on knows from exact solutions of the associated eigenvalue problem that one has a limit in which one has harmonic oscillator levels in the bottom of the potential well , all of which are doublets with a small tunnel splitting.@xcite it is tempting to think that one can provide a field theory in one space and time dimension in which the single potential minimum of the rotor becomes a periodic potential in a lattice .
then the bands would reflect the tunnel splitting .
so far , we have considered the form of the profiles @xmath79 and have seen that at least qualitatively , their form can be understood from simple degenerate perturbation theory .
the manifold @xmath89 considered consists of those states which give the lowest energies in the @xmath137 limit . for the case of @xmath41 , for @xmath90 these are the states shown in fig .
5 , namely , @xmath305 and for @xmath7 sites , there are @xmath257 of them . for @xmath25
one starts at @xmath90 from the states @xmath306 and one gets a subspace of @xmath307 states .
all these states have in common that they contain a single wall .
states which do not belong to this `` one - particle subspace '' @xmath89 have a large gap with the ground state .
it is now remarkable that this clear separation of the subspace @xmath89 from all other states remains intact even for _ finite _ values of @xmath48 .
the subspace @xmath89 has a simple analogy for ( anti)periodic boundary conditions . in these cases
, @xmath65 commutes with the ( modified ) translation operator @xmath67 ( or @xmath308 ) and is thus decomposed into block diagonal form , the blocks being labeled by the eigenvalues of @xmath67 ( or @xmath308 ) .
@xmath89 corresponds to the set of lowest eigenstates of the blocks of @xmath65 . while the gap of the lowest excited state _ outside _
@xmath89 with the ground state is finite ( e.g. @xmath309 for @xmath41 and when @xmath82 ) , the gaps inside @xmath89 scale with @xmath304 . if @xmath310 denotes the @xmath311 gap , we expect the scaling @xmath312 with @xmath313 , as obtained in eq .
( [ pertth ] ) .
finite - size estimates for @xmath314 are then obtained from , with @xmath315 @xmath316 the extrapolation towards the limit @xmath317 was carried out with the bst extrapolation algorithm@xcite .
we illustrate the convergence of the finite - size data with a few examples for @xmath25 in table [ tab1 ] .
the convergence towards @xmath313 is even clearer for @xmath41 .
we now consider the finite - size amplitudes @xmath318 the motivation for this comes from systems being precisely at a critical point . in that case
, it is known that for one - dimensional quantum chains finite - size amplitudes defined as @xmath319 are related@xcite to the anomalous dimensions of the scaling operators of the model and this leads to simple patterns of the spectrum of the amplitudes ( see e.g. ref . for more information ) .
although we are not working here at a critical point , we observe a simple structure of the excited states within @xmath89 for @xmath77 out of the critical region .
namely @xmath320 evidence for this is presented in table [ tab2 ] for @xmath41 and table [ tab3 ] for @xmath25 which give the bst extrapolated@xcite estimates for the @xmath321 .
it is remarkable that the _ same _ expression should hold for the gaps ratios of both the @xmath41 and the @xmath25 models , even though @xmath48 is not very small and/or @xmath116 .
for @xmath48 small and @xmath82 , this result can be recovered from the energies ( [ pertth ] ) in lowest order of perturbation theory . turning to the oscillatory phase
, we note that the spectrum of the states within the subspace @xmath89 is different from that in the f phase .
here there are important differences between @xmath7 even and @xmath7 odd . for simplicity , we restrict attention to spin @xmath41 and first concentrate on the line @xmath90 .
the structure of the levels we are going to find is shown in fig .
9 . for @xmath7
even , the spectrum of the low - lying states corresponding to @xmath89 is grouped into doublets .
while the doublet splitting is of order @xmath115 , the gaps between doublets are of order @xmath118 .
for @xmath7 odd , on the other hand , the energy levels are doubly degenerate , with gaps between them of order @xmath118 . to see this , we look for a scaling behavior of the energy gaps @xmath323 and define finite - size estimates for the exponent @xmath314 using ( [ thetafs ] ) . in table
[ tab5 ] , we give our results for @xmath324 . for @xmath7 even , estimates of @xmath314 from the lowest six gaps ( @xmath325 ) are given .
( for some of the higher gaps , the finite - size data for @xmath7 small are not yet in the asymptotic regime and are thus discarded . ) we observe that the lowest gap scales with an exponent @xmath326 , while all the higher gaps scale with @xmath313 . for @xmath7 odd ,
on the other hand , each energy level is doubly degenerate even for finite @xmath7 , i.e. @xmath327 , @xmath328 .
estimates of @xmath314 for @xmath329 are given in table [ tab5 ] and we find @xmath313 throughout .
similar results are also found for @xmath330 and @xmath331 .
next , we study the finite - size amplitudes @xmath332 , with @xmath315 .
estimates for the lowest six amplitudes extrapolated to @xmath77 are shown in table [ tab6 ] , obtained for @xmath7 even or odd , respectively .
we find that for @xmath7 even , the estimates for pairs of amplitudes are quite close to each other and are consistent with being equal . furthermore , their numerical values are near to the ones found for @xmath7 odd .
taken together , the present data suggest the simple picture that in the @xmath77 limit , the amplitudes @xmath321 should become doubly degenerate and independent of the evenness or oddness of @xmath7 . finally , when looking for the amplitude ratios @xmath333 , we find the following pattern @xmath334 which is consistent with the very same structure ( [ amprat ] ) of the low - lying amplitudes found on the @xmath82 line
this finding is certainly consistent with the relationship between the order parameter profiles observed earlier ( see fig .
7 ) between systems on the @xmath90 and @xmath82 lines .
these findings can be reproduced from the perturbative arguments of section iii.b when @xmath35 is small .
for @xmath7 odd , the eigenvalue problem then reduces to one treated there , but with @xmath335 sites and @xmath336 and each state being twofold degenerate .
for @xmath7 even , we get two distinct problems , one with @xmath337 sites and one with @xmath338 sites . using the expression ( [ pertth ] ) for the energies , we find pairs of levels with a splitting of order @xmath115 between them .
qualitatively , the same structure persists throughout the o phase .
we illustrate this in fig . 10 , where we show for @xmath339 and @xmath340 sites the dependence of the first few eigenvalues on @xmath48 .
because @xmath7 is not very large , the statements we shall make about the spectrum apply most strongly to the lowest levels and less well to the higher levels .
we see that inside the o phase , the lowest energies occur in pairs which oscillate around another . in the ground state
, there are for @xmath7 sites @xmath335 level crossings for @xmath7 odd and @xmath337 level crossings for @xmath7 even , respectively .
these level crossing do not continue into the f phase .
finally , we point out that at the f / d transition , the energies combine into a @xmath341 representation of the @xmath342 virasoro algebra , as predicted from conformal invariance@xcite . in agreement with that prediction
, we observe in fig .
10 close to the critical point @xmath343 an ( approximately ) equidistant level spacing with the degeneracies @xmath344 .
( actually , because of finite size effects , this structure is realized at @xmath345 rather than at @xmath346 . )
this kind of level crossings in the ground state of finite - lattice quantum system has been first investigated for _ periodic _ boundary conditions@xcite .
there is was noted that the level crossings always occurred between the @xmath39 even and odd sectors of the model .
here we find that the level crossing persist even if the @xmath39 symmetry is broken by our chosen `` up - down '' boundary conditions ( [ bc ] ) .
applying finite - size scaling@xcite to the location @xmath347 of the level crossings , we expect ( but did not test ) a scaling @xmath348 . for periodic boundary conditions
it is known@xcite that @xmath178 .
( @xmath349 gives the o / f transition line ) .
it is obvious that the fact that the width of the wall for the quantum model is of order the system size is a result of a quantum superposition of states each of which have a narrow wall .
it is tempting , therefore , to introduce a measure of the intrinsic " width of the wall , which is the minimum width obtainable within the subspace of states under consideration .
we emphasize that this concept depends on the subspace of states begin considered .
as we vary @xmath35 , the strength of the @xmath3-@xmath3 coupling , it seems plausible that the character of the low lying particle " states may vary .
thus , it would be of interest to introduce a measure of the intrinsic width applicable to this case . for this purpose
consider the quantity @xmath350 defined as @xmath351 where the maximum is to be taken over all possible quantum states @xmath96 in the subspace @xmath93 .
( as long as the site @xmath97 is not near an end , @xmath350 will depend only weakly on @xmath97 . )
if the states consist of sharp walls , then @xmath350 will be unity .
so we use this quantity to define the intrinsic width , @xmath352 , via @xmath353 .
this type of definition is somewhat similar to the inverse participation ratio introduced to characterize localized and extended states of an electron in a random potential.@xcite here we take the subspace @xmath93 to be the particle " subspace of states .
this definition is appropriate as long as the subspace of particle states can be unambiguously separated from the continuum .
this requirement suggests that @xmath354 .
but , we can also use this idea when @xmath35 is nonzero .
we should mention that the maximization required in eq .
( [ max ] ) is easy to carry out .
one simply forms the matrix , @xmath355 , ( in any representation ) where @xmath356 and @xmath350 is the largest eigenvalue of @xmath355 .
when @xmath350 is less than unity , one also obtains information on the shape of the wall via the eigenvalue spectrum of @xmath355 . for instance , if the wall has width 3 , and if the three largest eigenvalues of @xmath355 are degenerate , then one would conclude that the intrinsic wall profile is linear .
one context in which this concept provides some information is when we consider the large @xmath0 limit of the transverse ising model . in that case
we have already seen that the low lying particle spectrum splits into bands corresponding roughly to values of @xmath357 . when we apply eq .
( [ max ] ) taking the subspace @xmath93 to be the lowest band of @xmath7 states with @xmath296 , we expect to find @xmath358 .
we in fact verified this expectation in our numerical calculations .
they showed that one obtains two nearly degenerate maximal eigenvalues of magnitude essentially equal to @xmath359 .
there are two eigenvalues because there are two wavefunctions which can maximize @xmath350 since the minimum intrinsic wall width is two lattice spacings .
we see then , that the prescription of eq .
( [ max ] ) coincides with the perturbative calculations for small @xmath48 .
first we summarize the result for the classical ( @xmath15 ) system .
note that @xmath360 , for small @xmath5 .
so to move the wall through one or more lattice constants requires an energy @xmath361 \ .\ ] ] we have numerically studied the energies of the classical domain wall when its center is either at a lattice site or midway between lattice sites .
we find that the barrier energy @xmath362 does not change sign as @xmath5 is increased as might be suggested by eq .
( [ barrier ] ) , but rather @xmath362 decreases monotonically to zero as @xmath5 is increased towards its critical value of unity .
we carried out similar calculations for quantum systems with @xmath41 and @xmath25 .
we found that the difference in energy when the center of the wall moved from a lattice site to a point midway between lattice sites was too small to be accessible to double precision arithmetic .
since , for a chain of length @xmath363 , we can not imagine a parameter on such a scale , we assert that the quantum barrier energy is zero .
practically , we performed this test as follows .
for sufficiently large lattice sizes @xmath7 , we expect for the ground state energies @xmath364 ( i.e. wall between two sites ) and @xmath365 ( i.e. wall at a site ) @xmath366 where @xmath367 is the ground state energy per spin and independent of the boundary conditions . for @xmath41 , @xmath368 where @xmath58 is a complete elliptic integral@xcite
. the desired wall energy difference should be proportional to @xmath369 .
this in turn can be estimated from the quantities @xmath370 and we note that @xmath371 and @xmath372 should have opposite signs .
we give in table [ tab4 ] estimates for both @xmath371 and @xmath372 for both @xmath41 and @xmath25 for @xmath81 .
we chose this value of @xmath48 to be small enough that is was for from criticality but large enough so that @xmath373 should be easily observable if nonzero . taking the entries at face value
, the wall energies must be less than @xmath374 .
moreover , since @xmath375 and @xmath376 should have opposite signs , the values given in table [ tab4 ] appear to be entirely due to the leading finite - size corrections and we find that @xmath377 decrease rapidly with increasing @xmath211 .
so we get an upper bound @xmath378 for both @xmath41 and @xmath25 .
this result is not surprising . in the limit when the width of the domain wall diverges ( as it does for @xmath83 )
, the barrier energy should vanish . whether it vanishes algebraically or
exponentially would require further numerical or analytic analysis .
here we discuss the various regimes which exist at at large @xmath0 .
it is clear that bands of energy levels exist on various scales .
it is useful to discuss the consequences of these energy scales in terms of scaling functions . for notational simplicity , in this subsection @xmath379 will denote a scaling function of @xmath2 which
, in general , will be different in different appearances .
we first discuss the thermodynamic properties of the wall . for this purpose
we will consider the entropy associated with the wall , @xmath380 , defined as the entropy with `` up - down '' boundary conditions minus that with `` up - up '' boundary conditions .
we emphasize that the scaling of the bulk thermodynamics , i.e. , the extensive part of the thermodynamic potentials is independent of the boundary conditions and only depends on the bulk thermodynamic variables.@xcite at temperatures small compared to @xmath8 the entropy with `` up - up '' boundary conditions will be zero , because the ground state is nondegenerate .
the entropy with `` up - down '' boundary conditions will be that due to the band of states with just one domain wall .
first there is an extreme low temperature limit in which the entropy is vanishingly small : @xmath381 where we always set @xmath382 , the boltzmann constant equal to unity . in this limit only the lowest single state of the tunneling band found in sec .
ivb comes into play . at higher temperatures
the tunneling band is activated , so that @xmath383 here @xmath384 is the entropy of a one dimensional tight binding band of @xmath7 sites . by noting the form of the density of states one can see that for small @xmath2 , @xmath385 and as @xmath386 , @xmath387 .
the upper limit of this regime is defined so that essentially only the lowest subband of sec .
ivb is activated .
next , there is a regime when the temperature is high enough that all states of these subbands are equally populated , but only one domain wall exists . in this discussion
it is assumed that @xmath388 , so that the band of particle states is well separated from the states with more than one wall . in this regime
one has @xmath389 in this case , @xmath387 for @xmath386 and @xmath390 for @xmath391 .
finally , we have the classical regime when @xmath392 . in that regime
any effects due to the fact that @xmath393 and @xmath394 do not commute with one another becomes unimportant . in that regime then , the partition function for the quantum model of eq .
( [ hamzz ] ) is the same [ up to corrections of relative order @xmath395 as that of the classical model in which the operators are interpreted as classical variables according to eq .
( [ class ] ) .
the effect of a random potential is somewhat different . in the present context a random potential
would be created by a field which independently for each site assumes random values from a distribution of width @xmath396 .
it is known@xcite that in the presence of a random potential all states are localized .
that means that in the presence of a random field the eigenfunctions are localized and therefore the width of the wall in the ground state remains finite in the limit @xmath83 .
nevertheless , the width of the wall , @xmath185 , can be expressed in terms of scaling functions of the same variables as describe the thermodynamics except that here @xmath396 replaces @xmath14 .
thus @xmath397 and @xmath398 since we know that @xmath185 can not be less than the intrinsic width , we see that @xmath399 .
similarly , we have @xmath400 in this case @xmath401 and @xmath402 , the intrinsic wall width in this regime . for any larger @xmath396
, @xmath403 remains equal to unity .
we may summarize our conclusions for the ising model in a transverse field as follows . 1 .
although an exact analysis of domain walls in the transverse ising model can in principle be accomplished using the jordan - wigner transformation , it is much simpler to consider the nature of degenerate perturbation theory in the presence of `` up - down '' boundary conditions . within degenerate perturbation theory for @xmath19 and
@xmath25 the hamiltonian governing the position of the center of the domain wall is isomorphic with a hopping model with a hopping matrix element @xmath404 .
we verified this in detail with analysis of the scaling of the energy levels with chain length , @xmath7 .
2 . for spins @xmath405 , @xmath60 and @xmath406 the ground state profiles
when spins at opposite ends of a chain of length @xmath7 are fixed to be antiparallel show relatively little dependence on the transverse field @xmath48 ( as long as @xmath48 is less than the critical value , @xmath407 , above which ordering in @xmath6 is destroyed ) . for small @xmath48
the profiles at large but finite @xmath0 are not very different from those at small @xmath0 . in all cases ,
the profiles are quite well obtained from the hopping model .
3 . the profiles in the oscillatory phase show much larger finite size corrections than those in the ferromagnetic phase . in the limit @xmath83 the profiles in the oscillatory phase may still show some oscillatory behavior , unlike in the ferromagnetic phase . also , our results for the profiles show an interesting finite - size effect in which spatial parity appears to be broken .
4 . in the spectrum of the oscillatory phase at @xmath90 one finds an unusual pattern for the gaps between the ground state and the first few excited states . while for @xmath7 even , the energies form doublets , each with a splitting which scales with @xmath7 as @xmath408 , and with gaps between the doublets which scale as @xmath304 . for @xmath7 odd , the energies are doubly degenerate and the gaps scale with @xmath7 as @xmath304 .
this behavior was explained in terms of dimer excitations which occur in the limit @xmath409 when @xmath410 .
5 . for large @xmath0 the quantum transverse ising model the hamiltonian describing the position of the center of the domain wall gives
rise to band splittings analogous to those found for an electron in a periodic potential@xcite .
the lowest sub - band is described by a hopping model in which the center of the domain wall tunnels from one site to a neighboring sites .
the associated tunneling matrix element is of order @xmath411 , where @xmath412
. it would be interesting to recover this result within a field theory for one space and one time dimension .
6 . for the classical ( @xmath27 ) model
the domain wall is extremely narrow for small @xmath48 .
we have determined the domain wall energy for the case when i ) the center of the domain wall is at a lattice site and ii ) the center of the domain wall is midway between two lattice sites .
the difference between these energies is the pinning energy which must be exceeded to move the domain wall through the lattice . for the quantum model
we find this pinning energy vanishes as @xmath83 due to the fact that the width of the domain wall diverges in this limit . 7 . in attempting to relate quantum problems to classical problems
it is necessary to define an intrinsic width , @xmath352 , of a quantum domain wall .
we have defined @xmath352 as the minimum possible width attainable using wave functions from a manifold @xmath93 . at temperatures
large compared to the band but small compared to the activation energy for creation of an additional wall , the manifold @xmath93 is the hopping band . in that case , the intrinsic width of the quantum system is essentially equal to the width of the analogous classical domain wall .
* acknowledgements * the authors appreciate helpful advice from d.b .
abraham , r. b. stinchcombe , t. c. lubensky , i. affleck and j.l .
abh thanks the department of theoretical physics of oxford university for its kind hospitality .
abh is supported by an epsrc visiting fellowship and mh is supported by a grant from the ec programme human capital and mobility " .
mc acknowledges support by the polish agency kbn ( grant number 2 p302 127 07 ) .
f. family and t. vicsek , editors , `` dynamics of fractal surfaces '' , world scientific , singapore ( 1991 ) ; g. forgacs , r. lipowsky and th .
m. nieuwenhuizen , in `` phase transitions and critical phenomena '' , edited by c. domb and j. l. lebowitz , vol .
14 , academic press , ny ( 1991 ) .
this argument is correct to leading order in @xmath286 .
it neglects overlap between @xmath413 and other states of @xmath93 , notably states similar to @xmath96 , but in which @xmath414 has the value @xmath415 .
however the band of the @xmath416 states is separated from the band of @xmath296 states by an energy approximately equal to @xmath1 .
but since the relevant matrix element is of order @xmath417 and is off - diagonal , it will enter at order @xmath418 \sim s\tau^2 $ ] .
thus this effect is negligible compared to the terms we do treat , at least for large @xmath0 . m. n. barber in _ phase transitions and critical phenomena _ , vol .
c. domb and j. lebowitz , ( academic press new york 1983 ) , p. 145 ; v. privman ( ed ) , _ finite size scaling and numerical simulation of statistical systems _ , ( world scientific , singapore 1990 ) .
.finite - size estimates of the exponent @xmath314 for the spin1 model for two values of @xmath48 .
the line labeled @xmath419 gives the @xmath77 estimates and the numbers in brackets give the estimated uncertainty in the last given digit .
[ tab1 ] [ cols="^,^,^,^,^,^,^",options="header " , ] comparison of the magnetization profiles @xmath79 for the spin-@xmath405 ( boxes ) , the spin-1 ( full circles ) and the spin-@xmath406 ( open circles ) models .
the full curve gives the profile of eq .
( [ simpro ] ) which is correct to first order in @xmath48 .
the data are for @xmath82 and a ) @xmath81 and b ) @xmath85 . density of states ( per site ) @xmath424 as a function of energy @xmath58 ( relative to the center of the manifold ) within the manifold @xmath93 for @xmath269 .
the middle subband contains @xmath266 states .
the other subbands contain @xmath7 states .
the asymmetry of the subbands rapidly decreases as @xmath0 increases . inside each subband , near its edge at @xmath425 , @xmath426 for @xmath83 .
spectrum of @xmath36 for spin @xmath41 and @xmath339 as a function of @xmath48 for @xmath340 sites .
the two vertical lines indicate the location of the transitions between the o / f phases and the f / d phases , respectively , for @xmath77 . | we discuss several problems concerning domain walls in the spin @xmath0 ising model at zero temperature in a magnetic field , @xmath1 , applied in the @xmath2 direction .
some results are also given for the planar ( @xmath3-@xmath4 ) model in a transverse field .
we treat the quantum problem in one dimension by perturbation theory at small @xmath5 and numerically over a large range of @xmath5 .
we obtain the spin density profile by fixing the spins at opposite ends of the chain to have opposite signs of @xmath6 .
one dimension is special in that there the quantum width of the wall is proportional to the size @xmath7 of the system .
we also study the quantitative features of the ` particle ' band which extends up to energies of order @xmath5 above the ground state . except for the planar limit
, this particle band is well separated from excitations having energy @xmath8 involving creation of more walls . at large @xmath0
this particle band develops energy gaps and the lowest sub - band has tunnel splittings of order @xmath9 .
this scale of energy gives rise to anomalous scaling with respect to a ) finite size , b ) temperature , or c ) random potentials .
the intrinsic width of the domain wall and the pinning energy are also defined and calculated in certain limiting cases .
the general conclusion is that quantum effects prevent the wall from being sharp and in higher dimension would prevent sudden excursions in the configuration of the wall . |
solar jets which are small - scale plasma ejections along open field lines or the legs of large - scale coronal loops have been extensively investigated ( e.g. , schmieder et al .
1988 ; shibata et al .
1994 ; zhang et al .
2000 ; liu & kurokawa 2004 ; cirtain et al .
2007 ; jiang et al . 2007
; tian et al . 2014 ) .
they often recur from the regions of evolving photospheric magnetic flux , such as satellite spots of an active region ( ar ) , quiet regions , and coronal holes ( e.g. , chae et al . 1999 ; chen et al .
2008 ; pariat et al .
2010 ; yang et al .
2011a , 2011b ) .
the detailed statistical properties of x - ray jets were studied by shimojo et al .
( 1996 ) and savcheva et al .
( 2007 ) . probably due to the release of magnetic twist ,
helical or untwisting motions are often observed in jets ( e.g. , jibben & canfield 2004 ; liu et al . 2009 ; shen et al . 2011 ; chen et al . 2012 ) .
sometimes , jets are associated with large - scale solar eruptions , such as filament eruption and coronal mass ejections ( cmes ; e.g. liu et al . 2005 ; jiang et al .
2008 ; guo et al .
2010 ; shen et al .
2012 ) . by examining many x - ray jets in _ hinode_/x - ray telescope
coronal x - ray movies of the polar coronal holes , moore et al .
( 2010 ) found that nearly all solar polar x - ray jets can be divided into two different types : standard jets and blowout jets .
the standard x - ray jets are evidently produced by reconnection between an emerging magnetic arch at the base of the jet and the ambient unipolar open magnetic field of the coronal hole , during which the interior of the base arch is insert and does not participate in the eruption ( e.g. , yokoyama & shibata 1995 ; canfield et al .
1996 ; pariat et al . 2009 ) .
the blowout x - ray jets are produced by blowout eruption of the base small filament , a miniature version of the blowout eruption of the sheared - core magnetic arcade in a cme eruption .
the blowout jet model has been supported by some observational studies ( e.g. , hong et al .
2011 ; liu et al . 2011 ; shen et al .
2012 ; pucci et al . 2013 ; young & muglach 2014 ) .
however , the detailed process of the interaction between the erupting filament and the ambient open field is unclear .
the super solar ar 12192 during 2014 october consisted of the largest sunspot group in the past 24 years ( e.g. , sun et al .
2015 ; thalmann et al .
2015 ; yang et al .
it is so far the most intensely flaring region of cycle 24 . according to the statistics of chen et al .
( 2015 ) , ar 12192 produced 6 x - class and 29 m - class flares from october 18 to 29 , but only one of which was followed by a cme . in this study , using the observations of the new vacuum solar telescope ( nvst ; liu et al . 2014 ) and the _ solar dynamics observatory _ ( _ sdo _ ; pesnell et al . 2012 ) , we investigated the only one cme - associated flare from the periphery of ar 12192 , which was caused by a blowout jet ( chen et al .
our observations present the unprecedentedly clear process of the interaction between the erupting small filament and the ambient open " field , which is consistent with the blowout jet model . here and
in the following text , we use quotation marks around the word open " , considering that the
open " loops , which appear as rapidly fan - out extreme - ultraviolet ( euv ) structures , might be large - scale loops reaching large heights as calculated by thalmann et al .
the nvst at the _ fuxian solar observatory _ has three channels to image the sun , i.e. , h@xmath0 , tio , and the g band .
the h@xmath0 channel is used to observe the solar chromosphere with a bandwidth of 0.25 .
the h@xmath0 data used in this study were obtained in 6562.8 from 07:50:12 ut to 09:17:33 ut on 2014 october 24 with a cadence of 12 s. the nvst was pointed to ar 12192 with a field - of - view ( fov ) of 152@xmath1@xmath2157@xmath1 and a pixel size of 0.@xmath1168 .
the data are first calibrated , including flat field correction and dark current subtraction , and then reconstructed using speckle masking ( weigelt 1977 ; lohmann et al .
1983 ) .
we also adopt the atmospheric imaging assembly ( aia ; lemen et al .
2012 ) multi - wavelength images and the helioseismic and magnetic imager ( hmi ; scherrer et al . 2012 ; schou et al .
2012 ) data on board the _ sdo_. we choose the aia 131 , 171 , 304 and 1600 images , obtained from 07:00 ut to 09:20 ut on october 24 with a pixel size of [email protected] and a cadence of 12 s. we use the hmi full - disk line - of - sight ( los ) magnetograms with a 3 min cadence , i.e. , one frame in four , from 2014 october 23 12:00 ut to 2014 october 24 12:00 ut .
by investigating the evolution of the hmi magnetograms , we found that new magnetic flux started to emerge from 2014 october 21 in the southeast of the main negative polarity sunspot in ar 12192 .
about 3 days later , a blowout jet occurred at the emerging flux region , triggering an m4.0 flare and a cme .
the flare and the associated jet observed by the _ sdo _ and nvst near the peak time were shown in figure 1 .
the goes soft x - ray 1@xmath38 flux showed that the m4.0 flare initiated at 07:37 ut , and reached its peak at 07:48 ut ( see the red curve in panel ( b ) ) .
as shown in panel ( a ) , the east flare ribbon was located at the negative polarity fields and the west ribbon was at the positive polarity fields .
the appearances of the jet and the filament eruption were similar in higher - temperature wavelength of 131 and lower - temperature wavelengths of 171 and 304 ( panels ( b)-(c ) and ( e ) ) .
the two ribbons of the flare in the chromosphere were observed in 1600 image ( panel ( d ) ) . the jet ( the bright structure denoted by arrows a " ) and the filament ( the dark structure indicated by arrow b " ) were clearly observed in the h@xmath0 image ( panel ( f ) ) .
unfortunately , the observations of nvst started late and we can only use the _ sdo _ data to study the first eruption .
the temporal evolution of the first eruption is shown in figure 2 ( also see the accompanying animation movie_1.mp4 ) .
the filament was in the east - west direction , with a projected length of about 60 mm .
the east footpoint of the filament was anchored in the negative polarity magnetic fields and the west one is at the positive polarity fields ( panel ( a ) ) .
the blowout jet appeared at about 07:38 ut , and the material of the bright features was simultaneously ejected and moved towards the south ( indicated by the arrows in panels ( b)-(c ) ) .
the most distinct change of the filament is that it began to rise ( denoted by the arrow in panel ( d ) ) . from 07:44:19
ut , the projected height of the filament increased by about 5 mm in two minutes .
the region with euv brightenings was extended as seen in the aia 304 images and the flare reached its peak at 07:48 ut ( panel ( e ) ) .
some filament material was ejected outward ( denoted by the arrow in panel ( f ) ) . at 07:59 ut , the flare had greatly decayed ( panel ( g ) ) , and the filament erupted ( indicated by the arrow in panel ( h ) ) . at the decaying phase of the flare , partial material of the erupting filament fell back to the solar surface nearby its east footpoint . at about 08:12 ut , a cme was observed in the fov of lasco c2 .
the angular width of the cme was about 90and the average velocity was approximately 500 km s@xmath4 .
figure 3 displays the second eruption process of another small filament observed by aia 171 and 304 . the general appearance and the surrounding environment before the second eruption
is shown in panel ( a ) .
the second filament erupted at ( 40@xmath1 , -430@xmath1 ) about twenty minutes after the first filament eruption .
we can see that another blowout jet was associated with the filament eruption .
the blue arrows in panels ( b1 ) and ( c1 ) point to the filament before eruption .
a set of open " field lines were rooted in vicinity of the east footpoints of the filament , both anchoring in the negative polarity magnetic fields .
the open " lines ( denoted by the arrows in panel ( a ) ) were outlined by the falling dark material of the former jet .
the west footpoint of the filament was shielded by the material of the dark features of the open " lines , and seemed to be located at the positive polarity fields . at about 08:41 ut , the interaction between the
open " lines and the filament began , and meanwhile the euv brightenings were enhanced ( panels ( b2 ) and ( c2 ) ; also see the animation movie_2.mp4 ) .
the interaction continued and then the filament erupted finally ( panels ( b3 ) and ( c3 ) ) . with the high - resolution observations of the nvst ,
the detailed evolution process of the second eruption was analyzed ( see the accompanying animation movie_3.mp4 ) .
the filament and the open " lines are denoted in figure 4(a ) by arrows a " and b " , respectively . by comparing the h@xmath0 and 171 images , we find that the structure denoted by arrow
b " in panel figure 4(a ) corresponds to the
open " lines indicated by the blue arrows in figure 3(a ) .
therefore the structure b " in figure 4(a ) represents a set of open " lines .
the footpoint of the
open " lines was located at the negative polarity magnetic fields , lying at the northwest of the filament ( figure 4(a ) ) . then the open " lines began to interact with the filament and they gradually reconnected . at 08:40:03 ut , many
open " lines within the area marked by the brackets disconnected from their initial footpoints ( panel ( b ) ) .
the open " lines continued moving eastward and reconnecting with the filament . during the reconnection process , the interaction area of the filament and
the open " lines was brightened up as observed in euv lines , e.g. , aia 171 line ( denoted by the green arrow in figure 3(b2 ) ) . at 08:43:53 ut , the
open " lines totally connected with the filament , and a new system of the filament and open " lines was formed , as displayed in panel ( c ) .
after that , the filament started to untwist , triggering the rotation of the new system and leading to the brightening ( panel ( d ) ) .
the highly twisted system rotated continuously ( panel ( e ) ) , and lasted for about twenty minutes .
the rotation angle caused by the untwisting motion was about 8@xmath5 ( see next paragraph for detail ) .
at the late phase of the rotation , the twist of the system was significantly released ( panel ( f ) ) . the comparison of panels ( a ) and ( f ) showed that the footpoints of the
open " lines were shifted to the east end of the filament . as the system rotated , the magnetic twist propagated outward along open " lines ( see the accompanying animation movie_4.mp4 ) . to show the propagation process
well , we take a bright helical structure as an example . at 08:53:35 ut
, the projected position of the helical structure was located at ( 40@xmath1 , -475@xmath1 ) , as denoted by the arrow in panel ( g ) .
about four minutes later , the helical structure reached to a new position ( 25@xmath1 , -600@xmath1 ) ( indicated by the arrow in panel ( h ) ) .
the average moving velocity of the structure was estimated to be about 400 km s@xmath4 . for detailed analysis of the untwisting process of the filament and the open " lines
, we focus on the marked rectangle regions in figure 4(e ) , track the shift movement of dark fine structures , and mark the position changes of the structures with time ( see figure 5 ) .
the dark threads in the h@xmath0 images are considered to represent flux tubes of the filament . in order to measure the rotation angle of the filament , we draw a red line which is perpendicular to the filament spine . as the filament rotated ,
the intersection ( marked by the blue arrows ) between each dark thread and the red line shifted from left to right ( panels ( a1)-(a4 ) and ( b1)-(b4 ) ) , indicating a clockwise rotation .
we deduce the rotation angle from the shift distance . to illustrate how we measure the rotation of the filament , a cartoon is inserted into panel ( a ) .
the filament is considered to be a circular cylinder .
the width of the filament is @xmath6 and the distance from the intersection to the filament left edge is @xmath7 .
we define @xmath8 , and then the rotation angle @xmath9 between @xmath10 and @xmath11 is @xmath12 .
this measurement method works under the assumption that the filament consisting of dark threads rotates rigidly . because we do not trace and measure the movement of dark features along the field lines
, the material flow will not affect our results .
for one analyzed thread , it was on the left edge of the filament at the beginning ( panel ( a1 ) ) .
about 24 seconds later , the intersection position moved to the west for about 3@xmath1 ( shown in panel ( a2 ) ) . until 08:53:34 ut , the intersection moved about 6@xmath1 ( panel ( c ) ) .
we calculate the movement of the intersecting position every 12 seconds , until the thread moved to the right edge of filament ( indicated in panel ( a4 ) ) .
since this thread could not be observed at the far side of the filament , we choose another thread and track its evolution ( marked by the arrows in panels ( b1)-(b4 ) ) .
thus we can obtain the rotation angle and angular speed of the filament in two minutes , as plotted in panel ( c ) .
the angular velocity is not uniform , revealing the acceleration process followed by a deceleration .
the minimum and maximum angular speeds are 30 min@xmath4 and 230 min@xmath4 , respectively .
the average angular velocity is 100 min@xmath4 .
the rotation process lasted for about twenty minutes , the total twist angle of 8@xmath5 is reasonable since the angular velocity is not uniform .
generally , many mechanisms can result in the eruption of flux ropes , such as kink instability ( trk & kliem 2005 ) , and torus instability ( kliem & trk 2006 ) .
however , according to some previous studies , even for large twists , low - lying filaments , especially those with small length , still can keep stable ( e.g. , vrsnak 1990 ; isenberg & forbes 2007 ) .
with the _ sdo _ observations , we study the only one cme - associated flare in ar 12192 . there were two filament eruption processes at the edge of the ar .
the magnetic flux emerged continuously at the southeast of the negative main sunspot in this super ar , forming a twisted filament system at this area .
the first filament eruption caused a blowout jet , and produced an m4.0 flare and led to the only one cme . using the nvst high tempo - spatial resolution data
, we particularly study the evolution of the second filament eruption . the filament and
the open " lines reconnected with each other and formed a new system , the details of which were clearly observed by the nvst .
subsequently , the filament untwisted , and resulted the rotation of the new system .
the total rotation angle is about 8@xmath5 and the average angular speed is about 100 min@xmath4 . as the largest ar since
1990 november , noaa 12192 is the most flare productive ar . except for one m4.0 class flare , all the x - flares and the other m - flares which originated from the ar core were not associated by any cme .
according to the recent studies about this ar , the confined flares were affected by the strong confinement of the overlying magnetic fields , and the decay index of the background fields was below the typical onset threshold for the torus instability and thus no cme was produced ( sun et al .
2015 ; thalmann et al . 2015 ; chen et al .
however , the m class flare in the present study was located at the edge of the ar instead of in the core region , and a cme was produced .
we note that this flare was triggered by the blowout jet , and the flare became an eruptive flare due to the interaction between the flare material and open " fields , leading to the cme ( thalmann et al .
2015 ) . in the standard jet model ,
the jet is caused by the reconnection between the emerging small - scale loops and the pre - existing open " lines . while according to the blowout jet model , there exists a filament which is involved into the reconnection and eruption .
although many authors ( e.g. , hong et al .
2011 ; shen et al .
2012 ; pucci et al .
2013 ) have reported the observational evidences for the later model , the interaction process between the filament and the open " field lines have not been well observed . in this paper , the filament can be identified in multi - wavelengths , such as aia 304 , 171 , and 131 , and especially in the nvst h@xmath0 line . as revealed by the h@xmath0 image in figure 4(a ) , the location of one end of the filament was close to the
open " lines .
then the interaction between them began ( figure 4 ( b ) ) .
about 16 minutes later , their topology had been significantly changed , and the new connection between the filament and the open " lines was created ( figure 4 ( c ) ) .
thanks to the high - quality h@xmath0 data from the nvst , this detailed interaction process is , to our knowledge , first reported in the present paper . as pointed out by zhang & low ( 2005 ) , the accumulation of magnetic helicity in the corona can build up magnetic energy which will be responsible for the cme . when the stored free energy is higher than the aly limit ( aly 1984 , 1991 ) , eruptions take place . as a fraction of helicity in a system ,
magnetic twist can be inferred by filament structure .
thus we try to study magnetic twist through observing the evolution of filament structures . in the present study
, we show a scenario of twist transfer process between different systems and propagation from the lower atmosphere to the higher layers . as shown in figure 4(a ) , the twist was initially stored in the helical filament .
after the interaction between the filament and the open " lines , a new system connecting the filament and the open " lines was formed .
then the new system rotated due to the untwist of the filament , and thus the twist was transferred from the closed filament system to the open " system . during the rotating movement ,
the twist probably propagated from the lower atmosphere to the upper layers .
we speculate that the twist might be finally spread by the cme to the interplanetary space . | using the multi - wavelength images and the photospheric magnetograms from the _ solar dynamics observatory _ , we study the flare which was associated by the only one coronal mass ejection ( cme ) in active region ( ar ) 12192 . the eruption of a filament caused a blowout jet , and then an m4.0 class flare occurred .
this flare was located at the edge of ar instead of in the core region .
the flare was close to the apparently open " fields , appearing as extreme - ultraviolet structures that fan out rapidly . due to the interaction between flare materials and
open " fields , the flare became an eruptive flare , leading to the cme .
then at the same site of the first eruption , another small filament erupted . with the high spatial and temporal resolution h@xmath0 data from the new vacuum solar telescope at the _ fuxian solar observatory _ , we investigate the interaction between the second filament and the nearby open " lines .
the filament reconnected with the
open " lines , forming a new system . to our knowledge
, the detailed process of this kind of interaction is reported for the first time .
then the new system rotated due to the untwisting motion of the filament , implying that the twist was transferred from the closed filament system to the open " system .
in addition , the twist seemed to propagate from the lower atmosphere to the upper layers , and was eventually spread by the cme to the interplanetary space . |
for a decade , the passive remote sensing techniques based upon electromagnetic sources of opportunity have become more and more popular for detection and characterization purposes .
the most significant reason for this enthusiasm lies in the fact that the passive systems take advantage of the numerous existing electromagnetic emitting sources to come up with new approaches for remote sensing or surveillance applications . as a matter of fact , among all the possible sources of opportunity ,
the global navigation satellite systems gnss ( gps , galileo , glonass , ... ) appear among the most relevant solutions since the electromagnetic signals emitted ( in l - band ) are reliable , available all over the world , deterministic and perfectly known ( for direct paths ) .
it is worth noting that the use of gnss signals for passive observation has been applied to a significant extent to the field of oceanography since the year 2000 . indeed
, several studies have been carried out to estimate oceanographic parameters using the reflected gnss signals measured from ground - based systems , see for example @xcite , or airborne systems , see for example @xcite .
these systems can be used for the measurement of wind velocity and wind direction @xcite , surface roughness and its effects @xcite or even to estimate the salinity @xcite . for airborne and long range systems , the precision in time involves that the resolution cell easily exceeds a few hundred square meters in surface area .
then , the scattering by each individual sea wave can not be discriminated and described . in these cases ,
the sea surface is considered as a very large stochastic rough surface induced by the complex interaction between the fluid and the wind . in this context
, the reflected gnss is mostly seen as a tool to extract the statistical characteristics of a maritime environment ( root mean square deviation of the sea surface height for instance ) . for ground - based systems ( see @xcite for instance )
, the studies mainly focus on the delay / doppler analysis of the gps signals .
the modulation of the gps signal makes the sharp analysis on long coherent period quite difficult and limits the resolution in the time frequency domain .
the recent maritime opportunity passive systems ( mops ) @xcite research project considers a local maritime domain observed in the vicinity of the sea surface .
the prime objective of this project is to obtain a very high resolution analysis ( very long coherent period and high precision in time ) of the gps phase for the direct and reflected signals in order to estimate the sea surface movements . from preliminary studies
it appears that the obtained delay / doppler maps show quite complex patterns that remain difficult to interpret .
however , these patterns clearly suggests that this passive gnss system should be able to detect the movement of individual sea waves . the general purpose of this paper is closely related to this research project .
it investigates the possibility of observing the movement and the deformation of a sea surface from the reflected signal in l - band . in practice
, we investigate the connections between a deterministic time evolving surface and the time - frequency representation of the signal scattered by this surface .
the idea is to take advantage of these representations to extract oceanographic parameters from doppler and micro - doppler signatures . in a broad outline ,
a sea surface can be seen as a sum of ` sinusoidal ' surfaces at several scales .
basically , it can be expected that the large scale , corresponding to the gravitational waves , will be related to the global motion of the sea and the small scales , corresponding to the capillary waves directly induced by the wind , to the roughness of the sea surface . in this paper , we will concentrate on the interpretations of the time - frequency signatures of the signal scattered by the evolving sea surface .
these interpretations should be of a great importance for complementary works dealing with the estimation of oceanographic parameters .
we can expected to estimate , as already noted , the movements and the deformations of a sea surface ( i.e. the local sea state and the sea roughness ) but also the local wind velocity and direction .
first , our approach consists in generating time evolving ( two - dimensional ) surfaces and computing the field of a plane wave in l - band scattered by these surfaces .
the numerical model is based upon the method of moment ( mom ) which is a standard approach used to quantify the scattering by sea surfaces , in l - band .
then , the time - frequency signatures of the scattered signal are obtained using wigner - ville representations .
finally , making comparisons with the signatures produced by canonical surfaces ( moving sinusoids for instance ) , we provide interpretations of the complex signatures corresponding to more realistic sea surfaces .
the sequel of the paper is organized as follows . in section [ sec : em_model ] the electromagnetic model used to compute the scattered electromagnetic field is introduced .
section [ sec : sea_model ] describes the model applied to generate realistic evolving sea surfaces .
then , section [ sec : tfoverview ] presents a short overview of the time - frequency representations used in this paper .
sections [ sec : sinus ] and [ sec : sea ] give an interpretation of the time - frequency representations of the signal reflected from canonical and modeled sea surfaces .
finally , section [ sec : conclusion ] gives some concluding remarks and proposes some future works dealing with the extraction of the characteristics of the sea surface from the time - frequency representations .
roughly speaking , the gnss signal can be considered as a plane right hand circular polarization ( rhcp ) incident wave . for the sake of simplicity , we assume in the following that the satellites are at the zenith position , so the incident wave propagates in the nadir direction , see figure [ fig : configuration ] . in this configuration ,
the electromagnetic modeling consists in the evaluation of the scattering of a plane rhcp wave by a dielectric interface . as a matter of fact , due to the salinity of the sea water , the surface
is assumed to be a perfect electric conductor ( pec ) .
it is noteworthy that a circular polarized plane wave can be split into two linear polarized plane waves : transverse magnetic ( tm ) and transverse electric ( te ) polarization .
an appropriate approach to modeling the scattering by tm and te polarized wave is to consider the boundary integral equation method ( biem ) and the estimation using functional basis : method of moments ( mom ) . for the tm polarization ( @xmath0 ) ,
the modeling of the scattering is based upon the electric field integral equation ( efie ) @xcite : @xmath1 where @xmath2 is the incident field , @xmath3 is the permeability of free space , @xmath4 represents the sea surface , @xmath5 is the angular frequency of the electromagnetic wave , @xmath6 is the surface current , @xmath7 is the 2d green function , @xmath8 and @xmath9 are the source and observation points on the sea surface ( @xmath10 ) .
it is worth noting that the incident wave is known and is assumed to be tapered to avoid numerical truncation effects .
so , the amplitude @xmath11 of incident field will be given by @xcite : @xmath12 where @xmath13 is the tapering parameter and the additional factor in the phase @xmath14 is : @xmath15}{\left(kg_t\right)^2}.\ ] ] the 2d green function @xmath7 is also perfectly known , and the only unknown function of the integral equation ( [ e : efie ] ) is the surface current @xmath6 . when the surface is computed , the scattered field is determined using the following relation : @xmath16 where @xmath17 is the position of the observer above the sea surface .
for the te polarization , the electrical field @xmath18 belongs to the plane ( @xmath19 ) and the magnetic field is in the form ( @xmath20 ) .
so the modeling of the scattering in te polarization is based upon the magnetic field integral equation ( mfie)@xcite : @xmath21d\vec{\rho}\,',\ ] ] where @xmath22 is the normal vector to the sea surface at the position @xmath9 , @xmath23 is the gradient related to the second variable of the green function , @xmath24 is the incident field and @xmath25 is the surface current . in the same way as the tm polarization case , the integral equation ( [ e : mfie ] ) determines the surface current @xmath25 .
so , the magnetic field @xmath26 received by the observer is computed using a relation similar to equation ( [ e : mfie_bis ] ) , and then the electric field received by the observer can be evaluated . considering both tm and te polarizations and a standard method of moment ( mom )
approach to solve the integral equations ( [ e : efie ] and [ e : mfie ] ) ( a detailed presentation can be found in @xcite ) , we can simulate the scattering of a gnss signal ( rhcp plane wave ) by a sea surface .
the last point for the electromagnetic modeling that needs to be said is that the receiver antenna of the observer is left hand circular polarized ( lhcp ) to optimize the recording of the reflected signal .
so the actual received signal will be the lhcp component of the scattered field .
since the previously described electromagnetic model estimates the scattered field for any surface , the last step to obtain a complete simulation is the generation of the sea surfaces .
physically , a sea surface is induced by a complex and stochastic interaction between the fluid and the wind .
fortunately , a statistical description of the energy spectral decomposition of the sea surface can be approximated by empirical and/or semi - theoretical models for different wind speeds and wind directions .
one of the most consistent with the experimental data has been developed by elfouhaily et al .
this sea spectrum is in the form : @xmath27 where @xmath28 represents the isotropic part of the spectrum modulated by the angular function @xmath29 . @xmath30 and @xmath31 respectively denote the spatial wave number and the wind direction , see figure [ fig : elf ] .
( a ) ( b ) to generate a realistic sea surface associated to a given weather condition ( wind speed and wind direction ) , we expand the surface into a continuous sum of sinusoidal curves . for each spatial wave number @xmath30 ,
the mean amplitude is estimated using the square root of the sea spectrum and the phase is randomly set between 0 and @xmath32 . in practice ,
the convolution of the square root of the spectrum with a unitary white gaussian random signal generates a one - dimensional profile ( a statistical realization of the sea surface ) that represents an ocean surface for given weather conditions . to describe the time evolution of the sea surface thus obtained , we associate a time wave number @xmath33 for each @xmath30 using the phase dispersion relation : @xmath34 where @xmath35 is the acceleration due to gravitation and @xmath36 is the amplitude of the sinusoidal curve .
finally , we can generate a realistic time - evolving surface and using the electromagnetic model we can also estimate the signal recorded by the receiver .
now , we will demonstrate that these simulated signals analyzed with time - frequency tools lead to many physical interpretations .
the numerical simulation of the electromagnetic waves intercepted by an observer above a time - varying sea surface provides complex non - stationary signals . a relevant approach
to fully describe the nature of these non - stationary signals consists in a time - frequency ( tf ) analysis . indeed ,
a time - frequency distribution ( tfd ) shows how the spectral content of the signal evolves with time , thus providing an ideal tool to dissect , analyze and interpret non - stationary signals .
most of time , the time - frequency distribution ( tfd ) maps the energy of a one - dimensional time - domain signal into a two - dimensional function of time and frequency . in this article
, we will stress the fact that the feature extraction based on time - frequency analysis of an electromagnetic field above the sea can be exploited to point out useful information about the signal fluctuation induced by the movement of the surface .
a great variety of methods for obtaining a tfd have been defined , most notably the short time fourier transform , the wavelet transform and the wigner - ville distribution . for more details
the reader can refer for example to @xcite and the time - frequency toolbox ( tftb ) .
this article deals with the wigner - ville distribution .
the wigner - ville distribution ( wvd ) of a signal @xmath37 , denoted by @xmath38 , is defined as @xmath39 where @xmath40 is the analytic associate of @xmath37 ( i.e the hilbert transform of the real signal @xmath37 ) .
the main properties of the wvd are : it is always real - valued , it preserves time and frequency shifts and it satisfies the marginal properties . unfortunately , the wvd shows interference terms .
these interference terms are troublesome since they may overlap with signal terms and thus make it difficult to visually interpret the wvd image .
basically , the interference between two points in the time - frequency plane correspond to the appearance of a third point located at the geometrical midpoint . besides , the interference terms oscillate perpendicularly to the line joining the two interfering points , with a frequency proportional to the distance between these two points . in order to reduce the interference terms , the pseudo - wigner - ville distribution ( pwvd ) can be used .
this transform is described in the following .
the pwvd is defined as @xmath41 where @xmath42 is a regular window .
this windowing is equivalent to a frequency smoothing of the wvd .
it leads to the attenuation of the interference terms .
however , an excessive window leads to a loss in properties and may damage the joint - time - frequency resolution .
as already mentioned , the electromagnetic field scattered by a realistic time - varying sea surface induces complex non - stationary signals received by the observer .
so , in a first phase , we suggest generating and analyzing the received signal when the simulated sea corresponds to a canonical moving surface ( a sinusoid for instance ) . in this section , we will attempt to provides several interpretations of the time - frequency signatures for simple or more sophisticated canonical reflecting surfaces .
this step strikes us as necessary to understand and explain the main features obtained in the time - frequency domain for more complex reflecting surfaces . in what follows
, we will considers the increasing complexity of the simulated reflecting surface : from the case of a simple sinusoidal surface to the case of the linear superposition of several sinusoidal surfaces .
let the sea surface be approximated , for its global shape and motion , by a simple sinusoid @xmath43 defined as @xmath44 the main parameters of this surface are the amplitude @xmath45 , the velocity @xmath46 , the angular frequency @xmath47 or the wavelength @xmath48 and the wavenumber @xmath49 . in ( [ eq : sin ] ) ,
@xmath50 stands for the time and @xmath51 for the position . to come closer to a realistic sea model
, the parameters of this sinusoidal model should be inferred from the sea surface model introduced in section [ sec : sea_model ] . for given weather conditions ( wind speed and wind direction ) , the wavenumber of the sinusoid @xmath52 is fitted to the abscissa @xmath53 of the maximum of omnidirectional elevation spectrum component , see figure [ fig : elf ] . since the wavenumber @xmath52 is set , the angular frequency @xmath54 can be computed using the phase dispersion relation ( [ e : dispersion ] ) .
the wavelength @xmath55 and the velocity @xmath46 can then be deduced . to determine the amplitude of the sinusoid
, we generate realistic sea surfaces from the elfouhaily spectrum .
then , the standard deviation @xmath56 of the wave height is computed . in physical oceanography ,
the global characteristic is given by the significant wave height ( swh or hs ) which is defined traditionally as the mean wave height ( trough to crest ) of the highest third of the waves ( @xmath57 ) .
in practice , the swh is usually defined as four times the standard deviation of the surface elevation : @xmath58 .
this is why the sinusoid amplitude is set to @xmath59 .
the estimated parameters defining the sinusoidal surfaces are given in table [ tab : beaufort ] .
note that , in this paper , only the beaufort scales from 1 to 5 are considered .
.sinusoidal sea surface parameters . [
cols="^,^,^,^,^,^ " , ] these simplifying assumptions and the estimated parameters have been experimentally validated by comparing the obtained evolving sinusoidal surfaces , for several values of the beaufort scale , with the global shape and motion of several sea surfaces ( generated using the section [ sec : sea_model ] model ) at the same beaufort scales .
figure [ fig : ex_surface ] shows a @xmath60 m sinusoidal surface at time @xmath50 and the corresponding sea surface for scale 3 .
the main visual difference is the wavelength of both surfaces .
it clearly appears from figure [ fig : ex_surface ] that the canonical surface corresponds to only one wavenumber and the simulated sea surface to a mixture of a wide frequency spectrum .
however , the sinusoidal surface obtained remains broadly an acceptable model in the first instance .
considering the previous sinusoidal surface and the electromagnetic model presented in section [ sec : em_model ] , we are able to simulate the signal measured at the receiver over a period of @xmath61s .
then , applying a pseudo - wigner - ville distribution ( pwvd ) , a two - dimensional representation of this signal can be achieved in the time - frequency domain .
figure [ fig : ex_surface_sin ] presents the time - frequency representation of the received signal reflected from surfaces corresponding to the levels 2 , 3 and 4 in the beaufort scale . at first sight
, the time - frequency feature seems very close to a sinusoid .
this is mainly due to the fact that the scattering phenomenon becomes time - periodic in the case of a sinusoid surface .
nevertheless , a closer examination shows that the time - frequency feature looks more like a distorted sinusoid , see figure [ fig : tf_sinus_surf_3 ] .
we must not forget that the tf representation shows the variations of the doppler frequency observed by the receiver induced by the motion of the sinusoid surface . finding the relation between the evolving sea surface elevation and the doppler variations is the key challenge in analyzing the pwvd and interpreting the time - frequency signature .
in order to give a useful interpretation , let us denote by @xmath62 , @xmath63 , @xmath64 and @xmath65 four significant points observed in the pwvd of the signal received for a sinusoid surface , see figure [ fig : tf_sinus_surf_3 ] .
the abscissa of these points are the time delays @xmath66s , @xmath67s , @xmath68s and @xmath69s .
the points @xmath63 and @xmath65 respectively correspond to a maximum and a minimum of the time frequency signature and the points @xmath62 and @xmath64 correspond to two doppler zero - crossings .
( middle ) .
the white circles point out the time - frequency signature at four different times : @xmath66s , @xmath67s , @xmath68s and @xmath69s . ]
the doppler frequency is zero ( @xmath62 and @xmath64 ) when a maximum or a minimum of the sinusoidal surface is under the receiver position ( see figure [ fig : sinus_surf_3 ] ) .
in fact , due to the tapering of the incident field , the incident wave mainly impinges on central area of the sea surface and using the approximation of the ray theory , the main electromagnetic contribution coming from the sea surface is induced by the specular point situated at the top ( or the bottom ) of the sea surface , see figure [ fig : zero_doppler ] .
the velocity associated to this specular point ( red arrow in figure [ fig : zero_doppler ] ) is strictly perpendicular to the observer direction , so the doppler shift is zero .
note that in figure [ fig : zero_doppler ] , and also in figure [ fig : max_doppler ] , the black arrows stand schematically for the intensity ( length of the arrow ) and direction of the incident wave .
numerically , if we remember that the wavelength of the sinusoidal surface is @xmath70 , the velocity is @xmath71 and the time period is @xmath72 , we note that @xmath73 exactly corresponds to half of the @xmath74 period .
( a ) ( b ) on the other hand , the doppler frequency is maximal or minimal ( @xmath63 and @xmath65 ) when the amplitude ( see figure [ fig : sinus_surf_3 ] ) of the sinusoidal surface above the observer is close to zero ( i.e. the inflexion point of the sinus ) .
figure [ fig : max_doppler ] illustrates this configuration and shows the velocity vector at this point .
however , the extrema of the doppler signature do not strictly correspond to the zero - crossings of the sea surface , and @xmath75s@xmath76s is an approximation of half of the @xmath74 period .
thus we can say that the doppler signature is a distorted sinusoid .
furthermore , a simple calculation shows that the velocity vector at the point above the observer does not correspond to the doppler shift at point @xmath63 or @xmath65 .
( a ) ( b ) in the sequel , we show that it is possible to make a further , detailed analysis of the time - frequency signature by drawing analogies from the ray theory and explaining the shape of the signature by the presence of specular points .
more precisely , the feature in the time - frequency domain for this kind of surface is due to the motion of the specular point which reflects maximum energy in the receiver direction . to illustrate this , figure [ fig : sinus_spec_3 ] shows the pwvd of the received signal after reflection on the sinusoidal surface and the evolution ( black line ) of the doppler frequency due to the single moving specular point ( for the considered sinusoidal surface only one specular point exists at each time ) . in practice ,
the specular point is geometrically determined in the central lobe of the incident wave and its doppler frequency is computed as : @xmath77 where @xmath78 and @xmath79 are respectively the distances emitter - target and target - receiver . at l - band , for a simple sinusoidal surface and the considered setup ,
the mom model and the single point source ( specular point ) approximation have significant differences in quantitative terms .
however , in terms of qualitative description , both approaches provide very similar time frequency signatures .
this can be easily illustrated by generating the signal @xmath80 measured at the receiver due to the specular point considered as a source point .
this signal is obtained from @xmath81,\ ] ] where @xmath82 is the distance covered by the electromagnetic wave , @xmath83 the frequency of the incident wave and @xmath84 the electromagnetic wave number . figure [ fig : simu_scat ] shows the pwvd obtained using this ` point source model ' .
the feature amplitudes are quite different since a basic source model does not take into account the complex phenomena of attenuation .
nevertheless , the pwvd in figure [ fig : simu_scat ] is very nearly comparable to that obtained in figure [ fig : sinus_spec_3 ] . .
] specifically , the specular point interpretation offers a more thorough and intuitive analysis of the sinusoid feature in the time - frequency domain ( see figure [ fig : sinus_3_zoom ] ) .
indeed , the figures [ fig : distortion ] show the different positions of the central specular point for a half and a full period .
the cyclic movement of the specular point clearly appears .
it should nevertheless be noted that the motion of the specular point follows an elliptical path with a non - uniform speed .
the specular point travels at a lower speed in the vicinity of the top of the sinusoidal sea surface ( @xmath62 ) than near the bottom ( @xmath64 ) . )
.,width=377 ] ( a ) ( b ) finally , the non - linear variations of the velocity when the specular point goes over the minima or the maxima of the sinusoidal surface is the reason for the time - frequency feature corresponding to a distorted sinusoid : @xmath85 . obviously , the monochromatic model of the sea surface is far from any acceptable level for a realistic interpretation .
according to the spectral model of the sea surface previously presented ( see section [ sec : sea_model ] ) , a sea is characterized by relevant physical phenomena at various scales : the surface with large wavelength and high amplitude stands for the gravity wave ( large - scale roughness ) and the surface with smaller wavelength and amplitude stands for the capillary and short gravity waves ( small - scale roughness ) . in order to come closer to reality
, we consider the same previous sinusoidal surface to which a second sinusoid with smaller amplitude and wavelength is added
. figure [ fig:2sinus_surf ] shows examples of such surfaces for a beaufort scale 3 . for this new
generated surface , the parameters of the second sinusoid correspond to the amplitude of the first divided by three , the velocity remains the same and the wavelength is divided by four .
the time - frequency representation of the simulated signal obtained is given in figure [ fig:2sinus_a3 ] .
in the same manner as outlined above , the two sinusoid models can lead to a specular point interpretation .
nevertheless , the sum of sine curves with different wavelengths involves local curvature phenomena and several specular points may appear in the central area . in figure [ fig:2sinus_a3 ] ( bottom ) the doppler frequencies associated to each specular point
are superimposed ( in black ) on the time - frequency distribution .
it is worth noting that they seem to generate a global continuous doppler curve in the time - frequency domain . in figure [ fig:2sinus_spec_zoom ]
we have highlighted the doppler frequency associated with the present specular points at each time with different color .
moreover , the same color is used to highlight the motion of a given specular point throughout a period . in figure [ fig:2sinus_a3 ] ( bottom ) the evolution of the doppler frequency due to the specular point in the monochromatic sea model ( same as figure [ fig : sinus_spec_3 ] )
is also added in white .
the doppler curve due to the two sinusoids can be seen as the so - called @xcite micro - doppler phenomenon ( compare to the doppler phenomena associated with the global signature due to the single sinusoid with larger amplitude ) .
it appears clearly that there is a global trend highlighted by the white curve which is due to the motion associated with the main sinusoidal component ( monochromatic model ) .
in addition , the association of the two sinusoidal components induces oscillatory variations around this main trend .
in the present case , since the ratio between both sinusoids is an integer , the sea surface considered is deterministic and periodic .
so , the tfr is also periodic .
let us consider a zoom into the tfr focusing on one period ( see figure [ fig:2sinus_spec_zoom ] ) .
this can be associated with figure [ fig:2sinus_surf ] which shows the surface at several times and the position of the corresponding specular points ( red star ) .
highlighted by the black , blue and red dots is the doppler associated to each specular point and at each time.,width=377 ] closer examination of the time - frequency representation reveals the number of specular points and their motion .
for instance , at the time @xmath86s ( see figure [ fig:2sinus_spec_zoom ] ) , three different intersections exist on the doppler curve that can be related to the time - frequency components of three specular points , see figure [ fig:2sinus_surf ] . on the other hand , at the time @xmath87s ,
there is only one intersection on the doppler curve , and there exists only one specular point , see figure [ fig:2sinus_surf ] . despite the good agreement between mom and the specular point approximation ,
it is important to add that the comparison must not be pushed too far in the case of a sea surface with multi sinusoid components . in the same way as for the monochromatic sea surface ( see figure [ fig : simu_scat ] ) , figure [ fig:2sinus_scat ] shows the tfr of the electromagnetic field computed using the specular point source model .
overall , this image has many features in common with figure [ fig:2sinus_a3 ] . however ,
significant differences can be pointed out between both images .
this can be explained largely by the fact that multi sinusoid components involve local curvatures upon the sea surface .
so , the electromagnetic scattering induces multi path reflection and more complex interactions that are not taken into account by the specular point approximation . with a view to improving our sea surface model ,
the surface considered now consists of the sum of three sinusoids with different wavelengths and amplitudes and above all with different velocities .
it is still considered a surface with parameters corresponding to beaufort scale 3 .
the two added sinusoids are fixed so that the wavelength is respectively divided by 3 and 5 , the amplitude is divided by 2 and 5 and the velocity is multiplied by 1.25 and 1.6 . in previous cases , the movement of the sea can be summarized into a global translation at a constant speed , see figures [ fig : sinus_surf_3 ] and [ fig:2sinus_surf ] .
in reality , the sea surface has to be considered as a dispersive medium for the sea waves ( speed depends upon the wavelength , see equation ( [ e : dispersion ] ) ) .
this dispersion induces deformations over time in addition to the translation .
this last issue introduces new physical phenomena and modifies the tf features . the tfr obtained for the three sinusoid models
is presented in figure [ fig : tf_3sinus ] .
as in the case of two sinusoid models , the tfr allows us to view the micro - doppler phenomena ( oscillatory variations around the main trend ) .
nevertheless , the structure of these oscillations seems slightly more complex , in the present case .
once again , to understand the tfr , we can make use of the specular point approximation .
figure [ fig : tf_3sinus_scat ] shows the tfr of the electromagnetic field approximated with the specular point source model . despite significant differences , we do see that the global shape is very similar to figure [ fig : tf_3sinus ] .
therefore the specular point approximation can , to large extent , provide a relevant description .
the tf representation for three sinusoid sea surfaces brings about quite complex phenomena that deserve more sustained analysis .
thus figure [ fig : tf_3sinus_zoom ] shows zooms of two areas of interest in figure [ fig : tf_3sinus ] : from @xmath88s to @xmath89s and from @xmath90s to @xmath91s .
in addition , figures [ fig : tf_3sinus_surf1 ] and [ fig : tf_3sinus_surf2 ] show the sea surface with the specular points at several successive times in these two periods . between @xmath92s and @xmath93s
there is only one specular point inducing a doppler curve that appears to be oscillating . in figure
[ fig : tf_3sinus_surf1 ] , we can see that in a first step ( @xmath92s/2.14s ) the specular point moves away from the receiver with a speed greater than the global sea movement .
then doppler frequency is negative and lower than that obtained for a harmonic sea surface . in a second step ( @xmath94s/2.26s )
the specular point locally tends to move closer to the receiver . the doppler frequency increases and even reaches a positive maximum . in the last step ( @xmath95s/2.39s ) , the specular point speeds up the shift to the right , and doppler frequency returns to negative domain .
this oscillation can not be explained by a global translation of the sea surface but must be seen as the consequence of the local sea surface deformations related to dispersion .
the deformations due to dispersion increase the complexity of the local curvature observed at the sea surface . between @xmath96s and @xmath97s ,
the number of specular points changes over time .
figure [ fig : tf_3sinus_surf2 ] shows that there is only one specular point from @xmath96s to @xmath98s .
then , a new specular point appears at @xmath99s .
finally this new point forks into two new specular points .
both specular points move closer to the observer but with different velocities . in the same way
, we can see that dispersion and the local deformations of the sea surface periodically lead to the disappearance of one or more specular points .
somehow we must highlight the fact that the doppler curves associated to each specular points still form a continuous global doppler curve . after considering the scattering by canonical surfaces , in this section , we investigate the tfr of the electromagnetic signal obtained for the scattering by realistic sea surfaces .
clearly , the surfaces generated using the theory presented in section [ sec : sea_model ] are far more complex that those obtained using sinusoids or the sum of several sinusoidal functions .
however , we are going to find most of the previous expounded phenomena that influence the time - frequency features .
figures [ fig : tf_mer_234 ] shows the tfr obtained from the signal reflected from sea surfaces with respectively beaufort scale 2 , 3 and 4 . for the sea surface with beaufort scale 2 , from the receiver point of view , the surface is almost flat which is why the time frequency feature is mainly focused upon the zero - doppler frequency line .
the oscillations around this strip , which can be seen as micro - doppler phenomena , are linked with the small oscillation of the surface .
for the two other sea states , the surface can no longer be considered to be almost flat and then the tf feature appears as a far more structured geometry .
these tf representations suggest that there is great potential for feature extraction and specialized remote sensing applications . at first glance
, we can see that the spreading of the doppler frequency is in close conjunction with the sea state .
this is illustrated in figure [ fig : dist_nrj ] which shows the ( normalized ) power distribution according to the doppler frequency obtained from several sea states .
it can also be notated that the micro - doppler signature is directly related to the surface roughness and the fluid dynamics of the sea surface .
figure [ fig : tf_mer_3_spec ] shows the tfr and ( stacked ) the evolution ( blue points ) of the doppler frequencies of the specular points .
the beaufort scale of the sea surface considered is 3 .
it is clear that the understanding of this tfr image is much more complicated than canonical cases .
moreover , it is difficult to assert whether the doppler curves related to each specular point form a continuous global curve in a present case .
however , the cloud of points corresponding to each doppler specular point bears similarities with the doppler curve obtained in the two or three sinusoid surface cases . in order to improve readability , figure [ fig : tf_mer_3_spec_b ]
focuses on a restricted time domain and figure [ fig : tf_mer_3_surf_spec ] represents the snapshots of the sea surface and the computed specular points ( red stars ) at specific times : @xmath90s , @xmath100s , @xmath101s , @xmath102s , @xmath103s and @xmath104s . .
black lines show the times corresponding to the sea surface drawn in figure [ fig : tf_mer_3_surf_spec].,width=453 ] first of all , it is noticeable that the tfr induced by a realistic sea surface looks like the doppler curves observed in canonical cases at certain times ( @xmath101s for instance ) .
these situations most often occur when specular points are fewer and form a gathering of points as in figure [ fig : tf_mer_3_surf_spec ] . in these situations , it becomes easy to make comparisons with the canonical cases previously studied and it may give rise to the very simple physical interpretations given in section [ sec : sinus ] .
conversely , when the distribution of the specular points spreads over a larger region ( between @xmath90s and @xmath100s ) , the time frequency signature becomes more complex and is probably the consequence of multi - scale phenomena .
furthermore , figure [ fig : tf_mer_3_surf_spec ] shows that realistic sea surfaces often cause non uniform specular point distributions . in certain cases ,
the distribution of specular points is structured as different clusters . in the latter case ,
the tfr is characterized by multi component electromagnetic interactions . in any case , it is reasonable to assume that deterministic or statistical analysis of the tfr related to a given sea surface could provide significant information about the fluid dynamics of the sea . in a first instance , the analysis could consist in an extraction of the global tfr structure ( pseudo periodic structure ) . in a second phase ,
the analysis could focus upon the specific tfr structures related to the multi - scale phenomena .
progress will be made in future works on this matter .
this paper investigates the feature in the time - frequency domain of l - band signals reflected from sea surfaces .
the data used have been generated using a mom method and the tfr have been obtained from the tftb toolbox .
it has been shown that the signatures of the reflected signals in the tf domain are linked with the physics of the fluid dynamics .
these proposed results provide the basis for interpreting the scattering of l - band waves by time varying sea surfaces .
more precisely , it has been shown that the specific patterns ( looking like sigmoid functions ) can be linked to the specular reflections of the incident wave on the sea surface .
in addition , we have seen that local oscillations of the time frequency curve could be related to dispersion phenomena of the fluid surface .
this preliminary study opens up new opportunities for gnss signals ( l - band signals ) as remote sensing systems for dynamic sea surfaces .
future works will focus on the estimation of oceanographic parameters from these time - frequency analyses of reflected signals .
eventually , this study will later be extended to the analysis of non - linear sea or breaking waves @xcite .
this work was supported by ` gis europle mer ' and ensta bretagne .
j. f. marchan - hernandez , m. vall - llossera , a. camps , n. rodriguez - alvarez , i. ramos - perez , e. valencia , x. bosch - lluis , m. talone , j. m. tarongi , and m. piles . ground - based gnss - r measurements with the pau instrument and their application to the sea surface salinity retrieval : first results . in _
ieee international geoscience and remote sensing symposium ( igarss ) _ , volume 4 , pages 530533 , 2008 .
j. l. garrison , j. k. voo , s. h. yueh , m. s. grant , a. g. fore , and j. s. haase .
estimation of sea surface roughness effects in microwave radiometric measurements of salinity using reflected global navigation satellite system signals .
, 8(6):11701174 , 2011 .
j. k. voo , j. l. garrison , s. h. yueh , m. s. grant , a. g. fore , j. s. haase , and b. clauss
. application of reflected global navigation satellite system ( gnss - r ) signals in the estimation of sea roughness effects in microwave radiometry . in _ 11th specialist meeting on microwave radiometry and remote sensing of the environment ( microrad ) _ , pages 140145 , 2010 .
zavorotny , a.g .
voronovich , s.j .
katzberg , j.l . garrison , and a. komjathy .
extraction of sea state and wind speed from reflected gps signals : modeling and aircraft measurements . in _
ieee international geoscience and remote sensing symposium ( igarss ) _ , volume 4 , pages 15071509 , 2000 .
v. u. komjathy , a.and zavorotny , p. axelrad , g. h. born , and j. l. garrison .
signal scattering from sea surface : wind speed retrieval using experimental data and theoretical model .
, 73(2):162174 , 2000 .
d. r. thompson , l. a. linstrom , r. f. gasparovic , and t. m. elfouhaily .
surface roughness estimation from gps sea reflections . in
_ ieee international geoscience and remote sensing symposium ( igarss ) _ , volume 2 , pages 12781280 , 2002 . | passive remote sensing techniques have become more and more popular for detection and characterization purposes .
the advantage of using the global navigation satellite systems ( gnss ) are the well known signals emitted and the availability in most areas on earth . in the present paper , l - band signals ( including gnss signals )
are considered for oceanographic purposes .
the main interest in this contribution is the analysis of the signal reflected by an evolving sea surface using time - frequency transforms .
the features which occur in this domain are examined in relation to the physical phenomena : interaction of the electromagnetic waves with the moving sea surface .
* qualitative analysis of the time - frequency signature induced by a reflected l - band signal from time evolving sea surfaces * arnaud coatanhay and alexandre baussard + _ ^1^ensta bretagne / lab - sticc ( umr cnrs 6285 ) , brest , france + ( dated : december 17 , 2013 ) _ |
the control theory is the analysis of controlled dynamical systems .
these systems are varied : they can be differential , stochastic or discrete .
the optimal control theory concerns the determination of a control optimizing a certain cost .
consequently , this theory is strongly connected to the 18@xmath0 century classical mechanic ( variational principles , euler - lagrange equations , etc . , see @xcite ) .
since the second world war , this theory has a considerable development and one can find applications in many domains : celestial mechanic @xcite , biology @xcite , hydroelectricity @xcite , economy @xcite , etc .
the subject is widely treated and one can find a lot of references on the subject , see for example @xcite . the fractional calculus , _ i.e. _ the mathematical field dealing with the generalization of the derivative to any real order , plays an increasing role in many varied domains as economy @xcite or probability @xcite .
fractional derivatives also appear in many fields of physics ( see @xcite ) : wave mechanic @xcite , viscoelasticity @xcite , thermodynamics @xcite , fluid mechanic in heterogeneous media @xcite , etc .
a natural question then arises : _ can we develop optimal control theories for fractional differential systems ? _ recently , a subtopic of the fractional calculus gains importance : it concerns the variational principles on functionals involving fractional derivatives .
this leads to the statement of fractional euler - lagrange equations , see @xcite .
a direct consequence is the emergence of works concerning a particular class of fractional optimal control problems , see @xcite and references therein .
these studies usually use the lagrange multiplier technique allowing to write these problems as problems of optimization without constraint of _ augmented functionals_. with a calculus of variations , authors then obtain a necessary condition for the existence of an optimal control .
this condition is commonly given as the existence of a solution of a system of fractional differential equations called _ fractional pontryagin s system_. + in this paper , we first give a new presentation of this result .
precisely , making an additional assumption ( see condition ) , we rewrite _ directly _ these fractional optimal control problems as simpler problems of optimization without constraint of functionals _ depending only on the control_. with a calculus of variations , we finally conclude to the same necessary condition but with a new presentation .
although the method here used is considerably inspired by the lagrange multiplier technique , it allows us to give a complete proof using only classical mathematical tools adapted to the fractional case : _ calculus of variations _ , _ gronwall s lemma _ , _ cauchy - lipschitz theorem _ and _ stability under perturbations of differential equations_. nevertheless , the explicit computation of controls satisfying the above necessary condition needs the resolution of the fractional pontryagin s system which is a main drawback .
indeed , solving a fractional differential equation is in general very difficult .
consequently , in this paper , we suggest a deviously way in order to get informations on the solutions of a fractional pontryagin s system .
precisely , we study the existence of classical conservation laws , _ i.e. _ functions which are constant on each solution .
indeed , constants of motion , generally associated to physical quantities , give strong informations on the solutions in the phase space for example . moreover ,
they also can be used in order to reduce or integrate the system by quadrature .
previous results in this direction have been obtained by torres and frederico in @xcite .
however , in each of these papers , the conservation law is not explicit but implicitly defined by a functional relation . in this paper , inspired by a recent result obtained in @xcite , we prove a fractional noether s theorem providing an explicit conservation law for fractional pontryagin s systems exhibiting a symmetry . in this paper
, we also suggest : * a link between classical optimal control problems and their fractional versions via the stanislavsky s formalism ( see @xcite ) ; * a solved fractional example allowing to test numerical schemes in the strict fractional case . + the paper is organized as follows .
firstly , we remind some classical definitions and results concerning fractional calculus ( section [ section1 ] ) .
then , we present the class of fractional optimal control problems studied and the usual strategy using the lagrange multiplier ( section [ section21 ] ) .
then , assuming an additional condition on the problem , we develop a new proof leading to the same result with a different presentation , see section [ section22 ] .
we then suggest a way in order to transit from a classical optimal control problem to its fractional version via the stanislavsky s formalism ( section [ section3 ] ) .
subsequently , we suggest some examples with , in particular , a solved fractional example allowing to test numerical schemes in the strict fractional case ( section [ section4 ] ) .
finally , in section [ section5 ] , we state a fractional noether s theorem for fractional pontryagin s systems admitting a symmetry . technical proofs of lemmas are provided in appendix [ appa ] .
let us introduce the following notations available in the whole paper .
let @xmath1 be two reals , let @xmath2 , @xmath3 denote two dimensions and let @xmath4 be the euclidean norm of @xmath5 and @xmath6 .
the fractional calculus concerns the extension of the usual notion of derivative from non - negative integer orders to any real order . since 1695 , numerous notions of fractional derivatives emerge over the year , see @xcite . in this paper
, we only use the notions of fractional integrals and derivatives in the sense of riemann - liouville ( 1847 ) and caputo ( 1967 ) whose definitions are recalled in this section .
we refer to @xcite for more details . + let @xmath7,{\mathbb{r}}^d)$ ] and @xmath8 .
the left ( resp .
right ) fractional integral in the sense of riemann - liouville with inferior limit @xmath9 ( resp .
superior limit @xmath10 ) of order @xmath11 of @xmath12 is defined by : @xmath13a , b ] , \ ; i^{\alpha}_- g ( t ) : = \dfrac{1}{\gamma ( \alpha ) } \di \int_a^t ( t - y)^{\alpha -1 } g(y ) \ ; dy\ ] ] respectively : @xmath14 where @xmath15 denotes the euler s gamma function .
let us remind that @xmath16 ( resp .
@xmath17 ) is continuous and can be continuously extended by @xmath18 in @xmath19 ( resp .
@xmath20 ) . let us note that @xmath21 ( resp .
@xmath22 ) coincides with the anti - derivative of @xmath12 vanishing in @xmath19 ( resp .
@xmath20 ) . for @xmath23 ,
let @xmath24 . +
now , let us consider @xmath25 .
the left ( resp .
right ) fractional derivative in the sense of riemann - liouville with inferior limit @xmath9 ( resp .
superior limit @xmath10 ) of order @xmath26 of @xmath12 is then given by : @xmath13a , b ] , \ ; { d^\alpha_-}g(t ) : = \dfrac{d}{dt } \big ( i^{1-\alpha}_- g \big ) ( t ) \quad \big ( \text{resp . }
\quad \forall t \in [ a , b [ , \ ; { d^\alpha_+}g(t ) : = -\dfrac{d}{dt } \big ( i^{1-\alpha}_+ g \big ) ( t ) \big),\ ] ] provided that the right side terms are defined . + in the riemann - liouville sense , the strict fractional derivative of a constant is not zero .
caputo then suggests the following definition . for @xmath25 , the left ( resp .
right ) fractional derivative in the sense of caputo with inferior limit @xmath9 ( resp .
superior limit @xmath10 ) of order @xmath26 of @xmath12 is given by : @xmath13a , b ] , \ ; { { } _ { { \rm c } } d^\alpha_-}g(t ) : = { d^\alpha_-}\big ( g - g(a ) \big ) ( t ) \quad \big ( \text{resp . }
\quad \forall t \in [ a , b [ , \ ; { { } _ { { \rm c } } d^\alpha_+}g(t ) : = { d^\alpha_+}\big ( g - g(b ) \big ) ( t ) \big),\ ] ] provided that the right side terms are defined .
let us note that if @xmath27 ( resp .
@xmath28 ) , then @xmath29 ( resp .
@xmath30 ) .
+ in the _ classical case _ @xmath31 , the fractional derivatives of riemann - liouville and caputo both coincide with the classical derivative .
precisely , modulo a @xmath32 term in the right case , we have @xmath33 . + finally , let us remind the following important result .
if @xmath34,{\mathbb{r}}^d)$ ] and @xmath25 , then we have : @xmath35a , b ] , \ ; { { } _ { { \rm c } } d^\alpha_-}g(t ) = i^{1-\alpha}_- \dot{g } ( t ) \quad \text{and } \quad \forall t \in [ a , b [ , \ ; { { } _ { { \rm c } } d^\alpha_+}g(t ) = - i^{1-\alpha}_+ \dot{g } ( t ) , \ ] ] where @xmath36 denotes the derivative of @xmath12 .
consequently , in this case , we have @xmath37 ( resp .
@xmath38 ) is continuous and can be continuously extended in @xmath19 ( resp .
@xmath20 ) . let us precise that the properties and results developed in this section are well - known and already proved in many references , see @xcite for example .
however , these results are not always exactly presented as we need . in this case
, we give a detailed proof for the reader s convenience .
+ first , we remind two basic results concerning the fractional integrals .
they are proved in @xcite both using the fubini s theorem .
the first one yields the _ semi - group property _ of the fractional integral operators : [ prop1 ] let @xmath7,{\mathbb{r}}^d)$ ] and @xmath39 , @xmath40 .
then , we have : @xmath41 , \ ; i^{\alpha_1}_- \circ i^{\alpha_2}_- g ( t ) = i^{\alpha_1 + \alpha_2}_- g(t ) \quad \text{and } \quad i^{\alpha_1}_+ \circ i^{\alpha_2}_+ g ( t ) = i^{\alpha_1 + \alpha_2}_+ g(t).\ ] ] the following second property is occasionally called _ fractional integration by parts_. it is very useful for calculus of variations involving fractional derivatives : [ prop2 ] for any @xmath42 , @xmath43,{\mathbb{r}}^d)$ ] and any @xmath44 , we have : @xmath45 let us introduce the following convention : a function defined on @xmath46a , b]$ ] ( resp .
@xmath47 ) is said to be an element of @xmath48,{\mathbb{r}}^d)$ ] if and only if it is continuous on @xmath46a , b]$ ] ( resp .
@xmath47 ) and if it can be continuously extended in @xmath19 ( resp .
@xmath20 ) . from section [ section11 ] , we can give the following examples : * for @xmath7,{\mathbb{r}}^d)$ ] and @xmath44 , we have @xmath16 , @xmath49,{\mathbb{r}}^d)$ ] ; * for @xmath34,{\mathbb{r}}^d)$ ] and @xmath25 , we have @xmath37 , @xmath50,{\mathbb{r}}^d)$ ] .
now , we prove some results of composition between the left fractional operators
. one can easily derive the analogous versions for the right ones .
[ prop3 ] let @xmath7,{\mathbb{r}}^d)$ ] and @xmath25 .
then , @xmath51 is an element of @xmath48,{\mathbb{r}}^d)$ ] and for any @xmath52 $ ] , we have : @xmath53 let us assume additionally that @xmath54,{\mathbb{r}}^d)$ ] . then , @xmath55 is an element of @xmath48,{\mathbb{r}}^d)$ ] and for any @xmath52 $ ] , we have : @xmath56 let us prove the first result .
since @xmath57 , we have for any @xmath58a , b]$ ] : @xmath59 hence , @xmath51 is continuous on @xmath46a , b]$ ] and can be continuously extended by @xmath60 in @xmath19 .
now , let us prove the second result .
it is obvious for @xmath61 .
now , let us consider @xmath62 . since @xmath54,{\mathbb{r}}^d)$ ] , we have @xmath63,{\mathbb{r}}^d)$ ] .
combining the first result and equality , we have for any @xmath58a , b]$ ] : @xmath64 hence @xmath55 is continuous on @xmath46a , b]$ ] and can be continuously extended by @xmath18 in @xmath19 .
the proof is completed . finally , we prove the following fractional cauchy - lipschitz type theorem .
let us precise that a version of this theorem is proved in ( * ? ? ?
* part 3.5.1 , corollary 3.26 , p.205 ) .
it gives the existence and the uniqueness of a global solution of a fractional differential equation .
[ thmfcl ] let @xmath25 , @xmath65 , { \mathbb{r}}^d)$ ] and @xmath66 .
let us assume that @xmath67 satisfies the following lipschitz type condition : @xmath68 , \ ;
\vert f(x_1,t)-f(x_2,t ) \vert \leq k \vert x_1 - x_2 \vert .\ ] ] then , the following fractional cauchy problem : @xmath69 has an unique solution in @xmath48,{\mathbb{r}}^d)$ ] .
the solution @xmath12 is an element of @xmath70 } ( [ a , b],{\mathbb{r}}^d)$ ] ( where @xmath71 $ ] denotes the floor of @xmath72 ) and @xmath12 satisfies : @xmath73 , \ ; g(t ) = a + i^\alpha_- \big ( f(g , t ) \big ) ( t).\ ] ] firstly , let @xmath7,{\mathbb{r}}^d)$ ] . from property
[ prop3 ] , we conclude that @xmath12 is solution of the fractional cauchy problem if and only if @xmath12 satisfies equality . now , let us define : @xmath74{lrcl}\varphi : & { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^d ) & \longrightarrow & { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^d)\\&g & \longmapsto & { \begin{array}[t]{lrcl}\varphi ( g ) : & [ a , b ] & \longrightarrow & { \mathbb{r}}^d\\&t & \longmapsto & a + i^\alpha_- \big ( f(g , t ) \big ) ( t ) .
\end{array } } \end{array}}\ ] ] by induction , we prove that for any @xmath75 and any @xmath42 , @xmath43,{\mathbb{r}}^d)$ ] , we have : @xmath41 , \ ; \vert \varphi^n ( g_1)(t ) - \varphi^n ( g_2)(t ) \vert \leq \dfrac { k^n(t - a)^{n\alpha}}{\gamma ( 1+n\alpha ) } \vert g_1 - g_2 \vert_{\infty}.\ ] ] consequently , we have for any @xmath75 and any @xmath42 , @xmath43,{\mathbb{r}}^d)$ ] : @xmath76 since @xmath77 , there exists @xmath75 such that @xmath78 is a contraction .
consequently , @xmath79 admits an unique fix point in @xmath48,{\mathbb{r}}^d)$ ] .
the proof is completed .
from now and for all the rest of the paper , we consider @xmath25 and @xmath66 . in this section , we first give a brief presentation of the class of fractional optimal control problems interesting us in this paper .
finally , we briefly remind the usual strategy ( using the lagrange multiplier technique ) leading to a necessary condition for the existence of an optimal control , see @xcite and references therein .
+ an optimal control problem concerns the optimization of a quantity depending on parameters given by a controlled system .
precisely , the aim is to find a control optimizing a certain cost .
such a control is called _
optimal control_. many of these studies lead to the use of the lagrange multiplier technique .
for example , we refer to @xcite for controlled systems coded by ordinary differential equations .
+ in this paper , we work in the following framework .
we are interested in systems controlled by the following fractional cauchy problem : @xmath80 where @xmath81 denotes the control .
our aim is to find a control @xmath81 optimizing a quantity of the form : @xmath82 where @xmath83 is solution of .
+ the common strategy is first to write this problem as a problem of optimization under constraint : @xmath84 then , the lagrange multiplier technique consists in the study of the critical points of the following _ augmented functional _ : @xmath85 where @xmath86 is commonly called _
lagrange multiplier_. finally , with a calculus of variations , such a strategy leads to the following result : a necessary condition for @xmath87 to be a solution of is that there exists a function @xmath86 such that the following _ fractional pontryagin s system _ holds : @xmath88 { d^\alpha_+}p = \dfrac{\partial h}{\partial x } ( q , u , p , t ) \\[10pt ] \dfrac{\partial h}{\partial v } ( q , u , p , t ) = 0 \\[10pt ] \big ( q(a),p(b ) \big ) = ( a,0 ) , \end{array } \right.\ ] ] where @xmath89 .
we refer to @xcite and references therein for more details .
+ in this paper , we are going to assume that @xmath90 satisfies a lipschitz type condition , see condition .
this assumption will allow us to write _ directly _ the initial problem as a simpler problem of optimization without constraint of the initial functional which is then _ only dependent of the control _
@xmath81 . finally , a simple calculus of variations leads us to the same result but with a new presentation .
let us remind that the method here developed is widely inspired from the lagrange multiplier technique .
nevertheless , it allows us to give a complete proof only using classical mathematical tools adapted to the fractional case : _ calculus of variations _ , _ gronwall s lemma _ ,
_ cauchy - lipschitz theorem _ and _ stability under perturbations of differential equations_. in this section , we first give rigorously the definitions concerning the class of fractional optimal control problems described in section [ section21 ] : * the elements denoted @xmath91,{\mathbb{r}}^m)$ ] are called _ controls _ ; * let @xmath90 be a @xmath92 function of the form : @xmath74{lrcl}f : & { \mathbb{r}}^d \times { \mathbb{r}}^m \times [ a , b ] & \longrightarrow & { \mathbb{r}}^d\\&(x , v , t ) & \longmapsto & f(x , v , t ) .
\end{array}}\ ] ] it is commonly called the _ constraint function_.
we assume that @xmath90 satisfies the following lipschitz type condition .
there exists @xmath93 such that : @xmath94 , \ ; \vert f(x_1,v , t ) - f(x_2,v , t ) \vert \leq m \vert x_1 - x_2 \vert ; \ ] ] * for any control @xmath81 , let @xmath95 } ( [ a , b],{\mathbb{r}}^d)$ ] denote the unique global solution of the following fractional cauchy problem : @xmath96 @xmath97 is commonly called the _ state variable _ associated to @xmath81 .
its existence and its uniqueness are provided by theorem [ thmfcl ] and condition ; * finally , with this condition on @xmath90 , the fractional optimal control problem described in section [ section21 ] can be rewritten as the simpler problem of optimization of the following _ cost functional _ : @xmath74{lrcl}{\mathcal{l}}^\alpha : & { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^m ) & \longrightarrow & { \mathbb{r}}\\&u & \longmapsto & \di \int_{a}^{b } l ( q^{u,\alpha},u , t ) \ ; dt , \end{array}}\ ] ] where @xmath98 is a _ lagrangian _ , _
i.e. _ a @xmath92 application of the form : @xmath74{lrcl}l : & { \mathbb{r}}^d \times { \mathbb{r}}^m \times [ a , b ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & l(x , v , t ) .
\end{array}}\ ] ] hence , the existence and the uniqueness of @xmath97 for any control @xmath81 allow us to rewrite _ directly _ the initial problem as a simpler problem of optimization without constraint of the cost functional @xmath99 : we do not need to introduce an augmented functional with a lagrange multiplier .
moreover , let us note that @xmath99 is only dependent of the control .
+ a control optimizing @xmath99 is called _
optimal control_. a necessary condition for a control @xmath81 to be optimal is to be a _ critical point _ of @xmath99 , _
i.e. _ to satisfy : @xmath100,{\mathbb{r}}^m ) , \ ; d{\mathcal{l}}^\alpha(u)(\bar{u } ) : = \lim\limits_{\eps \rightarrow 0 } \dfrac{{\mathcal{l}}^\alpha(u+\eps \bar{u})-{\mathcal{l}}^\alpha(u)}{\eps } = 0.\ ] ] in the following , we then focus on the characterization of the critical points of @xmath99 . with an usual calculus of variations , we obtain the following lemma [ lem1 ] giving explicitly the value of the gteaux derivative of @xmath99 : [ lem1 ] let @xmath81 , @xmath101,{\mathbb{r}}^m ) $ ] . then , the following equality holds : @xmath102 where @xmath103 } ( [ a , b ] , { \mathbb{r}}^d ) $ ] is the unique global solution of the following linearised cauchy problem : @xmath104 \bar{q}(a ) = 0 .
\end{array } \right.\ ] ] see appendix [ appad ] .
this last result does not lead to a characterization of the critical points of @xmath99 yet . then , let us introduce the following elements stemming from the lagrange multiplier technique : * let @xmath105 be the following application @xmath74{lrcl}h : & { \mathbb{r}}^d \times { \mathbb{r}}^m \times { \mathbb{r}}^d \times [ a , b ] & \longrightarrow & { \mathbb{r}}\\&(x , v , w , t ) & \longmapsto & l(x , v , t)+w \cdot f(x , v , t ) .
\end{array}}\ ] ] @xmath105 is commonly called the _ hamiltonian _
associated to the lagrangian @xmath98 and the constraint function @xmath90 ; * for any control @xmath81 , let @xmath106 } ( [ a , b ] , { \mathbb{r}}^d ) $ ] denote the unique global solution of the following fractional cauchy problem : @xmath107 p(b ) = 0 .
\end{array } \right.\ ] ] @xmath108 is commonly called the _ adjoint variable _ associated to @xmath81 . its existence and its uniqueness
are provided by the analogous version of theorem [ thmfcl ] for right fractional derivative . since @xmath109
, we can write @xmath110 .
let us note that , for any control @xmath81 , the couple @xmath111 is solution of the following _ fractional hamiltonian system _ : @xmath112 { d^\alpha_+}p = \dfrac{\partial h}{\partial x } ( q , u , p , t ) .
\end{array } \right.\ ] ] finally , the introduction of these last elements allows us to prove the following theorem : [ thmfinal ] let @xmath91,{\mathbb{r}}^m)$ ] .
then , @xmath81 is a critical point of @xmath113 if and only if @xmath114 is solution of the following _ fractional stationary equation _ : @xmath115 let @xmath81 , @xmath101,{\mathbb{r}}^m)$ ] . from lemma [ lem1 ]
, we have : @xmath116 then : @xmath117 from theorem [ thmfcl ] , since @xmath118 is solution of and @xmath108 is solution of , we have : @xmath119 and @xmath120 then , using the fractional integration by parts given in property [ prop2 ] , we obtain : @xmath121 then : @xmath122 the proof is completed by the dubois - raymond s lemma .
let us note that lemma [ lem1 ] is proved in appendix [ appad ] using lemmas [ lemgronf ] , [ lemappa1 ] and [ lemappa2 ] .
these last three lemmas are respectively a fractional gronwall s lemma , a result of stability of order @xmath123 and a result of stability of order @xmath124 for the fractional cauchy problem .
hence , the proof of theorem [ thmfinal ] is only based on classical mathematical tools adapted to the fractional case . + from theorem [ thmfinal ] , we retrieve the following result provided in @xcite and references therein : [ corfinal ] @xmath99 has a critical point in @xmath48,{\mathbb{r}}^m)$ ] if and only if there exists @xmath125 } ( [ a , b],{\mathbb{r}}^d ) \times { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^m ) \times { \mathscr{c}}^{[\alpha ] } ( [ a , b],{\mathbb{r}}^d)$ ] solution of the following _ fractional pontryagin s system _ : @xmath126 { d^\alpha_+}p = \dfrac{\partial h}{\partial x } ( q , u , p , t ) \\[10pt ] \dfrac{\partial h}{\partial v } ( q , u , p , t ) = 0 \\[10pt ] \big ( q(a),p(b ) \big ) = ( a,0 ) .
\end{array } \right.\ ] ] in the affirmative case , @xmath81 is a critical point of @xmath99 and we have @xmath127 . in practice
, we are going to use more corollary [ corfinal ] than theorem [ thmfinal ] , see examples in section [ section4 ] .
let us note that the fractional pontryagin s system is made up of the fractional hamiltonian system , the fractional stationary equation and initial and final conditions . in the case
@xmath61 , the fractional derivatives coincide with the classical one .
consequently , the fractional optimal control problem studied coincides with the classical one .
then , in this case , corollary [ corfinal ] is nothing else but the classical theorem obtained in @xcite : [ corfinalclass
] @xmath128 has a critical point in @xmath48,{\mathbb{r}}^m)$ ] if and only if there exists @xmath129,{\mathbb{r}}^d ) \times { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^m ) \times { \mathscr{c}}^{1 } ( [ a , b],{\mathbb{r}}^d)$ ] solution of the following _ pontryagin s system _ : @xmath130 \dot{p } = - \dfrac{\partial h}{\partial x } ( q , u , p , t ) \\[10pt ] \dfrac{\partial h}{\partial v } ( q , u , p , t ) = 0 \\[10pt ] \big ( q(a),p(b ) \big ) = ( a,0 ) .
\end{array } \right.\ ] ] in the affirmative case , @xmath81 is a critical point of @xmath128 and we have @xmath131 .
let us take the constraint function @xmath132 satisfying . in this case , the fractional optimal control problem studied is : @xmath133 using corollary [ corfinal ] , if this problem has a solution , then there exists a solution @xmath125 } ( [ a , b],{\mathbb{r}}^d ) \times { \mathscr{c}}^0 ( [ a , b],{\mathbb{r}}^m ) \times { \mathscr{c}}^{[\alpha ] } ( [ a , b],{\mathbb{r}}^d)$ ] of the fractional pontryagin s system here given by : @xmath134 { d^\alpha_+}p = \dfrac{\partial l}{\partial x } ( q , u , t ) \\[10pt ] \dfrac{\partial l}{\partial v } ( q , u , t ) + p = 0 \\[10pt ] \big ( q(a),p(b )
\big ) = ( a,0 ) . \end{array } \right.\ ] ] in the case of the existence of an optimal control @xmath81 for problem , we obtain that the state variable associated @xmath97 is solution of the following _ fractional euler - lagrange equation _ : @xmath135 according to the works of agrawal in @xcite , we then obtain that @xmath97 is a critical point of the following _ fractional lagrangian functional _ : @xmath136 we refer to @xcite for more details concerning fractional euler - lagrange equations .
as seen in introduction , there is a large development in fractional calculus and consequently concerning optimal control problems with controlled systems coded by fractional differential equations . in this section
, we present a link between a classical optimal control problem and its fractional version via the _ stanislavsky s formalism _ , see @xcite . only for this section , we assume that @xmath137=[0,\tau ] $ ] with @xmath138 and @xmath139 .
the governing equation for a fluid in a homogeneous porous medium is usually the classical richards equation and it is derived from physical principles , see @xcite .
nevertheless , the fractional richards equation plays an important role in the study of the behaviour of a fluid in a heterogeneous medium , @xcite .
a natural question then arises : _ how physically understand the emergence of the fractional order when the medium changes ?
_ let us sketch an example of answer .
a heterogeneous porous medium presents a lot of stochastic heterogeneities implying complex geometric structure and then inducing stochastic retention zones , @xcite .
hence , we can assume that a change of medium induces a _ modification of the time variable_. therefore , following the work of stanislavsky consisting in the introduction of a stochastic internal time ( a `` slow '' time ) , we remind that the emergence of a fractional derivative can be the effect of a change of time .
let us give more details .
however , we refer to @xcite for a complete study
. + let us consider @xmath140 a stochastic process of probability density function @xmath141 .
@xmath140 represents the stochastic internal time ( the `` slow '' time ) .
stanislavsky assumes that the laplace transform @xmath142 of @xmath141 with respect to its first variable satisfies : @xmath143(s)={\rm e}_{\alpha,1}(-st^\alpha),\ ] ] where @xmath144 is the mittag - leffler function defined in appendix [ appaa ] .
a possible construction of @xmath140 is given in @xcite .
+ stanislavsky studies the dynamical effects of this change of time . precisely , for any @xmath145,{\mathbb{r}}^d)$ ] and any @xmath146 $ ] , he studies the function @xmath147 defined by : @xmath148 where @xmath149 designates the mean value .
the main property is the following result proved in @xcite : [ lemstan ] let @xmath150,{\mathbb{r}}^d)$ ] .
then , we have : @xmath151 , \ ; \mathbb{e } \big ( \dot{g } ( t_t ) \big ) = { { } _ { { \rm c } } d^\alpha_-}\big ( \mathcal{f}_\alpha ( g ) \big ) ( t).\ ] ] hence , a classical derivative is transformed into a fractional one under the introduction of the stochastic internal time of stanislavsky .
now , let us talk about the consequences of the works of stanislavsky on the class of fractional optimal control problems studied in this paper .
we use the notations and definitions of section [ section22 ] .
+ let us assume that @xmath98 and @xmath90 are autonomous , @xmath152 and @xmath153 . then , let us assume that @xmath90 satisfies the following condition : @xmath154,{\mathbb{r}}^d ) \times { \mathscr{c}}^0 ( [ 0,\tau],{\mathbb{r}}^m ) , \ ; \forall t \in [ 0,\tau ] , \\ \mathbb{e}\big ( f \big ( q(t_t),u(t_t ) \big ) \big ) = f \big ( \mathbb{e } \big ( q(t_t ) \big),\mathbb{e}\big ( u(t_t ) \big ) \big ) .\end{gathered}\ ] ] such a condition is satisfied by linear autonomous constraint functions , as the examples in section [ section4 ] .
according to lemma [ lemstan ] , one can prove that for any control @xmath81 , @xmath155 satisfies : @xmath156 precisely , we have for any control @xmath81 , @xmath157 . + now , let us consider the optimal control problem studied in section [ section22 ] in the classical case : @xmath158,{\mathbb{r}}^d ) \longmapsto \di \int_0^\tau l(q^{u,1},u , t ) \ ; dt.\ ] ] let us assume that the controlled system can be prone to changes inducing a modification of the time variable as it is the case with porous media and the richards equation .
we are then interested in the optimization of the following `` slow '' cost functional : @xmath159,{\mathbb{r}}^d ) \longmapsto \di \int_0^\tau l\big ( \mathcal{f}_\alpha ( q^{u,1}),\mathcal{f}_\alpha ( u ) \big ) \ ; dt\ ] ] which is relied to the following _ fractional _ cost functional : @xmath160,{\mathbb{r}}^d ) \longmapsto \di \int_0^\tau l ( q^{u,\alpha},u ) \ ; dt\ ] ] as its restriction to the image of @xmath161 .
once an optimal control of the functional found , one can be interested in its projection on the image of @xmath161 in order to approach an optimal control of the functional .
+ hence , the stanislavsky s formalism is an example relying a classical optimal control problem to its fractional version via the introduction of a stochastic internal time .
in this section , we are going to study some examples of optimal control problems studied in section [ section22 ] both in classical and fractional cases .
classical linear - quadratic examples are often studied in the literature because they are used for tracking problems .
the aim of these problems is to determine a control allowing to approach as much as possible reference trajectories , ( * ? ? ?
* part 1.4 , p.49 ) . in this section ,
we study such an example , ( * ? ? ?
* part 4.4.3 , example 3 , p.53 ) . more generally , a quadratic lagrangian is often natural ( for example in order to minimize distances ) and even if the differential equations are frequently non linear , we are often leaded to study linearised versions .
+ in this section , we take @xmath162 .
we consider an evolution problem of two nondescript populations @xmath163 and @xmath164 interacting during a time interval @xmath137 $ ] .
we control the injection or the discharge of the population @xmath164 in the system : we then control the value of the population @xmath164 in real time .
the interaction between @xmath163 and @xmath164 is governed by the following linear differential equation : @xmath165 for any control @xmath164 and any real @xmath166 , we denote by @xmath167 the unique solution of associated to @xmath164 and satisfying the initial condition @xmath168 .
let us assume that , for an initial condition @xmath169 , we know experimentally a satisfactory control @xmath170 such that the evolution of the population @xmath171 is healthy with respect to a nondescript constraint .
+ the problem is finally the following : _ assume that the initial condition is modified ( _ i.e. _
@xmath172 ) , what is the control @xmath164 minimizing the difference between @xmath164 and @xmath170 plus the difference between @xmath173 and @xmath171 ? _ more precisely , we are looking for a minimizer of the following functional : @xmath174,{\mathbb{r } } ) \longmapsto \dfrac{1}{2 } \di \int_a^b ( \xi - \xi_x)^2 + ( z_{y,\xi } -z_{x,\xi_x})^2 \ ; dt.\ ] ] with a change of variable @xmath175 , @xmath176 , @xmath177 and giving the following quadratic lagrangian and the following linear constraint function : @xmath178{lrcl}l : & { \mathbb{r}}^2 \times [ a , b ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & ( x^2+v^2)/2 \end{array } } \quad \text{and } \quad { \begin{array}[t]{lrcl}f : & { \mathbb{r}}^2 \times [ a , b ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & x+v , \end{array}}\ ] ] the problem is reduced to find an optimal control @xmath81 for the cost functional @xmath128 associated to @xmath98 , @xmath90 and @xmath179 .
see section [ section22 ] for notations and definitions .
+ according to theorem [ corfinalclass ] , we are interested in solving the pontryagin s system here given by : @xmath180 hence , with @xmath181 , it is sufficient to solve the following linear cauchy problem : @xmath182 using matrix exponentials , one can prove that it exists an unique solution @xmath87 to cauchy problem given by : @xmath41 , \ ; \left\lbrace \begin{array}{l } q(t ) = a \big [ \cosh \big ( \sqrt{2}(t - a)\big)+\dfrac{1-r}{\sqrt{2 } } \sinh \big ( \sqrt{2}(t - a)\big ) \big ] \\[10pt ] u(t ) = a \big[\dfrac{1+r}{\sqrt{2 } } \sinh \big ( \sqrt{2}(t - a)\big)- r \cosh \big ( \sqrt{2}(t - a)\big ) \big ] \end{array } \right.\ ] ] where @xmath183 finally , we conclude that @xmath81 is the unique critical point of @xmath128 . hence , if there exists @xmath164 minimizing , then : @xmath41 , \ ;
\xi(t ) = \xi_x ( t ) + u(t).\ ] ] in this section , we consider the previous example of section [ section41 ] in the strict fractional case @xmath184 .
let us consider @xmath185 and the couple @xmath186 given in equation .
we want to express a critical point @xmath81 of the cost functional @xmath99 associated to @xmath98 and @xmath90 .
see section [ section22 ] for notations and definitions .
+ therefore , according to corollary [ corfinal ] , we look for @xmath187 solution of the fractional pontryagin s system here given by : @xmath188 hence , from @xmath189 , it is sufficient to solve the following linear fractional cauchy problem : @xmath190 although the fractional cauchy problem is linear , the presence of the left and the right fractional derivatives is a main drawback to the explicit computation of a solution . more generally , this characteristic implies very big difficulties in order to solve the most of fractional hamiltonian systems . finally , the resolution of is the most of time still an opened problem .
+ nevertheless , in this case , we can hope getting informations on the solutions of in dimension @xmath191 with the help of the fractional noether s theorem proved in section [ section5 ] .
indeed , in this case , the fractional pontryagin s system admits a symmetry and a constant of motion can be obtained .
we refer to section [ section5 ] for more details . in this section ,
we give a solved fractional example in the sense that we give explicitly the unique critical point of the cost functional @xmath99 .
nevertheless , the fractional pontryagin s system is still not completely resolved .
+ let us take @xmath185 , @xmath137=[0,1]$ ] and @xmath192 .
let us consider the following lagrangian @xmath98 and the following linear constraint function @xmath90 : @xmath193{lrcl}l : & { \mathbb{r}}^2 \times [ 0,1 ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & ( v^2/2 ) + \gamma ( 1-t)^\beta x \end{array } } \quad \text{and } \quad { \begin{array}[t]{lrcl}f : & { \mathbb{r}}^2 \times [ 0,1 ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & \lambda x + \mu v , \end{array}}\ ] ] where @xmath194 , @xmath195 , @xmath196 , @xmath197 are parameters in @xmath198 .
we are looking for a critical point @xmath81 of @xmath99 associated to @xmath98 and @xmath90 .
see section [ section22 ] for notations and definitions .
+ according to corollary [ corfinal ] , we have to solve the fractional pontryagin s system here given by : @xmath199 from theorem [ thmfcl ] , the cauchy problem @xmath200 with @xmath201 admits an unique solution for any @xmath202,{\mathbb{r}})$ ] .
then , from @xmath203 , it is sufficient to solve the following fractional cauchy problem : @xmath204 with the change of unknown @xmath205 for any @xmath206 $ ] .
the unique solution of this last fractional cauchy problem is given in ( * ? ? ?
* chap.3 , p.137 ) by : @xmath207 , \ ; p_0 ( t ) = \di \int_0^t ( t - y)^{\alpha-1 } { \rm e}_{\alpha,\alpha } \big ( \lambda ( t - y)^\alpha \big ) \gamma y^\beta \ ; dy,\ ] ] where @xmath208 is the mittag - leffler function defined in appendix [ appaa ] . in order to get a better formulation , we make a change of variable which gives us : @xmath209 , \ ; p_0 ( t ) & = & \gamma \di \int_0^t ( t - y)^{\beta } y^{\alpha-1 } { \rm e}_{\alpha,\alpha } ( \lambda y^\alpha ) dy \\ & = & \gamma \gamma ( \beta + 1 ) i^{\beta + 1}_- \big ( y^{\alpha-1 } { \rm e}_{\alpha,\alpha } ( \lambda y^\alpha ) \big ) ( t ) \\[5pt ] & = & \gamma \gamma ( \beta + 1 ) t^{\alpha+\beta } { \rm e}_{\alpha,\alpha+\beta+1 } ( \lambda t^\alpha ) .\end{aligned}\ ] ] we refer to @xcite for more details concerning the calculations . finally , we obtain the unique critical point @xmath81 of @xmath99 given by : @xmath207 , \ ; u(t ) = - \mu \gamma \gamma ( \beta + 1 ) ( 1-t)^{\alpha+\beta } { \rm e}_{\alpha,\alpha+\beta+1 } \big ( \lambda ( 1-t)^\alpha \big ) .\ ] ] although we obtain the unique critical point @xmath81 of @xmath99 , let us note that this example does not provide a completely solved fractional pontryagin s system : the state variable @xmath97 is still unknown
. however , this example allows to test the quality of numerical schemes giving approximations of critical points of cost functionals @xmath99 . this will be done in a further paper for a variational integrator of .
as we have seen with concrete examples in sections [ section42 ] and [ section43 ] , fractional pontryagin s systems are the most of time not resolvable explicitly .
this is a strong obstruction in order to express explicitly a critical point of a cost functional @xmath99 . in this section ,
we suggest a deviously way in order to get informations on the solutions of and consequently on the critical points of the cost functional @xmath99 associated .
precisely , we are interested in the existence of conservation laws for fractional pontryagin s systems admitting a symmetry ( see definition [ defsym ] ) . + indeed , in 1918 , noether proved the existence of an explicit conservation law for any classical euler - lagrange equation admitting a symmetry .
we refer to @xcite for more details .
adapting her strategy to the fractional case , cresson , torres and frederico proved in 2007 a preliminary result giving a conservation law for any fractional euler - lagrange equation admitting a symmetry , see @xcite . nevertheless , the conservation law obtained was not explicit but only given implicitly via a functional relation . from this first result ,
we have formulated in @xcite a fractional noether s theorem providing an _ explicit _ formulation of this conservation law via a _
transfer formula_. + at the same time , in @xcite , torres and frederico applied a similar strategy for fractional pontryagin s system admitting a symmetry .
nevertheless , this conservation law is also given implicitly via a functional relation . in this section
, we are going to formulate a fractional noether s theorem providing an _ explicit _ formulation of this conservation law via an other _ transfer formula_. + we first review the definition of a one parameter group of diffeomorphisms : let @xmath75 . for any real @xmath210 ,
let @xmath211{lrcl}\phi ( s,\cdot ) : & { \mathbb{r}}^n & \longrightarrow & { \mathbb{r}}^n \end{array}}$ ] be a diffeomorphism .
then , @xmath212 is a one parameter group of diffeomorphisms of @xmath213 if it satisfies : 1 .
@xmath214 ; 2 .
@xmath215 ; 3 .
@xmath216 is of class @xmath92 .
usual examples of one parameter groups of diffeomorphisms are given by translations and rotations .
the action of three one parameter groups of diffeomorphisms on an hamiltonian allows to define the notion of a symmetry for a fractional pontryagin s system : [ defsym ] for @xmath217 , let @xmath218 be a one parameter group of diffeomorphisms of @xmath5 , @xmath219 and @xmath5 respectively .
let @xmath98 be a lagrangian , @xmath90 be a constraint function and @xmath105 be the associated hamiltonian .
@xmath105 is said to be @xmath220-invariant under the action of @xmath221 if it satisfies for any @xmath187 solution of and any @xmath222 : @xmath223 from this notion , torres and frederico proved in @xcite the following result : [ lemtorr ] let @xmath98 be a lagrangian , @xmath90 be a constraint function and @xmath105 be the associated hamiltonian .
let us assume that @xmath105 is @xmath220-invariant under the action of three one parameter groups of diffeomorphisms @xmath221 .
then , the following equality holds for any solution @xmath187 of : @xmath224 then , from this result , torres and frederico defined a notion of _ fractional - conserved quantity_. nevertheless , this result did not provide exactly a constant of motion . in this section
, using a transfer formula , we are going to write the left term of equation as an explicit classical derivative and then we obtain a real constant of motion .
let us provide this transfer formula : [ lemtransfer ] let @xmath42 , @xmath225,{\mathbb{r}}^d)$ ] satisfying the following condition ( c ) : the sequences of functions @xmath226 and + @xmath227 converge uniformly to @xmath18 on @xmath228 $ ] .
then , the following equality holds : @xmath229.\ ] ] in @xcite , under a similar condition than condition ( c ) , we have proved : @xmath230 \ ; \text{and } \ ; - g_1 \cdot { d^\alpha_+}g_2 = \dfrac{d}{dt } \left [ \di \sum_{r=0}^{\infty } g_1^{(r ) } \cdot i^{r+1-\alpha}_+ g _ 2\right].\ ] ] consequently , writing @xmath231 , the proof is completed .
a discussion is provided in @xcite concerning the condition ( c ) : one can prove that this condition is satisfied for any couple of analytic functions for example . then , combining lemmas [ lemtorr ] and [ lemtransfer ] , we prove : [ thmnoetherfrac ] let @xmath98 be a lagrangian , @xmath90 be a constraint function and @xmath105 be the associated hamiltonian .
let us assume that @xmath105 is @xmath220-invariant under the action of three one parameter groups of diffeomorphisms @xmath221 .
let @xmath187 be a solution of and let @xmath12 denote @xmath232 .
if @xmath12 and @xmath86 satisfy condition ( c ) , then the following equality holds : @xmath233 = 0 .\ ] ] this theorem then provides an explicit algorithmic way to compute a constant of motion for any fractional pontryagin s systems admitting a symmetry .
nevertheless , the conservation law is only given as a series of functions : in most cases , it is not easily computable .
however , an arbitrary closed approximation of this quantity can be obtained with a truncation .
+ let us note that the fractional linear - quadratic example developed in section [ section42 ] gives a concrete example of fractional pontryagin s system admitting a symmetry : [ ex4 ] let us consider @xmath234 , the following quadratic lagrangian and the following linear constraint function @xmath74{lrcl}l : & { \mathbb{r}}^2 \times { \mathbb{r}}^2 \times [ 0,1 ] & \longrightarrow & { \mathbb{r}}\\&(x , v , t ) & \longmapsto & ( \vert x \vert^2 + \vert v \vert^2 ) /2 \end{array } } \quad \text{and } \quad { \begin{array}[t]{lrcl}f : & { \mathbb{r}}^2 \times { \mathbb{r}}^2 \times [ 0,1 ] & \longrightarrow & { \mathbb{r}}^2\\&(x , v , t ) & \longmapsto & x+v .
\end{array}}\ ] ] then , let us consider the three one parameter groups of diffeomorphisms given by the following rotations : @xmath235{lrcl}\phi_i : & { \mathbb{r}}\times { \mathbb{r}}^2 & \longrightarrow & { \mathbb{r}}^2\\&(s , x_1,x_2 ) & \longmapsto & \left ( \begin{array}{cc } \cos ( s \theta_i ) & - \sin ( s \theta_i ) \\
\sin ( s \theta_i ) & \cos ( s \theta_i ) \end{array } \right ) \left ( \begin{array}{c } x_1 \\ x_2 \end{array } \right ) , \end{array}}\ ] ] for @xmath217 and where @xmath236 , @xmath237 and @xmath238 . with these parameters
, one can prove that the hamiltonian @xmath105 associated to @xmath98 and @xmath90 is @xmath220-invariant under the action of @xmath221 .
consequently , the fractional pontryagin s system given in ( in dimension @xmath124 ) is not resolvable but admits a symmetry and then admits an explicit conservation law given by the fractional noether s theorem [ thmnoetherfrac ] . as said previously ,
this constant of motion is not explicitly computable .
however , truncating the series , one can provide an approximation of this quantity .
let us recall the definition extracted from @xcite of the mittag - leffler function @xmath239 of parameters @xmath39 , @xmath40 : @xmath240 now , let us give the following lemma proved in @xcite . for the reader s convenience , we recall the proof : [ lemgronf ] let @xmath7,{\mathbb{r}})$ ] and @xmath241 .
let us assume that there exist @xmath242 , @xmath243 such that : @xmath244 , \ ; 0 \leq g(t ) \leq
k_1 i^\alpha_- g(t ) + k_2.\ ] ] then , @xmath12 satisfies : @xmath245,\ ; 0 \leq g(t ) \leq k_2 { \rm e}_{\alpha,1 } \big ( k_1 ( t - a)^\alpha \big).\ ] ] using property [ prop1 ] , assumption implies by induction that for any @xmath75 and any @xmath52 $ ] : @xmath246 since @xmath12 is continuous on @xmath137 $ ] and non - negative , @xmath12 is bounded by @xmath18 and by a constant @xmath247 .
hence , using the stirling asymptotic formula , we have for any @xmath52 $ ] : @xmath248 finally , making @xmath249 tend to @xmath250 in inequality , the proof is completed .
[ lemappa1 ] let @xmath81 , @xmath101,{\mathbb{r}}^m)$ ] .
then , there exists a constant @xmath252 such that : @xmath253 , \ ; \vert q^{u+ \eps \bar{u},\alpha } ( t ) - q^{u,\alpha}(t ) \vert \leq c_1 \vert \eps \vert .\ ] ] from theorem [ thmfcl ] , we have for any @xmath254 and any @xmath52 $ ] : @xmath255 hence , for any @xmath254 and any @xmath52 $ ] , we have : @xmath256 we have for any @xmath254 and any @xmath257 $ ] : @xmath258 from condition and with a taylor s expansion of order @xmath123 with explicit remainder , the following inequality holds for any @xmath254 and any @xmath257 $ ] : @xmath259 where @xmath260 \subset [ -m_1,m_1]^m$ ] with @xmath261 independent of @xmath262 since @xmath81 and @xmath263 are continuous on @xmath137 $ ] .
+ thus , since @xmath264 , @xmath97 and @xmath263 are continuous , there exists @xmath265 such that for any @xmath262 , we have : @xmath266 , \ ; \left\vert \dfrac{\partial f}{\partial v } \big ( q^{u,\alpha}(y),\xi^{\eps}(y),y \big ) \times \bar{u}(y ) \right\vert \leq m_2 \\
\text{and then , } \ ; \forall t \in [ a , b ] , \ ; i^{\alpha}_- \left ( \left\vert \dfrac{\partial f}{\partial v } ( q^{u,\alpha},\xi^{\eps},t ) \times \bar{u } \right\vert \right ) ( t ) \leq m_3,\end{gathered}\ ] ] where @xmath267 is independent of @xmath262 .
+ finally , from inequalities , and , we have : @xmath268 , \ ; \vert q^{u+ \eps \bar{u},\alpha } ( t ) - q^{u,\alpha } ( t ) \vert \leq m i^{\alpha}_- \big ( \vert q^{u+ \eps \bar{u},\alpha } - q^{u,\alpha}\vert \big ) ( t ) + m_3 \vert \eps \vert.\ ] ] using the fractional gronwall s lemma [ lemgronf ] , we conclude that : @xmath269 , \ ; \vert q^{u+ \eps \bar{u},\alpha } ( t ) - q^{u,\alpha } ( t ) \vert & \leq & m_3 \vert \eps \vert { \rm e}_{\alpha,1 } \big ( m ( t - a)^\alpha \big ) \\[10pt ] & \leq & m_3 \vert \eps \vert { \rm e}_{\alpha,1 } \big ( m ( b - a)^\alpha \big).\end{aligned}\ ] ] defining @xmath270 , the proof is completed .
[ lemappa2 ] let @xmath81 , @xmath101,{\mathbb{r}}^m)$ ] .
there exists a constant @xmath271 such that : @xmath253 , \ ; \vert q^{u+ \eps \bar{u},\alpha } ( t ) - q^{u,\alpha}(t ) - \eps \bar{q}(t ) \vert \leq c \eps^2 .\ ] ] where @xmath118 is the unique global solution of the following linearised cauchy s problem : @xmath272 \bar{q}(a ) = 0 .
\end{array } \right.\ ] ] the existence and the uniqueness of @xmath118 are given by theorem [ thmfcl ] and condition .
we proceed in the same manner than for lemma [ lemappa1 ] .
we have for any @xmath254 and any @xmath52 $ ] : @xmath273 with a taylor s expansion s of order @xmath124 with explicit remainder , we have for any @xmath254 and any @xmath257 $ ] : @xmath274 where for any @xmath257 $ ] : * @xmath275 \subset [ \vert q^{u,\alpha } \vert_\infty -c_1,\vert q^{u,\alpha } \vert_\infty + c_1 ] ^d$ ] according to lemma [ lemappa1 ]
. then , there exists @xmath261 such that @xmath276^d$ ] for any @xmath254 and any @xmath257 $ ] ; * @xmath277 \subset [ -m_2,m_2]^m$ ] with @xmath265 independent of @xmath262 and @xmath257 $ ] since @xmath81 and @xmath263 are continuous on @xmath137 $ ] .
since @xmath278 is continuous on the compact @xmath279^d \times [ -m_2,m_2]^m \times [ a , b]$ ] , there exists @xmath280 such that for any @xmath262 and any @xmath257 $ ] , we have : @xmath281 then , from lemma [ lemappa1 ] and since @xmath263 is continuous , there exists @xmath282 such that : @xmath283 , \ ; \left\vert \dfrac{1}{2 } \nabla^2 f \big(\xi^\eps_1(y),\xi^\eps_2(y),y \big ) \big ( q^{u+\eps \bar{u},\alpha } ( y ) - q^{u,\alpha}(y),\eps \bar{u}(y),0)^2 \right\vert \leq m_4 \eps^2 .\end{gathered}\ ] ] consequently , from inequality and since @xmath90 satisfies condition ( implying @xmath284 ) , we have for any @xmath262 and any @xmath257 $ ] : @xmath285 finally , using the previous inequality and inequality , we obtain for any @xmath262 and any @xmath52 $ ] : @xmath286 where @xmath287 is independent of @xmath262 . finally , from the fractional gronwall s lemma [ lemgronf ] , we conclude that : @xmath253 , \ ; \vert q^{u+\eps \bar{u},\alpha}(t ) - q^{u,\alpha}(t ) - \eps
\bar{q}(t)\vert \leq m_5 \eps^2 { \rm e}_{\alpha,1 } \big ( 2m(b - a)^\alpha \big).\ ] ] defining @xmath288 , the proof is completed . in this section ,
we prove lemma [ lem1 ] and consequently , we use the notations and definitions given in section [ section22 ] .
+ let @xmath81 , @xmath101,{\mathbb{r}}^m ) $ ] and @xmath103 } ( [ a , b ] , { \mathbb{r}}^d ) $ ] the unique global solution of . from lemma [ lemappa2 ]
, we have : @xmath289 where @xmath290 .
in particular , since @xmath118 is continuous , there exists @xmath261 such that : @xmath291 , \ ; [ q^{u,\alpha}(y),q^{u+\eps \bar{u},\alpha}(y ) ] \subset [ -m_1,m_1]^d.\ ] ] in the same way , since @xmath81 and @xmath263 are continuous , there exists @xmath265 such that : @xmath291 , \ ; [ u(y),u(y)+\eps \bar{u}(y ) ] \subset [ -m_2,m_2]^m.\ ] ] we have : @xmath292 with a taylor s expansion of order @xmath124 with explicit remainder , we have for any @xmath262 and any @xmath257 $ ] : @xmath293 where for any @xmath294 and any @xmath257 $ ] , @xmath275 \subset [ -m_1,m_1]^d$ ] and @xmath277 \subset [ -m_2,m_2]^m$ ] .
since @xmath98 is of class @xmath92 , we obtain easily that there exists @xmath280 such that for any @xmath262 and any @xmath257 $ ] : @xmath295 consequently , we have for any @xmath296 : @xmath297 hence : @xmath298 the proof is completed .
i. podlubny . ,
volume 198 of _ mathematics in science and engineering_. academic press inc . , san diego , ca , 1999 .
an introduction to fractional derivatives , fractional differential equations , to methods of their solution and some of their applications . | in this paper , we study a class of fractional optimal control problems .
a necessary condition for the existence of an optimal control is provided in @xcite and references therein .
it is commonly given as the existence of a solution of a fractional pontryagin s system and the proof is based on the introduction of a lagrange multiplier .
assuming an additional condition on these problems , we suggest a new presentation of this result with a proof using only classical mathematical tools adapted to the fractional case : _ calculus of variations _ , _ gronwall s lemma _ , _ cauchy - lipschitz theorem _ and _ stability under perturbations of differential equations_. in this paper , we furthermore provide a way in order to transit from a classical optimal control problem to its fractional version via the stanislavsky s formalism .
we also solve a strict fractional example allowing to test numerical schemes .
finally , we state a fractional noether s theorem giving the existence of an explicit constant of motion for fractional pontryagin s systems admitting a symmetry . * * optimal control ; fractional calculus ; noether s theorem . *
* 26a33 ; 49j15 . |
although early - type galaxies make up much of the stellar mass in the local universe ( e.g. fukugita , peebles , & hogan 1998 , coppi & zepf 2000 ) , there is no consensus on when or how they formed .
the possibilities range from monolithic collapse at high redshift to formation via galaxy mergers over a range of redshifts .
many of the integrated properties of elliptical galaxies can be accounted for by both scenarios , making it difficult to determine the formation histories of these galaxies .
for example , one of the salient observational features of the population of early - type galaxies is that their color - magnitude and mg@xmath2 relations have small scatter .
however , both early collapse models ( e.g. kodama & arimoto 1997 ) and hierarchical merging models ( e.g. kauffmann & charlot 1998 ) have been shown to be consistent with the observed correlations . both mergers and collapse models can also be made consistent with current observations of galaxies at high redshift .
the small number of very red galaxies in deep k - selected samples indicates that most ellipticals formed some of their stars at @xmath3 ( treu & stiavelli 1999 , barger et al .
1999 , zepf 1997 ) , with a few exceptions ( e.g. treu et al .
1998 , benitez et al . 1999 ) .
however , this only requires that some of the formation activity of ellipticals take place at lower redshifts , and does not explicitly favor one model or another .
a critical observation that must be addressed by any successful model of galaxy formation is the weak or absent stellar rotation in giant ellipticals .
in contrast to the dynamical insignificance of rotation in elliptical galaxies , the dynamics of spiral galaxies are dominated by rotation , resulting in about an order of magnitude more total angular momentum for spirals compared to the visible components of ellipticals ( e.g. silk & wyse 1993 ) . the large difference in angular momentum between elliptical and spiral galaxies
has long been recognized as one of the most fundamental problems in galaxy formation . moreover
, this difference between the angular momenta of spiral and elliptical galaxies is not expected to be true of their dark matter halos .
all extant theoretical models predict that the spin - up of galactic halos through tidal torques will be similar for all galaxies , so the range of halo angular momenta is smaller than the observed difference between elliptical and spiral galaxies ( e.g. heavens & peacock 1988 , barnes & efstathiou 1987 , warren et al .
1992 , eisenstein & loeb 1995 ) . in terms of the commonly used dimensionless spin parameter @xmath4 , where @xmath5 is the angular momentum e is the energy , and m is the mass , all halos have @xmath6 .
these calculations also find that the angular momentum of a halo is mostly independent of its mass or environment , so any differences between the angular momentum of elliptical and spiral galaxies can not be accounted for solely by their different masses or environments .
the rotation and density of spiral galaxies can be straightforwardly understood as the result of the collapse and spin - up of baryonic material inside a dark matter halo ( e.g. fall & efstathiou 1980 ) .
more specifically , for a protogalaxy with @xmath7 , a collapse factor of roughly ten produces both the observed rotation and the density of spiral disks .
however , there is no such simple explanation for the low angular momentum in the visible parts of elliptical galaxies . as described above
, calculations of the tidal torques on galaxy halos show that all halos have fairly similar specific angular momentum .
moreover , the similar stellar densities of ellipticals and spirals suggest that the gas in each galaxy type collapsed by roughly the same degree during the galaxy formation process . the comparable dark matter contributions within the visible regions of each galaxy type , as well as the extended nature of dark matter halos around both spirals and ellipticals , also imply that the gas collapse factors in both morphological types must have been similar .
since the collapse factors seem to be similar , and the initial values of @xmath8 are expected to be similar , the dramatic difference in the rotation in the visible regions of ellipticals and spirals is a critical constraint on the galaxy formation process .
specifically , spirals and ellipticals must have had different formation processes , with efficient angular momentum transport occurring during the formation of elliptical galaxies .
it has long been realized that galaxy mergers might provide a mechanism for transporting angular momentum out of the visible regions of elliptical galaxies ( e.g. fall 1979 , toomre 1977 , barnes & hernquist 1992 ) .
the hypothesis that elliptical galaxies form from major mergers while spiral galaxies have avoided a major merger in the recent past is therefore an attractive one .
however , as described above , it is difficult to demonstrate from observational data that elliptical galaxies today formed from mergers in the past .
some of the strongest evidence that elliptical galaxies form episodically comes from the study of the colors of globular clusters around nearby ellipticals .
these studies show that elliptical galaxies often have globular cluster systems with bimodal metallicity distributions ( ashman & zepf 1998 and references therein , see also kundu 1999 and gebhardt & kissler - patig 1999 ) .
the distinct globular cluster populations observed in many ellipticals suggest episodic formation and are inconsistent with simple monolithic collapse models .
in contrast , bimodal globular cluster systems were predicted by ashman & zepf ( 1992 ) for elliptical galaxies that formed from mergers of gas - rich spirals .
specifically , ashman & zepf ( 1992 ) predicted that elliptical galaxies would have a population of metal - poor globulars from the halos of the progenitor spirals , and a population of metal - rich clusters formed in the merger that made the elliptical .
subsequent observations have produced a large body of observational evidence indicating that globular clusters can form in gas - rich mergers ( see reviews by schweizer 1998 , ashman & zepf 1998 , and references therein ) .
thus both fundamental predictions of the merger model have been supported by subsequent observations - globular clusters are observed to form in mergers and the proposed product of these mergers , elliptical galaxies , show evidence for multiple populations of globular clusters .
the discovery of young globular clusters in mergers and of bimodality in the globular cluster systems of ellipticals were successfully predicted by the merger models .
however , for many elliptical galaxies the specific number and inferred metallicity of the blue , metal - poor globular clusters is not in detailed agreement with the predictions of the simple merger model . in the simplest merger picture , these metal - poor clusters come from the halo populations of the progenitor spirals , and therefore the metal - poor population in ellipticals should have the same metallicity and specific frequency ( number per stellar luminosity , or preferably , number per stellar mass ) . however , some elliptical galaxies such as m87 appear to have more metal - poor clusters than can be accounted for by combining the halo populations of spirals like the galaxy and m31 ( e.g. forbes , brodie , & grillmair 1997 , lee , kim , & geisler 1998 ) while other ellipticals such as ngc 3923 ( zepf , ashman , & geisler 1995 ) and ic 4051 ( woodworth & harris 2000 ) may either not have a substantial metal - poor globular cluster population or have one which is more metal - rich than the halo populations of the galaxy and m31 .
zepf et al .
( 1995 ) and ashman & zepf ( 1998 ) have noted that , in the context of the merger models , these observations suggest variations in the properties of the globular cluster systems of the progenitor spirals , as well as the possibility of significant accretion of low - metallicity globular clusters and associated dwarf galaxies . following the discovery of the bimodality in the globular cluster systems of elliptical galaxies , models were constructed to try to account for bimodality without major mergers .
one example is the proposal that elliptical galaxies simply form in two phases , an initial spatially extended phase of metal - poor globular globular cluster formation followed by the later formation of metal - rich clusters after dissipation and enrichment , with some internal mechanism that turns off globular cluster formation between the two globular cluster formation phases ( forbes et al .
1997 , harris , harris , & mclaughlin 1998 ) .
alternatively , it has also been proposed that the central component of elliptical galaxies undergoes standard dissipative collapse , and the metal - poor globular cluster population is accreted later ( cote , marzke , & west 1999 ) .
the key physical element of all models is that angular momentum conservation requires the dissipative collapse that leads to the formation of the metal - rich globular clusters to also result in significant rotation in the metal - rich globular cluster population .
the only way to avoid this conclusion is to have a mechanism to efficiently transport angular momentum outwards .
merger models have been shown to provide such a mechanism , while smooth dissipative collapse models do not naturally transport angular momentum efficiently .
therefore , the kinematics of the globular cluster systems provides a way to test if the physics of the formation of elliptical galaxies and their globular cluster systems is consistent with smooth dissipative collapse or if angular momentum is required to be transported outwards .
moreover , if angular momentum transport is indicated by observations in the inner regions of an elliptical , it might be possible to observe the angular momentum transferred outwards in the kinematics at very large radii , as possibly found for the m87 globular cluster system by kissler - patig & gebhardt ( 1998 ) based on the data of cohen & ryzhov ( 1997 ) .
measurements of the radial velocities of globular clusters also provide information about the mass distribution of the galaxy and the orbits of the clusters .
globular clusters are particularly useful for studying the dynamics of the outer halos of elliptical galaxies because they can be observed out to much larger radii than it is possible to obtain spectroscopy of the integrated light .
a very large number of velocities are required to independently determine the mass distribution and the orbits of the tracer particles in a completely non - parametric way ( e.g. merritt & tremblay 1994 ) . however ,
if the mass distribution inferred from x - ray observations and the assumption of hydrostatic equilibrium in the hot gas is adopted , the orbits of the globular clusters can be constrained .
conversely , if assumptions are made about the cluster orbits ( e.g. that they are isotropic ) , then the mass distribution can be estimated . in practice , a sensible approach is to check for consistency of the mass distribution determined via the x - ray observations of the hot gas with dynamical measurements given simplifying assumptions about the orbits , as each technique has its own systematic concerns which are mitigated if the independent approaches agree . in this paper
, we present the results of a spectroscopic survey of globular clusters around the virgo elliptical ngc 4472 .
we focus on the analysis of the kinematics of the globular cluster system and its implications .
a discussion of the ages and metallicities of the ngc 4472 globular clusters is given in beasley et al .
the observations and data reduction are described in @xmath9 .
the results of our analysis of these data are presented in @xmath10 . in @xmath11
we compare the velocity dispersions of the different globular cluster populations , in @xmath12 we analyze the rotation of the cluster populations , in @xmath13 we study the radial trends in the velocity dispersion and rotation , and in @xmath14 we examine the mass distribution of ngc 4472 inferred from our data .
the implications of these results are discussed in @xmath15 , and our conclusions given in @xmath16 .
we obtained spectra of globular cluster candidates around ngc 4472 using the multi - object spectrograph ( hereafter mos , crampton et al .
1992 , lefevre et al .
1994 ) on the canada - france- hawaii telescope over parts of three nights in may 1998 .
mos is an effective instrument for this study because its relatively large field of view allows for efficient multiplexing of the ngc 4472 globular cluster system , which extends out to many arcminutes from the center of the galaxy ( cf . rhode & zepf 2000 ) .
the candidate globular clusters were selected from the photometric study of the ngc 4472 globular cluster system by geisler et al .
we obtained spectra through four slitmasks , each of which contained slitlets for approximately 45 globular cluster candidates .
we preferentially targeted cluster candidates with magnitudes between @xmath17 , but included fainter clusters when useful for maximizing the number of targets in each mask .
about 2/3 of the slitlets yielded useful spectra , with the remainder having either low signal - to - noise or problematic sky subtraction .
the spectra were obtained with a 600 l / mm grism , giving a resolution of 2.2 per pixel , and a spectral range of @xmath18 .
the data were processed with flat fields taken at the beginning and end of the night , and wavelength calibrated using frequent hg arcs .
after extraction of the object spectra and sky subtraction , the radial velocities of the candidate globular clusters were determined by cross - correlation ( e.g. tonry & davis 1979 ) with six template stars of known velocity and spectral types ranging from f8v to k0iii .
the ` r ' value of the cross - correlation was required to be greater than 2.5 for a velocity measurement to be included in our final sample .
the final velocity for each object was determined by a weighted average of the velocities returned by the cross - correlation with each of the six template stars , where the weighting was given by the cross - correlation peak height of each template .
the data reduction procedures are described more completely in beasley et al .
( 2000 ) .
the formal errors returned by the cross - correlation task for each globular cluster are given in table 1 below .
these are typically 50 - 100 @xmath19 .
a check on these errors can be made by comparing the velocities for the 13 clusters for which we have velocities from both this analysis and our earlier work ( sharples et al .
1998 , hereafter s98 ) .
the result is a mean difference of @xmath20 @xmath19 with an uncertainty in any individual measurement of 78 @xmath19 .
much of the dispersion is driven by a large offset for one cluster , so the typical uncertainty may be smaller .
regardless , these uncertainties are much smaller than the velocity dispersion of ngc 4472 , so the measurement errors in the velocities of individual clusters do not have a significant effect on the results .
in table 1 , we list the positions , magnitudes , colors , and radial velocities for the 144 clusters around ngc 4472 for which we have spectra .
this table includes 100 new globular cluster velocities obtained in the observations described above , and 57 previously published by us ( s98 ) , which also included some from earlier work by mould et al .
for the 13 clusters with data from both datasets , the tabulated velocity is from the observation with the strongest cross - correlation .
these data form the basis for the analysis presented in the remainder of this paper . for completeness , we also list in table 2 the 20 objects for which we obtained spectra and found them to be foreground stars or background galaxies .
our @xmath21 success rate in spectroscopically confirming our targets as globular clusters is higher than earlier work .
this high success rate is likely due to selection of targets from high quality photometry , and bodes well for future spectroscopic programs .
in addition to analyzing the populations as a whole , we also investigate the kinematic properties of the metal - rich and metal - poor populations of globular clusters in ngc 4472 .
these populations were first identified through the discovery of bimodality in the color distribution of the ngc 4472 globular cluster system ( zepf & ashman 1993 and many subsequent papers ) . that these color differences are primarily due to metallicity is now confirmed by spectroscopy ( beasley et al .
further evidence for metallicity driving the color distribution is provided by analysis of the luminosity functions of the red and blue globular clusters ( e.g. puzia et al .
1999 , lee & kim 2000 ) . in order to divide our sample into populations of metal - rich and metal - poor clusters
, we apply the kmm mixture - modeling algorithim to the full dataset of geisler et al .
this objective analysis divides the sample at a color of @xmath22 .
clusters bluer than this are more likely to belong to the metal - poor population while those redder than this value are more likely to belong to the metal - rich population .
this is the same analysis and color cut used in s98 .
this assignment represents the best statistical estimate of whether a globular cluster belongs to the metal - rich or metal - poor population .
however not every cluster will necessarily be assigned to the population to which it truly belongs because of possible overlap in the color distributions of the different cluster populations as well as photometric errors .
this will tend to make it more difficult to detect physical differences between the metal - rich and metal - poor cluster populations , particularly for clusters near the boundary .
the full dataset is plotted in figure 1 .
this figure shows the distribution on the sky of the globular clusters for which we have velocities , and also includes a rough indication of the velocities of these objects , as well as their identification as a member of the metal - rich or metal - poor cluster population .
the bulk of this paper is aimed at a quantitative analysis of the velocities and positions of the clusters presented in table 1 and figure 1 .
one of the goals of this study was to compare the kinematics of the metal - rich and metal - poor globular cluster populations of ngc 4472 . in our earlier work with radial velocities for 57 globulars we found tentative evidence ( @xmath23 confidence level ) for a higher velocity dispersion in the metal - poor population compared to the metal - rich population ( s98 ) .
a difference in velocity dispersions between the metal - rich and metal - poor globular cluster populations is critical evidence that these populations have real physical differences , since the division between the metal - rich and metal - poor cluster populations was made by an objective analysis of the colors alone .
figure 2 shows the comparison of the velocity distribution of the red , metal - rich clusters , and the blue , metal - poor clusters around ngc 4472 .
this plot demonstrates that the new , larger dataset strongly confirms our original discovery that the metal - poor cluster population has a significantly higher velocity dispersion than the metal - rich cluster population .
specifically , a f - test indicates that the two populations have a different dispersion at @xmath24 confidence level .
the values of the velocity dispersion in each population are @xmath25 @xmath19 for the metal - rich clusters , and @xmath26 @xmath19 for the metal - poor clusters where the uncertainties have been determined from bootstrapping .
these are uncorrected for any rotation , but as we show below , the rotation is small and does not affect the dispersions significantly .
moreover , regardless of its origin , the difference in velocity dispersions provides dynamical evidence for the original distinction between the metal - rich and metal - poor cluster populations in ngc 4472 .
it is also interesting to note that the difference in dispersions between the metal - rich and metal - poor globular cluster populations in ngc 4472 does not necessarily imply that the populations have different orbital properties .
this is because there is observational evidence that the metal - rich cluster system in ngc 4472 is more spatially concentrated than the metal - poor cluster system ( e.g. lee et al .
this difference between the density profiles of the metal - poor and metal - rich globular cluster populations makes it possible for these populations to have similar orbital anisotropy but different velocity dispersions , and still follow the same gravitational potential . in detail
, the velocity dispersion difference we observe is modestly larger ( @xmath27 ) than that expected from the difference in the spatial distributions for the metal - poor and metal - rich clusters found by lee et al . ( 1998 ) . a second major goal of our spectroscopic study of the ngc 4472 globular cluster system was to determine the rotation of the system as a whole , as well as that of the metal - rich and metal - poor populations . as discussed in the introduction , determining the rotation of the cluster populations in ngc 4472 is invaluable for addressing questions about the formation history of the galaxy and the relationship of the different cluster populations to one another
. a straightforward way to estimate the rotation in the globular cluster system of ngc 4472 is to do a non - linear least - squares fit to the equation @xmath28 this determines the best fitting flat - rotation curve with the position angle of the line of nodes ( @xmath29 ) and the rotation velocity ( @xmath30 ) free parameters .
the results from this analysis are given in table 3 .
we find little or no rotation in the metal - rich population and modest evidence for rotation in the full sample and in the metal - poor population .
the line of nodes of the best fitting rotation solution appears to be similar to the position angle of the major axis of the galaxy ( @xmath31 ) .
the significance of the rotation detections can be tested by performing monte carlo simulations of the data .
these simulations retain the position angles of the objects but randomize the velocities ( cf .
roughly half of such simulations give a rotation signature about some axis equal to that observed in the metal - poor and total cluster samples .
however , the rotation in these simulations is not preferentially aligned with the major axis of the galaxy , as the real data is .
we can account for this by considering random samples with the position angle of the line of nodes fixed to be that observed for the isophotes the galaxy . in this case , only @xmath32 of the random simulations give a rotation signature as large as observed for the total and metal - poor cluster samples .
the metal - rich sample still does not show evidence for rotation even when the position angle is fixed .
in addition to the best fitting rotation solution , we determine an upper limit on the rotation by creating monte carlo samples of clusters with the same number of objects and the same velocity dispersion as the given populations , but with a given rotation velocity .
we define the @xmath1 confidence upper limit as the rotation velocity for which only 1/100 of such random samples gives a rotation signature as small as that observed .
this procedure produces strict upper limits on the rotation in the globular cluster system of ngc 4472 , especially the metal - rich population .
specifically , the @xmath1 upper limit on rotation for the full cluster sample is 120 @xmath19 , for the metal - poor cluster population the @xmath1 upper limit on rotation is 200 @xmath19 , and for the metal - rich population the rotation is less than 75 @xmath19 at @xmath1 confidence . these limits on rotation can be used to set upper limits on the ratio of @xmath33 , which is a common diagnostic of the importance of rotation . for the metal - rich population , the combination of the @xmath1 upper limit of 75 @xmath19 on the rotation with the velocity dispersion of 221 @xmath19 found earlier gives an upper limit of @xmath0 at 99% confidence .
this is remarkably low compared to the expectation of simple collapse models .
it is also dramatically lower than the metal - rich thick disk / bulge clusters of the milky way , which have @xmath34 .
although the value for ngc 4472 is a projected one , it is very unlikely that rotation is dynamically significant .
one way to see this is to compare the observed @xmath33 of the metal - rich cluster system in ngc 4472 to that expected for an oblate rotator ( e.g. binney & tremaine 1987 ) . for the observed ellipticity of about 0.2 for the metal - rich population ( lee et al .
1998 , rhode & zepf 2000 ) , an oblate rotator would have a @xmath33 of about 0.4 , mostly independent of projection .
this prediction for an oblate rotator is larger than the @xmath1 upper limit , strongly suggesting that the metal - rich globular cluster system of ngc 4472 is not supported by rotation .
a similar approach is to note that the inclination would have to be @xmath35 for @xmath36 to be consistent with @xmath37 .
moreover , such an inclination would require an intrinsic flattening of greater than e9 for the metal - rich system to match the observed e2 in projection .
the very small rotation in the metal - rich globular cluster system of ngc 4472 is problematic for models in which central spheroidal systems all form similarly , with the only difference being the mass of the central bulge / elliptical component .
instead , it requires efficient angular momentum transport for the metal - rich globular cluster population of ngc 4472 that did not occur during the formation of the metal - rich globular cluster population of the milky way .
such a difference is natural in models in which elliptical galaxies like ngc 4472 formed in major mergers , and spiral galaxies formed in such a way as to avoid disruptive events that lead to effective , large scale transport of angular momentum .
this is discussed in more detail in the final section of the paper .
in addition to the total velocity dispersions and rotation velocities derived in the two previous sections , radial trends in these quantities are also interesting .
for example , if the velocity dispersion can be traced out to large radii , constraints can be placed on the mass distribution of the halo , although some simplifying assumptions about the orbital anisotropy of the globular cluster population are required for modest sample sizes .
conversely , a mass distribution can be adopted based on other data such as x - ray observations , and then the observed dispersions can be used to constrain the orbital anisotropy at different radii .
similarly , changes in the rotation velocity as a function of radius can be used to trace the distribution of angular momentum within a galaxy . in order to determine the rotational velocity and velocity dispersion as a function of radius , it is necessary to average in some way the discrete radial velocities of individual clusters .
binning is not ideal because of its well - known sensitivity to the choice of bin centers and sizes .
the problem of estimating smooth functions from individual datapoints is a well - developed statistical field ( e.g. simonoff 1996 and references therein ) .
the application of these techniques to the problem of estimating the velocities and dispersion of the globular cluster systems as a function of radius is straightforward because our goal is to determine the overall trends in these systems and sharp changes are not expected . in this case , a gaussian smoothing kernel provides reliable results with the advantage of simplicity in application and interpretation .
a key question for any smoothing procedure is to determine the width of the smoothing kernel .
a common approach , called cross - validation , is to determine the kernel size that minimizes the summed total of the variances determined at each point , when that point is excluded from the calculation .
however , cross - validation gives little guidance when the sample in question is consistent with being a constant , because it is statistically preferable in this case to use all of the data to determine the value of the function in question at each radius .
this is exactly the case for the velocities of the globular cluster system of ngc 4472 , particularly the full sample and the metal - poor subsample .
therefore , while cross - validation clearly identifies the statistical consistency of the data with a flat rotation curve , it does not yield much help for determining the appropriate smoothing kernel for determining any dependence of the velocity dispersion on radius . because our goal is to determine the overall trend of velocity dispersion with radius
, we adopt a fairly broad smoothing , using a gaussian kernel with @xmath38 .
this kernel width is roughly equivalent to two or three independent radial `` bins . ''
smaller smoothing kernels produce the same overall trends , with additional bumps and wiggles .
a sign that these may be undersmoothed is that the velocity dispersion profiles from smaller kernels yield unphysical results when used in the jeans equation ( see section 3.4 ) .
figure 3 shows the results of our analysis of the radial variation of the rotation and velocity dispersion . because the line of nodes of rotation is consistent with the position angle of the galaxy ( section 3.2 ) ,
the data have been flipped about the minor axis to take full advantage of the available positions and velocities .
the upper and lower lines are @xmath39 bootstrap limits . in order to build the bootstrap samples in a way that is consistent with
the smoothing kernel used to determine the best fit , the probability of including an object in the bootstrap sample is proportional to the value of the kernel of that object at a given radius . because of the smoothing required to get rotation and velocity dispersions from individual radial velocities , the individual points along the curves are of course not independent of one another .
overall , the globular cluster population in ngc 4472 appears to have a slowly declining velocity dispersion with radius .
this is seen in the full sample , and the individual metal - rich and metal - poor samples are consistent with the same trend .
a sharper drop in the dispersion at our largest radial distances ( @xmath40 ) is also consistent with the data , as shown by the lower limit of the dispersion in figure 3 and also by analyses using smaller smoothing kernels .
further data at large radii is required to test this possibility . the slightly declining velocity dispersion in ngc 4472 contrasts to the slightly rising dispersion found for m87 over the same radial range ( e.g. cohen 2000 and references therein ) .
this difference may suggest that the rising dispersion in m87 is due to transition from the galaxy to the virgo cluster itself , as m87 is located at the center of the virgo cluster , while ngc 4472 is well away from the cluster center .
figure 3 also shows the rotational velocity of the total sample and the metal - poor population is consistent with a constant value at all radii , while the metal - rich sample shows slight evidence for increasing rotation outwards .
this can also be seen by performing a straight regression of velocity with major axis distance , which gives absolutely no trend for the total and metal - poor samples , and a @xmath39 result for increasing rotation with larger radii for the metal - rich sample . at all radii for which we have data
the rotation is significantly smaller than the dispersion .
this applies to the globular cluster sample as a whole , as well as to the metal - rich and metal - poor cluster populations .
the velocity dispersion profile determined in the previous section can be used to derive the mass distribution of ngc 4472 through the application of the jeans equation ( e.g. binney & tremaine 1987 ) .
the mass determined in this way depends on the anisotropy of the cluster orbits , as well as on the directly observed projected velocity dispersion and projected density profiles .
the simplest approach is to adopt isotropy for the cluster orbits , and to see how this compares to other estimates for the mass of ngc 4472 within the same radial range . in detail , the three dimensional luminosity density and velocity dispersion profiles are determined from the observed projected quantities by numerically integrating the abel equations .
spherical symmetry is assumed in this approach , which is not strictly true for ngc 4472 , but it not likely to be too far wrong given that the observed ellipticity of the globular cluster systems is @xmath41 ( e.g. lee et al .
1998 ) .
figure 4 shows the mass of ngc 4472 as a function of radius .
the lines are masses inferred from application of the jeans equation to the radial velocities of our globular cluster sample .
the central solid line is the best fit to the 144 radial velocities discussed in this paper .
the dotted lines are the @xmath39 lower and upper limits .
these uncertainties in the mass are based on the bootstrapped uncertainties in the radial velocity dispersion profile given in the previous section .
we do not include uncertainties in the density profile of the globular cluster system because they are much smaller than those for the velocity dispersion .
specifically , the surface density profile is taken from the fit of a de vaucouleurs law to the wide field ccd imaging study of rhode & zepf ( 2000 ) , which extends out to @xmath42 , well beyond our last data point .
different techniques for deriving the density profile from these data or comparisons to previous work produce variations that are much smaller than the uncertainty in the velocity dispersion .
the radial profile of the mass distribution of ngc 4472 derived from the radial velocities of the globular clusters can be compared to the mass inferred from x - ray observations of the hot gas around the galaxy .
the solid squares on figure 4 are the masses inferred from rosat observations of the hot gas around ngc 4472 ( irwin & sarazin 1996 ) .
the open squares represent points for which the assumption of hydrostatic equilibrium on which the x - ray masses are based may be uncertain because the x - ray isophotes are irregular at these radii .
the overall agreement between the mass inferred from the globular cluster velocities given isotropic orbits and the mass inferred from the analysis of x - ray observations of the hot gas assuming hydrostatic equilibrium is good .
this agreement suggests that both assumptions are probably roughly correct , and that the masses so derived are reasonably accurate . the conclusion that then follows is that the mass - to - light ratio of ngc 4472 is at least five times greater at radii of @xmath43 kpc than it is at radii of several kpc , indicative of a substantial dark halo around this galaxy
this is some of the strongest dynamical evidence to date for massive halos around elliptical galaxies . at small radii
, there is a hint that the mass inferred from the globular cluster dynamics may be slightly higher than that inferred from the hot gas .
this could either be the result of statistical uncertainties in either observation , a slight radial anisotropy in the globular cluster orbits at smaller radii , or a slight underestimation of the temperatures in the hot gas at small radii .
in any case the overall agreement between the two approaches is good , and more data will be required to extend this approach to larger radii and to test for any differences at smaller radii .
one of the primary goals of the study of the kinematics of the globular cluster systems of nearby ellipticals is to constrain the formation history of these galaxies .
the bimodal color distributions of the globular cluster systems of many ellipticals already indicate an episodic formation history for these galaxies , roughly along the lines predicted by merger models .
the kinematic data presented here confirm and extend the identification of two populations of globular clusters in ngc 4472 by showing that the population of globular clusters identified as metal - poor through their colors has a higher velocity dispersion than the population identified as metal - rich .
in addition to confirming the existence of a metal - rich and metal - poor globular cluster populations in ngc 4472 , our data help constrain the physical nature of the episodic formation history of this galaxy .
perhaps the most critical observational result is that there is little or no rotation in the metal - rich globular cluster system . despite the absence of rotation , the metal - rich system clearly underwent significant dissipation and collapse since it is more spatially concentrated than the metal - poor cluster population and much more so than the dark matter halo . given conservation of angular momentum , the metal - rich population would be naturally expected to spin up as it collapses .
in fact , significant rotation is observed observed in the metal - rich population of disk galaxies like the milky way and m31 .
thus the minimal rotation in the metal - rich globular cluster population of ngc 4472 requires angular momentum transport during the formation of this elliptical galaxy , and distinguishes the metal - rich globular cluster population in this elliptical from metal - rich cluster populations in the galaxy and m31 .
merger models have long been put forward as a way to transport angular momentum outwards ( e.g. toomre & toomre 1972 , barnes & hernquist 1992 and references therein ) .
a generally consistent picture can be constructed in which elliptical galaxies like ngc 4472 form in major mergers which create the metal - rich globular cluster population and transfer angular momentum outwards , while disk galaxies like the milky way and m31 have only had more minor mergers , which may lead to modest amounts of globular cluster formation but which are not as efficient at angular momentum transfer .
an additional implication of our data is that ngc 4472 has a dark halo that extends out to the limit of our dataset , about @xmath44 kpc .
however , this limit is only set by the field of view of the spectrographs we have used .
a wide - field ccd imaging study of the globular cluster system of ngc 4472 shows that it extends out to at least @xmath45 kpc ( rhode & zepf 2000 ) .
thus globular clusters offer a dynamical probe of the halo at very large radii which can be reached effectively with spectrographs with sufficient field of view .
this is particularly valuable for ngc 4472 , since its x - ray isophotes are irregular at radii greater than about @xmath46 kpc , indicating the hot gas may not be in hydrostatic equilibrium . without hydrostatic equilibrium ,
mass estimates from x - ray gas are problematic , leaving dynamical techniques the best hope for extending the understanding of the halo of ngc 4472 to very large radii .
in this paper , we present radial velocities for 144 globular clusters around ngc 4472 .
based on this dataset , we analyze the kinematics of the globular cluster system as a function radius and metallicity .
we find \1 .
the metal - poor globular clusters have a significantly higher velocity dispersion than the metal - rich clusters .
this kinematic difference confirms the identification of these two cluster populations within ngc 4472 , as they were originally selected based on their colors alone .
@xmath33 is much less than one over the radial range covered by our data ( @xmath47 kpc ) .
this is true for the globular cluster system as a whole and for the individual metal - rich and metal - poor cluster systems .
the upper limit on rotation is particularly small for the metal - rich globular cluster system , for which @xmath0 at the @xmath1 confidence level .
this absence of rotation requires either significant angular momentum transport outwards during the formation of ngc 4472 or a much lower initial spin of the halo of ngc 4472 than given by standard tidal torque calculations .
the requirement for significant angular momentum transport is particularly strong for the metal - rich population , which is concentrated towards the center of the galaxy and has a high metallicity .
these properties point to significant dissipation and collapse during the formation of the metal - rich cluster population , yet it has little or no rotation .
the absence of significant rotation in the metal - rich globular cluster population of ngc 4472 strongly distinguishes this metal - rich cluster population from the metal - rich cluster populations found in disk galaxies like the milky way .
thus the angular momentum transport required for the metal - rich cluster population in ngc 4472 appears to be associated with the physical differences between elliptical and spiral galaxies .
this may be explained if ellipticals form from major mergers , while spirals are those galaxies that have not had major mergers since the formation of their disk .
the velocity dispersion profile of the ngc 4472 slowly declines with radius , with a hint that the decline may steepen at large radius .
the mass derived from this dispersion profile and the assumption of isotropic orbits is consistent with that derived from x - ray observations of the hot gas .
this suggests that neither technique has dramatic systematic problems , and strongly indicates the presence of a dark matter halo around ngc 4472 .
we thank dave carter for his contributions to the early stages of the research project , david buote for making available his analysis of the rosat observations of ngc 4472 , and the referee for a careful reading of the paper .
we acknowledge useful conversations with ken freeman , melinda weil , and dean mclaughlin .
sez acknowledges support from the hellman family foundation , mab acknowledges the support of a pparc studentship and the use of starlink facilities at the university of durham , dg acknowledges support for this project from conicyt through fondecyt grant 1000319 and by the universidad de concepcin through research grant no .
99.011.025 - 1.0 , and and dah acknowledges support from an operating grant awarded by the natural sciences and engineering research council ( nserc ) of canada .
170 & 12 27 05.19 & 8 09 52.6 & 21.19 & 1.73 & 1243 @xmath48 282 & 12 27 32.88 & 8 10 24.1 & 20.38 & 1.65 & 751 @xmath49 463 & 12 27 17.87 & 8 10 55.5 & 19.93 & 1.59 & 342 @xmath50 637 & 12 27 10.03 & 8 11 24.1 & 19.95 & 1.54 & 814 @xmath51 647 & 12 27 04.56 & 8 11 25.4 & 21.15 & 1.34 & 1101 @xmath49 676 & 12 27 19.80 & 8 11 30.5 & 20.72 & 1.46 & 1304 @xmath52 714 & 12 27 34.88 & 8 11 35.4 & 20.32 & 1.55 & 1061 @xmath49 744 & 12 27 20.91 & 8 11 40.2 & 19.73 & 1.38 & 814 @xmath53 876 & 12 27 28.14 & 8 11 58.3 & 19.63 & 1.46 & 1487 @xmath54 995 & 12 26 46.05 & 8 12 10.0 & 20.90 & 1.54 & 828 @xmath55 1047 & 12 27 00.94 & 8 12 13.8 & 21.14 & 1.89 & 1083 @xmath56 1087 & 12 27 17.20 & 8 12 17.5 & 19.64 & 1.44 & 1070 @xmath53 1110 & 12 27 12.04 & 8 12 20.3 & 19.88 & 1.37 & 1626 @xmath54 1207 & 12 27 26.09 & 8 12 29.3 & 21.34 & 1.25 & 739 @xmath48 1234 & 12 26 46.82 & 8 12 32.2 & 20.81 & 1.47 & 1059 @xmath57 1255 & 12 27 20.37 & 8 12 34.4 & 19.71 & 1.35 & 816 @xmath58 1315 & 12 27 11.39 & 8 12 40.9 & 20.68 & 1.42 & 1370 @xmath59 1423 & 12 27 02.74 & 8 12 51.3 & 20.94 & 1.57 & 1641 @xmath53 1448 & 12 26 56.90 & 8 12 54.2 & 20.79 & 1.34 & 1353 @xmath60 1475 & 12 27 07.89 & 8 12 57.0 & 21.15 & 1.46 & 1018 @xmath61 1518 & 12 27 07.89 & 8 13 00.6 & 19.25 & 1.85 & 1050 @xmath62 1570 & 12 27 06.52 & 8 13 05.6 & 20.98 & 1.58 & 1034 @xmath63 1587 & 12 27 26.80 & 8 13 07.3 & 21.16 & 1.12 & 471 @xmath64 1650 & 12 27 23.30 & 8 13 13.6 & 20.85 & 1.95 & 1040 @xmath65 1712 & 12 27 07.54 & 8 13 19.0 & 20.36 & 1.34 & 1144 @xmath66 1731 & 12 27 28.81 & 8 13 21.1 & 20.71 & 1.82 & 1294 @xmath67 1749 & 12 27 14.86 & 8 13 22.6 & 20.92 & 1.98 & 1407 @xmath68 1764 & 12 27 12.15 & 8 13 23.6 & 20.82 & 1.72 & 855 @xmath61 1798 & 12 27 12.65 & 8 13 25.9 & 20.69 & 1.98 & 811 @xmath51 1982 & 12 27 9.12 & 8 13 42.0 & 20.89 & 1.01 & 648 @xmath69 2031 & 12 27 15.13 & 8 13 46.4 & 20.71 & 1.37 & 1352 @xmath70 2045 & 12 27 6.50 & 8 13 47.8 & 20.94 & 1.77 & 857 @xmath71 2060 & 12 27 7.15 & 8 13 48.3 & 20.62 & 1.29 & 1108 @xmath70 2140 & 12 27 21.65 & 8 13 55.5 & 20.45 & 1.80 & 784 @xmath51 2163 & 12 27 23.34 & 8 13 57.7 & 20.15 & 2.01 & 402 @xmath69 2195 & 12 26 58.16 & 8 14 00.8 & 21.33 & 1.14 & 1241 @xmath52 2256 & 12 27 14.99 & 8 14 05.8 & 21.24 & 1.90 & 954 @xmath72 2341 & 12 27 00.22 & 8 14 13.0 & 20.76 & 1.91 & 1001 @xmath73 2406 & 12 27 13.24 & 8 14 18.5 & 20.84 & 2.03 & 1244 @xmath74 2420 & 12 27 08.45 & 8 14 19.3 & 20.95 & 2.08 & 763 @xmath75 2452 & 12 27 25.28 & 8 14 22.5 & 21.49 & 1.40 & 1828 @xmath76 2482 & 12 27 10.16 & 8 14 24.7 & 21.58 & 2.08 & 767 @xmath52 2528 & 12 27 16.79 & 8 14 28.3 & 20.34 & 1.46 & 654 @xmath70 2543 & 12 27 20.34 & 8 14 29.3 & 20.27 & 1.36 & 1199 @xmath77 2569 & 12 27 11.33 & 8 14 31.7 & 20.12 & 1.89 & 1056 @xmath78 2634 & 12 27 07.09 & 8 14 38.0 & 19.70 & 1.56 & 1014 @xmath79 2753 & 12 27 13.65 & 8 14 47.0 & 20.88 & 1.19 & 945 @xmath80 2759 & 12 26 53.73 & 8 14 47.3 & 19.97 & 1.31 & 654 @xmath75 2817 & 12 27 30.93 & 8 14 50.6 & 21.05 & 1.50 & 665 @xmath81 3150 & 12 27 05.95 & 8 15 12.2 & 21.40 & 1.79 & 952 @xmath82 3307 & 12 27 29.92 & 8 15 21.4 & 20.25 & 1.53 & 1790 @xmath70 3361 & 12 27 01.91 & 8 15 26.6 & 20.34 & 1.55 & 1392 @xmath83 3628 & 12 27 00.59 & 8 15 44.0 & 21.22 & 1.90 & 1008 @xmath84 3635 & 12 27 24.04 & 8 15 44.5 & 20.86 & 1.34 & 936 @xmath85 3757 & 12 27 13.40 & 8 15 51.8 & 21.02 & 1.82 & 1220 @xmath86 3808 & 12 27 06.60 & 8 15 54.6 & 20.35 & 1.83 & 832 @xmath87 3909 & 12 27 33.58 & 8 16 00.4 & 21.43 & 1.36 & 1253 @xmath55 3980 & 12 27 03.05 & 8 16 05.4 & 21.15 & 1.28 & 1112 @xmath88 4168 & 12 27 07.64 & 8 16 19.4 & 20.36 & 1.68 & 1384 @xmath89 4210 & 12 27 34.25 & 8 16 22.9 & 20.53 & 1.62 & 1910 @xmath53 4386 & 12 27 19.17 & 8 16 34.7 & 19.83 & 1.94 & 1197 @xmath83 4513 & 12 27 09.78 & 8 16 42.8 & 20.10 & 1.85 & 908 @xmath90 4731 & 12 27 09.56 & 8 16 58.2 & 19.96 & 1.43 & 698 @xmath79 4780 & 12 27 20.80 & 8 17 00.9 & 19.52 & 1.95 & 971 @xmath88 4959 & 12 27 23.06 & 8 17 11.9 & 21.38 & 1.33 & 1449 @xmath89 5090 & 12 27 09.43 & 8 17 20.3 & 19.83 & 1.61 & 582 @xmath78 5323 & 12 27 16.29 & 8 17 33.9 & 20.33 & 0.82 & 1263 @xmath70 5456 & 12 27 12.49 & 8 17 41.4 & 19.26 & 1.39 & 737 @xmath70 5561 & 12 26 56.85 & 8 17 48.3 & 20.82 & 1.39 & 903 @xmath77 5564 & 12 27 34.86 & 8 17 48.5 & 21.26 & 1.37 & 862 @xmath91 5629 & 12 27 11.90 & 8 17 52.3 & 21.09 & 1.36 & 522 @xmath92 5707 & 12 27 12.63 & 8 17 55.9 & 21.47 & 1.44 & 1712 @xmath56 5750 & 12 27 12.08 & 8 17 58.9 & 21.50 & 1.47 & 1063 @xmath93 5856 & 12 27 09.09 & 8 18 06.4 & 21.03 & 1.13 & 1050 @xmath94 6108 & 12 27 24.66 & 8 18 22.8 & 21.49 & 1.42 & 913 @xmath49 6164 & 12 27 12.26 & 8 18 27.2 & 19.79 & 1.65 & 426 @xmath56 6231 & 12 27 16.55 & 8 18 32.3 & 20.77 & 1.82 & 1046 @xmath95 6284 & 12 27 25.39 & 8 18 36.8 & 19.44 & 1.57 & 569 @xmath71 6294 & 12 27 01.30 & 8 18 37.7 & 21.02 & 1.64 & 1034 @xmath94 6344 & 12 27 07.59 & 8 18 41.3 & 20.89 & 2.01 & 1220 @xmath95 6357 & 12 27 20.14 & 8 18 41.9 & 20.47 & 1.26 & 958 @xmath53 6388 & 12 27 23.15 & 8 18 44.5 & 20.18 & 1.36 & 1212 @xmath54 6394 & 12 27 03.03 & 8 18 45.0 & 21.40 & 1.42 & 760 @xmath96 6427 & 12 27 12.47 & 8 18 47.5 & 21.11 & 1.79 & 1141 @xmath95 6476 & 12 27 43.03 & 8 18 50.2 & 20.94 & 2.23 & 966 @xmath62 6485 & 12 27 22.50 & 8 18 51.2 & 21.07 & 1.53 & 510 @xmath97 6520 & 12 27 16.76 & 8 18 53.6 & 20.06 & 1.86 & 607 @xmath79 6564 & 12 27 10.77 & 8 18 57.3 & 20.03 & 1.34 & 1077 @xmath51 6615 & 12 27 41.49 & 8 19 01.3 & 20.48 & 1.47 & 1923 @xmath84 6696 & 12 27 21.46 & 8 19 08.5 & 20.08 & 1.59 & 561 @xmath98 6701 & 12 26 53.65 & 8 19 08.6 & 20.97 & 2.00 & 1092 @xmath99 6721 & 12 27 07.36 & 8 19 10.5 & 21.09 & 1.73 & 1180 @xmath88 6748 & 12 27 08.24 & 8 19 12.4 & 20.21 & 1.53 & 817 @xmath100 6872 & 12 27 09.21 & 8 19 20.8 & 20.15 & 1.46 & 870 @xmath101 6989 & 12 27 22.00 & 8 19 30.6 & 20.61 & 1.75 & 1009 @xmath54 7028 & 12 27 39.52 & 8 19 33.5 & 21.40 & 1.38 & 1548 @xmath49 7043 & 12 26 56.69 & 8 19 34.4 & 20.47 & 1.78 & 808 @xmath102 7095 & 12 27 29.05 & 8 19 39.8 & 21.43 & 1.56 & 1285 @xmath90 7197 & 12 27 08.44 & 8 19 48.2 & 20.94 & 1.50 & 782 @xmath95 7281 & 12 27 20.64 & 8 19 53.8 & 21.43 & 1.25 & 389 @xmath103 7340 & 12 27 17.40 & 8 19 59.0 & 20.91 & 1.77 & 1067 @xmath53 7364 & 12 26 58.73 & 8 20 01.4 & 21.36 & 1.36 & 1522 @xmath80 7399 & 12 27 01.33 & 8 20 05.1 & 20.35 & 1.40 & 1005
@xmath89 7430 & 12 26 53.05 & 8 20 06.5 & 20.76 & 1.41 & 862 @xmath58 7449 & 12 26 47.44 & 8 20 07.7 & 20.75 & 1.65 & 724 @xmath70 7458 & 12 27 12.24 & 8 20 08.7 & 20.75 & 1.84 & 807 @xmath79 7531 & 12 26 52.55 & 8 20 13.5 & 19.71 & 2.07 & 818 @xmath104 7616 & 12 27 16.55 & 8 20 20.2 & 21.35 & 1.49 & 600 @xmath55 7659 & 12 27 10.56 & 8 20 24.0 & 19.87 & 1.34 & 1520 @xmath89 7702 & 12 27 38.87 & 8 20 27.4 & 21.19 & 1.88 & 1388 @xmath105 7746 & 12 27 27.01 & 8 20 31.5 & 21.29 & 1.42 & 712 @xmath73 7784 & 12 27 23.23 & 8 20 34.4 & 19.20 & 1.52 & 868 @xmath67 7798 & 12 27 32.56 & 8 20 35.4 & 20.96 & 1.39 & 1340 @xmath49 7872 & 12 27 14.71 & 8 20 42.0 & 20.33 & 1.45 & 908 @xmath106 7886 & 12 26 55.98 & 8 20 43.8 & 20.62 & 1.58 & 1236 @xmath83 7889 & 12 27 25.49 & 8 20 43.5 & 18.85 & 1.58 & 614 @xmath70 7894 & 12 27 01.99 & 8 20 37.3 & 21.61 & 1.73 & 730 @xmath76 7914 & 12 27 18.07 & 8 20 46.4 & 21.20 & 1.33 & 1101 @xmath53 7938 & 12 27 11.91 & 8 20 47.7 & 20.92 & 1.44 & 1251 @xmath95 7945 & 12 26 51.93 & 8 20 48.2 & 19.79 & 1.59 & 651 @xmath83 8000 & 12 27 28.43 & 8 20 53.3 & 21.08 & 1.49 & 368 @xmath66 8090 & 12 27 13.14 & 8 21 00.9 & 20.51 & 1.46 & 903 @xmath58 8143 & 12 27 31.74 & 8 21 04.7 & 20.96 & 1.52 & 672 @xmath107 8164 & 12 27 26.33 & 8 21 06.9 & 21.32 & 2.04 & 738 @xmath66 8165 & 12 27 23.58 & 8 21 06.9 & 20.22 & 1.39 & 1027 @xmath81 8210 & 12 26 54.15 & 8 21 10.1 & 20.43 & 1.32 & 576 @xmath68 8273 & 12 27 01.10 & 8 21 15.9 & 20.60 & 1.43 & 784 @xmath55 8332 & 12 27 11.29 & 8 21 21.1 & 21.09 & 1.40 & 1226 @xmath80 8353 & 12 27 08.71 & 8 21 22.8 & 20.03 & 1.98 & 928 @xmath66 8357 & 12 27 04.34 & 8 21 22.9 & 20.26 & 1.45 & 981 @xmath63 8384 & 12 27 15.30 & 8 21 24.4 & 21.39 & 1.41 & 768 @xmath71 8596 & 12 27 34.37 & 8 21 44.6 & 19.73 & 1.31 & 888 @xmath51 8653 & 12 26 43.83 & 8 21 49.5 & 20.50 & 1.26 & 744 @xmath89 8712 & 12 27 40.99 & 8 21 54.7 & 20.93 & 1.49 & 817 @xmath108 8740 & 12 27 09.77 & 8 21 57.5 & 21.27 & 2.14 & 913 @xmath109 8890 & 12 27 15.78 & 8 22 14.1 & 20.41 & 1.88 & 870 @xmath70 8919 & 12 27 01.83 & 8 22 16.0 & 19.87 & 1.45 & 1014 @xmath70 9145 & 12 27 42.37 & 8 22 39.1 & 19.79 & 1.76 & 973 @xmath110 9360 & 12 26 46.61 & 8 23 06.8 & 21.00 & 1.67 & 1191 @xmath111 9414 & 12 27 06.09 & 8 23 14.3 & 20.82 & 1.74 & 832 @xmath66 9527 & 12 26 59.87 & 8 23 32.0 & 20.91 & 1.58 & 941 @xmath63 9666 & 12 27 18.87 & 8 23 51.3 & 20.04 & 1.74 & 811 @xmath112 9991 & 12 27 26.39 & 8 14 35.5 & 19.41 & 1.27 & 1040 @xmath70 9992 & 12 27 15.80 & 8 17 16.7 & 19.99 & 1.48 & 641 @xmath70 4415 & 12 27 09.11 & 08 16 36.7 & 20.45 & 2.17 & 45000 2597 & 12 27 12.91 & 08 14 34.2 & 20.50 & 1.24 & 20000 2431 & 12 27 19.22 & 08 14 20.4 & 21.44 & 1.28 & 14800 1746 & 12 26 47.59 & 08 13 22.5 & 20.70 & 1.61 & 222 2430 & 12 27 03.46 & 08 14 20.4 & 21.45 & 1.75 & @xmath113 7790 & 12 26 47.97 & 08 20 34.7 & 19.74 & 1.41 & 11334 8113 & 12 27 06.93 & 08 21 02.1 & 20.88 & 1.33 & 10982 7569 & 12 27 35.79 & 08 20 16.3 & 21.25 & 1.44 & 10706 7039 & 12 27 43.90 & 08 19 34.2 & 21.21 & 1.70 & 34686 8062 & 12 27 30.61 & 08 20 58.8 & 19.97 & 1.18 & 83 9103 & 12 27 36.56 & 08 22 35.6 & 20.97 & 1.35 & 224 3900 & 12 27 16.98 & 08 16 00.0 & 21.04 & 1.89 & 10600 3217 & 12 27 36.55 & 08 15 16.3 & 21.05 & 1.29 & 8868 1922 & 12 27 37.08 & 08 13 36.7 & 21.16 & 1.60 & 32278 1024 & 12 27 14.18 & 08 12 12.5 & 20.99 & 1.11 & 200 1621 & 12 27 27.58 & 08 13 11.0 & 19.69 & 1.41 & 159 2935 & 12 27 40.19 & 08 14 57.6 & 19.87 & 1.06 & 133 1955 & 12 27 11.63 & 08 13 39.4 & 21.04 & 1.20 & 31000 8224 & 12 27 17.37 & 08 21 11.4 & 21.34 & 1.79 & 11700 6302 & 12 27 33.74 & 08 18 38.2 & 20.49 & 1.95 & 12000 .results from non - linear fits of equation 1 to globular cluster samples taken from table 1 .
the second column gives the mean velocity for each sample .
the third column gives the least - squares fit to the rotation velocity and the fourth column gives the best - fit position angle .
the final column gives the @xmath1 upper limit on the rotational velocity . [ cols="^,^,^,^,^,^,^,^ " , ] | we present new radial velocities for 87 globular clusters around the elliptical galaxy ngc 4472 , and combine these with our previously published data to create a data set of velocities for 144 globular clusters around ngc 4472 .
we utilize this data set to analyze the kinematics of the ngc 4472 globular cluster system .
the new data confirms our previous discovery that the metal - poor clusters have significantly higher velocity dispersion than the metal - rich clusters in ngc 4472 .
we also find very little or no rotation in the more spatially concentrated metal - rich population , with a resulting upper limit for this population of @xmath0 at a @xmath1 confidence level .
the very small angular momentum in the metal - rich population requires efficient angular momentum transport during the formation of this population which is spatially concentrated and chemically enriched .
such angular momentum transfer can be provided by galaxy mergers , but has not been achieved in other extant models of elliptical galaxy formation that include dark matter halos .
we also calculate the velocity dispersion as a function of radius , and show that it is consistent with roughly isotropic orbits for the clusters and the mass distribution of ngc 4472 inferred from x - ray observations of the hot gas around the galaxy . |
ode - locking ( ml ) is a fascinating phenomenon that allows the generation of ultrashort pulses from a laser @xcite and that is still a subject of intense research .
passive ml ( pml ) is arguably the most successful approach and it is achieved by combining two elements , a laser amplifier which provides gain and a saturable absorber ( sa ) acting as a pulse shortening element , see @xcite for a review . pulsed emission with a fundamental period corresponding to the cavity round - trip time arises from the different dynamical properties of the sa and the amplifier , which open a short window for amplification around the pulse @xcite .
ml has led to the shortest and most intense optical pulses ever generated and pulses in the femtosecond range are produced by dye @xcite and solid state lasers @xcite .
large output powers in the watt range are commonly achieved from coupled vecsels - sesam @xcite configurations . on the other hand ,
the large gain of semiconductor materials allows building sub - millimeter monolithic pml lasers .
the round - trip of such short devices is typically of the order of @xmath1 ps and they can therefore reach repetition frequencies of several tens of ghz .
they have also the advantage of being compact , low cost and adaptable to many cavity geometries @xcite , yet their power is in the milli - watt range . in spite of the research efforts dedicated to the general understanding of the basic issues ,
several aspects of pml still represent a scientific challenge .
for instance , while the breadth of the active medium gain curve is well known to govern the pulse - width , it was only very recently understood that the sa also provides a strongly asymmetric and non linear filtering .
such an usually overlooked effect was recently proven to be the key mechanism explaining wavelength instabilities in pml lasers @xcite .
in addition , recent studies indicated that the optimal position of the sa in the cavity was not following intuitive rules @xcite .
similarly , it was commonly thought until recently that pml semiconductor lasers can not operate at low repetition rates due to the fast recovery time of their material gain ( @xmath2 ns ) which should limit them to high repetition frequencies ( @xmath3 ghz ) .
too long cavities result in the so - called regime of harmonic mode - locking in which several pulses circulate in the cavity , according to the background stability criterion @xcite . until recently , the record of the lowest frequency pml was experimentally attained around @xmath4mhz @xcite .
we have recently shown that _ _ , _ _ in the limit of cavity round - trip much longer than the gain recovering time , mode - locked pulses may coexist with the zero intensity background and can be interpreted as _ temporal localized structures _
( ls ) @xcite .
localized pulses can be independently addressed and used as elementary bits of information , hence the cavity can be used as an all - optical buffer , as shown in @xcite .
a single localized pulse can be activated within the cavity , independently from the cavity size , thus leading to arbitrary low repetition rates which allowed establishing a new low frequency record of @xmath5mhz @xcite .
the possibility of addressing individually localized pulses opens interesting possibilities for the optical generation of arbitrary trains of narrow pulses , which has a large number of potential applications in different domains , e.g. time - resolved spectroscopy , pump - probe sensing of material properties , generation of frequency combs , optical code division multiple access communication networks @xcite and lidar @xcite . in this manuscript ,
we show how localized pulses form in a passively mode - locked semiconductor broad - area vcsel coupled to a resonant saturable absorber mirror ( rsam ) .
the vcsel bias current is used as a control parameter for addressing localized pulses and for generating patterns of closely packed narrow pulses . the experimental evidences and
theoretical analysis pave the path towards an optical arbitrary pattern generator of short light pulses .
, scaledwidth=37.0% ] our vcsel is a @xmath6 nm device manufactured by ulm photonics @xcite .
its standalone threshold current ( @xmath7 ) is about @xmath8ma , though emission is localized only at the periphery of the device up to @xmath9ma , after which roll - off starts to occur .
the @xmath6 nm rsam ( batop gmbh ) has a transverse dimension of @xmath10mm@xmath11 and it exhibits a low unsaturated reflectivity of @xmath12 that increases up to @xmath13 when saturated .
the rsam saturation fluence is @xmath14j.@xmath15 .
these values are obtained at the rsam resonant wavelength which can be thermally tuned over @xmath16 nm ( between @xmath17 and @xmath18 ) .
the full width at half maximum ( fwhm ) of the rsam resonance is around @xmath19 nm and the saturable absorption recovery time is around @xmath20ps .
the setup is shown in fig .
[ setup ] .
both the vcsel and rsam are mounted on temperature controlled substrates which allow for tuning the resonance frequency of each cavity ; parameters are set for having the emission of the vcsel resonant with the rsam .
the light emitted by the vcsel is collected by a large numerical aperture ( 0.68 ) aspheric lens and a similar lens is placed in front of the rsam .
a @xmath21 reflection beam splitter allows for light extraction from the external cavity and to monitor both the vcsel and the rsam outputs .
intensity output is monitored by a @xmath22ghz oscilloscope coupled with fast @xmath1ghz detector .
part of the light is sent to two ccd cameras ; the first one records the near - field profile of the vcsel , while the second records the vcsel s far - field profile .
the external cavity length round - trip ( @xmath23 ) is fixed at @xmath24ns which corresponds to a @xmath25mhz fundamental repetition rate .
the combination of the vcsel and the rsam in self - imaging condition does not induce an appropriate ratio of the saturation parameters for obtaining pml .
notwithstanding , mode - locking was obtained when placing the rsam surface in the exact fourier transform plane of the vcsel near - field profile , i.e. when imaging the vcsel far - field profile onto the rsam surface @xcite . as a consequence , the vcsel profile was imaged onto itself after a single external cavity round - trip , but inverted ( i.e. a magnification factor of @xmath26 ) .
as shown in @xcite , such configuration leads to the generation of two opposed tilted plane waves , i.e. waves that have a propagation wave - vector slightly out of the cavity axis @xmath27 and which travel in the external cavity with an opposite transverse component and alternating each other at every round - trip .
each one of these plane waves gives birth to a train of mode - locked pulses separated by twice the external cavity round - trip ( @xmath28 ) , while the two trains are time shifted of @xmath23 .
intuitively , one understands that injecting the fourier transform of the vcsel near - field profile into the rsam strongly favors the emission of a tilted wave from all points of the vcsel , since the fourier transform of such an emission consists in a tight focused single spot that saturates easily the rsam . while this scheme leads to conventional mode - locking and harmonic mode - locking for cavity lengths and round - trips shorter than the gain recovery time ( @xmath29 ) , we operate the system in the regime of long cavities , i.e. @xmath30 , and for bias currents below the laser threshold . in these conditions , as shown in @xcite , several emission states , including the zero intensity ( off ) solution , coexist for the same values of bias current . 280 ma and @xmath24ns .
the panels a ) , b ) , c ) and d ) correspond to @xmath31 pulses per round - trip , respectively .
another configuration for @xmath32 is shown in e ) where the pulses are regularly distributed within the cavity .
the maximal number of pulse is @xmath33 as depicted in panel f ) .
right : experimentally obtained bifurcation diagram for the number of pulses per round - trip ( g ) as a function of the bias current .
the stability of each solution is indicated by the solid horizontal lines . ]
these emission states are characterized by a different number of pulses circulating in the cavity , and for the same number of pulses , a large variety of pulse arrangement can be observed . in fig .
[ traces]a - d ) few examples between the large number of coexisting emission states is represented , including @xmath34 and @xmath35 pulses circulating inside the cavity . for the @xmath32 pulses case
, we show two situations where they appear either grouped , see fig .
[ traces]d ) , or equally separated fig .
[ traces]e ) .
the case of @xmath33 pulses circulating within the cavity represents the largest number of pulses that can be obtained for the size of the external cavity chosen .
all the above listed pulsing solutions coexist in the parameter space for a wide range of vcsel pumping current @xmath36 .
the pulse width can not be determined precisely from the oscilloscope traces shown in fig .
[ traces ] , which are limited by our real - time detection system ( 10 ghz effective bandwidth ) .
however , an estimation of the pulse width can be obtained from the optical spectrum of the output , which exhibits a broad spectral peak whose fwhm is around 0.12 nm that corresponds , assuming a time - bandwidth product of 0.4 , to a pulse width of 10 ps fwhm .
the pulse was also detected by a 36 ghz detector , which confirms a pulse width of less than 12 ps fwhm considering the oscilloscope bandwidth limit .
finally , the pulse peak power has been measured to be @xmath20watt .
the multistability between a large number of different solutions is shown in fig .
[ traces]g ) , where we plot the bifurcation diagram of these solutions as a function of the pumping current .
figure [ traces]g ) is obtained increasing the parameter @xmath36 from @xmath37ma , where only the steady off solution is stable , up to the value where this solution loses its stability ( @xmath38ma ) and then sweeping it down until a periodic emission with @xmath33 pulses per round - trip appears , see fig .
[ traces]f ) . in analogy to spatial localized structures
@xcite , this state corresponds to the fully developed temporal pattern which is , together with the coexisting stable off solution , at the origin of the localized structures .
as @xmath36 is decreased , the state with @xmath33 pulses per round - trip loses its stability and the system bifurcates progressively towards states with smaller numbers of ls , until a single one is present in the cavity .
each state is spontaneously appearing as @xmath36 is scanned downward and , once a new state appears , we increase @xmath36 to explore its stability up to @xmath39 ma .
as far as the system remains on the same branch there is no change in the pulse arrangement , thus showing that , even if several arrangements are possible , once one is chosen , it is stable versus parameter changes . figure [ traces ] indicates that in this regime , mode - locked pulses in our system can be interpreted as temporal localized structures , i.e. any pulse is independent from the other and it can be individually addressed .
it is worthwhile noting that the localized character of the pulse implies that the pulse is localized not only in the intensity but also in all the variables describing the system .
this means that , after the short intensity pulse emission , the gain and saturable loss variables recover their steady state value on a longer time scale thereby defining the effective temporal extent of the ls .
this is the main difference between localized pulses and conventional mode - locked pulses . in the latter case , which is obtained for pumping levels above laser threshold
, the gain does not recover to its steady state value between pulses .
this is well explained by the background stability criterion @xcite which states that a pulsating solution is stable if losses are larger the gain between pulses .
since the laser is pumped above threshold , the only stable pulsating solution is the one having a number of pulses in the cavity large enough such that , in - between pulses , gain can not recover sufficiently to overcome the unsaturated loss level .
this implies that fundamental mode - locking is possible only for @xmath29 . in the case of localized pulses
, we are in the opposite limit and the gain can recover fully between pulses because the maximal gain level remains below the unsaturated loss level , even for an infinitely long interval between pulses .
thus , localized pml pulses have no lower limit in their repetition rate , as shown in fig .
[ traces]a ) where a repetition rate of 66.6 mhz is obtained . on the other hand
, the recovery time of the slowest variable , in our case the gain recovery time , determines the effective temporal extent of the localized pulse , since no other pulse can occur before the recovery of all the variables .
another important difference between localized pulses and conventional mode - locked pulses is that , for a given set of parameters , the energy of the localized pulses does not depend significantly on the number of pulses circulating inside the cavity .
instead in conventional mode - locking the energy of the system is shared between the pulses which circulate inside the cavity and fundamental mode - locking is preferred over harmonic mode - locking for maximum pulse energy .
f).[diagcavlong ] ] in order to represent the evolution of the localized pulses travelling within the cavity over many round - trips , we use the so - called space - time diagram representation where the temporal trace is folded onto itself at the cavity round - trip period .
accordingly , the round - trip number @xmath40 becomes a pseudo - time variable while the pseudo - space variable corresponds to the timing of the pulse modulo @xmath23 .
this representation is similar to the one used for following the evolution of pulses along optical fibers where fast and slow time scales are well separated and it has been proposed also for delayed system @xcite . in fig .
[ diagcavlong]a - d ) we show some space - time like diagrams obtained from the same time traces used for fig .
[ traces ] where all the situations depicted coexist for the same parameter values .
panel a ) shows the evolution of a single ls in the external cavity , while panel b ) , c ) and d ) represent the evolution of respectively @xmath41 and @xmath42 pulses . in all the situations ,
the pulses do not visibly jitter over the considered timescale of @xmath43s and they seems to be insensitive to the surrounding noise .
this can be attributed to the presence of the saturable absorber which acts as an effective noise eater . while the rsam opens a very short window of net gain around the pulse , the remaining low intensity emission from the vcsel
is absorbed by the rsam .
accordingly , the background shown in fig.[diagcavlong ] is homogeneous and corresponds to zero emission . in presence of localized structures , as the ones shown in fig .
[ diagcavlong ] , the possibility of forming bounded states of localized structures , also called molecules , naturally arises.when gathered in molecules , the pulses are not independent anymore , while the molecule itself becomes a compound localized structure which is independent from other localized structures that may travelling within the cavity .
such molecules have been widely observed in fiber lasers @xcite and in general only a few if not an unique equilibrium distance separates the `` atoms '' of a molecule .
the signature of such bound states of ls in a space - time like diagram is the presence of preferred distances between two neighbor pulses .
we have never found any preferred distances between close pulses in the space - time like diagrams that we have analyzed .
instead we have observed a _
continuum _ of temporal separations .
one can understand the absence of molecules in our system by the specificity of the interaction between neighbor ls .
indeed , it is known that the gain depletion induced by each ls produces a repulsive interaction @xcite that decays over a time scale @xmath44 .
the potentially attractive interaction mediated by the oscillating tails of the pulses vanishes after a few times the effective pulse width , i.e. a time scale of @xmath45ps . as such ,
this clear scale separation between the attractive and repulsive forces explain why ls molecules would be hard to obtain with semiconductor active materials
. it may be useful to compare our results with the extensive literature on mode - locking .
localized pulses in mode locked laser might have been observed in the past .
for example , several papers mention pml regimes obtained at _ exceedingly low _ frequencies with respect to the limits of the background stability criterion ; repetition rates of the order of few hundreds mhz were reported in@xcite .
we believe that these regimes could be explained in terms of ls . while it is difficult to know exactly the experimental conditions of these works , in @xcite it is clearly mentioned that the off solution coexists with the pml solution . the experimental result reported recently in @xcite , claiming mode - locking at a repetition rate of 85.7 mhz ,
can also be explained in terms of ls . in this work , neighboring pulses of @xmath46ps fwhm , separated of @xmath20ns ,
are interpreted theoretically as bound states .
while the experimentally observed distance corresponds to the vanishing of the repulsive interaction induced by the gain depletion @xcite , the theoretical analysis is performed after adiabatically eliminating the carrier dynamics , thus neglecting such repulsive interactions .
we believe that this approximation in @xcite , correct for fiber lasers but not for semiconductor devices , has led to a wrong interpretation of independent neighbor pulses in terms of bound states resulting of the locking of the oscillating tails of the pulses .
, scaledwidth=50.0% ] the most important property of localized structures is their mutual independence which allows for their individual addressing .
the localized pulses that we have obtained can be individually triggered by shining light pulses inside the cavity , as we have shown numerically in @xcite . by encoding an information bit in the form of a localized pulse
, our system can be operated as an all - optical buffer , similar to the one proposed using fiber kerr resonator with a driving field @xcite , but taking advantage of the compactness and fast response of semiconductor materials as in @xcite .
the buffer bit - rate depends on the minimal time interval between two independent pulses , i.e. the temporal extent of the ls which is , in our case , fixed by the gain recovery time . the experimental results shown in fig .
[ traces]f ) indicate that a separation of 1 ns is certainly sufficient to avoid interaction between pulses , thus leading to buffer bit rate of @xmath47ghz . while the all - optical addressing of the mode - lock localized pulses is the topic of ongoing research ,
the nucleation of a single pulse can be observed when perturbing mechanically the system , as shown in fig .
[ allumage ] . yet
this kind of perturbation does not allow any control of the switching ; it may lead to the nucleation of several pulses at the same time with random positions within the cavity .
notwithstanding , a higher degree of control can be implemented when perturbing the bias current of the vcsel . the main limitation using
this parameter comes from the frequency response of the vcsel to a modulated signal coupled onto the pumping current via a bias t. the electrical characteristics of the broad - area vcsel we used , together with the bandwidth of the bias - t used , limit the frequency modulation below 0.3 ghz .
accordingly , the individual addressing of localized pulses via the emission of electrical pulses suffers from the too long rise time of the perturbation .
nevertheless a sinusoidal modulation of the pumping current with a period corresponding to the cavity round - trip @xmath23 allows for the excitation of localized pulses in a narrow time interval within the round - trip time , i.e. a well - defined vertical band of the space - like variable , as plotted in fig .
[ diagcavlong ] .
the idea would be to modulate the current between a lowest value where the off solution is the only possible solution up to a highest value where the fully developed temporal pattern shown in fig .
[ traces]f ) is also the unique possible situation .
since this modulation is synchronous with the cavity round - trip , a portion of this monostable temporal pattern will appear only in correspondence with the modulation signal peak .
once the sinusoidal perturbation of the pumping current is removed , the portion of the monostable pattern will give way to a solution with one , two or more localized pulse depending on the continuous bias current value of the vcsel , see fig .
[ traces]g ) , and on the size of the the temporal pattern excited .
the result of this experiment is shown in fig .
[ buffertrace ] where the vcsel is biased at @xmath48ma , a current value for which for the sake of simplicity only the state with a single localized pulse per round - trip coexists with the off state . we prepare the system in the off state and then we apply a modulation having a peak to peak amplitude of 460 ma . in correspondence with the modulation peak ( where @xmath49ma ) , six localized pulses appear at a distance of 800 ps , which corresponds to the interpulse distance of fig .
[ traces]f ) , i.e @xmath50 .
once the modulation signal is removed , the continuous value of pumping current does not support the stable existence of six localized pulses and the system bifurcates towards the state with a single pulse per period .
this regime is stable and it persists indefinitely .
second , modulation is applied with a peak - to - peak amplitude of 460 ma and 6 ls appear close to the peak of the modulation ( green , middle line ) .
when we switch - off current modulation , only a single ls remains in the cavity ( blue , top line ) .
[ buffertrace ] ] the experiment reported above suggests that a current modulation can be used to obtain a pattern of closely packed localized pulses .
this functionality is appealing for many applications like e.g. ocdma @xcite and lidar @xcite , since narrow optical pulses ( < 10 ps ) are typically obtained using conventional pml systems which deliver a periodic train of pulses with no versatility in the pulse arrangement .
pulse patterning can be of course obtained by adding optical gates based on electro - optical modulators , but these solutions are expensive , energetically inefficient , and their implementation requires synchronization between the pulse source and the gating .
we implement a pulse pattern generator by driving the system pumping current with a forcing at a period equal to the cavity round - trip and which brings the system into the monostable region where only the periodic pattern of pulses is stable , according to fig .
[ traces]f ) .
the amplitude of the forcing together with the steady value of the bias current , fix the time interval during which the system is driven in this monostable region and , accordingly , this fixes the number of pulses which are excited . for a stable output
, the modulation period needs to match closely the cavity round - trip @xmath23 , which in our case requires a modulation frequency of @xmath51 = 66.584.030 hz .
we have verified that a stable output pattern is lost if the frequency mismatch exceeds @xmath1khz .
the results obtained are shown in fig .
[ address ] where we control the number of pulse emitted from one to five by changing the amplitude of the sinusoidal forcing .
the distance between pulses is fixed by the effective temporal extent of the pulses which , as explained above , depends on the carrier recombination time .
the close inspection of panel e ) and f ) reveals that the distance between pulses can change slightly within a pattern .
this can be explained by the use of a sinusoidal signal for bringing the system in the monostable region of fig .
[ traces]f ) , where a periodic pattern of pulses is emitted . in the portion of the sinusoidal signal where this occurs , the current is still increasing , thus leading to an effectively faster carrier recovery rate and therefore to closer pulses . for the same reason , pulse intensities vary slightly inside the pattern .
the appearance of an increasing number of pulses for larger amplitudes of the forcing , as well as the inversed phenomenon while decreasing the modulation strength , occurs with an high degree of hysteresis as seen in fig .
[ address]f ) , thus revealing multistability between the different situations depicted in fig .
[ address]a)-e ) ma , @xmath52ns , @xmath51 = 1/@xmath23 . as the modulation peak - to - peak amplitude
is increased a ) 297 ma , b ) 306 ma , c ) 349 ma , d ) 405 ma and e ) 454 ma , the number of pulses close to the peak of the modulation is incremented by one .
f ) the reverse sequence appears when decreasing the modulation amplitude .
the modulation value at which transition from @xmath53 pulses to @xmath54 pulses happens is slightly different with respect the modulation value at which the inverse transition from @xmath54 pulses to @xmath53 occurs , denoting a strong multistability , .
[ address ] ] while fig .
[ address ] discloses the proof - of - principle of a pulse pattern generator based on localized pulses , more complicated pulse pattern structures can be envisioned using other kind of modulation signals , as square or pulsed current modulation . even if these modes of operation require to solve the problem of the electrical coupling with the vcsel , we have tried to use a sinusoidal signal at a period corresponding to @xmath55 for addressing two patterns of pulses within the cavity . the result is shown in fig
. [ address2fsr ] , where two localized pulses are separated by half of the round - trip time , in correspondence to the modulation peaks . .
the bias is @xmath37ma , and the modulation amplitude @xmath56ma.[address2fsr ] ]
the proper study and the simulation of pml lasers is a demanding problem from the computational point of view : while pulses may form on a relatively short time scale of a few tens of round - trips , the pulse characteristics only settle on a much longer time scale @xcite .
if anything , the complex transverse dynamics @xcite also present in our case shall slow down the dynamics even further . in particular , we consider here extremely long cavities and therefore very long delays for which with the pulse to round - trip aspect ratio is @xmath57 .
the necessary presence of noise prevents the use of adaptive step - size algorithms , which would be particularly suitable for such strongly nonlinear problems .
hence , for the sake of simplicity , we use a purely temporal model as for instance the generic delay differential equation model of @xcite which generalizes haus model as it encompasses both the pulsating and the steady regimes .
while more detailed results could also be obtained with a traveling - wave approach @xcite , we found that such a simple model was sufficient to give a good qualitative agreement with the experimental results .
denoting by @xmath58 the amplitude of the optical field , @xmath59 the gain , and @xmath60 the saturable absorber losses , the model reads @xmath61a_{\tau}-a,\label{eq : vt1}\\ \dot{g } & \negthickspace=\negthickspace\negthickspace & g_{0}+\delta g\sin\left(\frac{2\pi t}{\tau_{m}}\right)-\gamma g - e^{-q}\left(e^{g}-1\right)\left|a\right|^{2},\label{eq : vt2}\\ \dot{q } & \negthickspace=\negthickspace\negthickspace & q_{0}-q - s\left(1-e^{-q}\right)\left|a\right|^{2},\label{eq : vt3}\end{aligned}\ ] ] where time has been normalized to the sa recovery time , @xmath62 and @xmath63 are the linewidth enhancement factors of the gain and absorber sections respectively , @xmath64 is the fraction of the power remaining in the cavity after each round - trip , @xmath65 is the pumping rate , @xmath66 and @xmath67 the amplitude and the period of the modulation of the gain , @xmath68 is the gain recovery rate , @xmath69 is the value of the unsaturated losses which determines the modulation depth of the sa , @xmath70 is the ratio of the saturation energy of the sa and of the gain sections and @xmath71 is the bandwidth of the spectral filter . in eq .
[ eq : vt1 ] the subscript @xmath23 denotes a delayed value of the variable , @xmath72 .
this delay renders the dynamical system infinitely - dimensional and it describes the spatial boundary conditions of a cavity closing onto itself . as such it governs the fundamental repetition rate of the pml laser .
we use standard parameter values that closely represents the experimental situation : @xmath73 , @xmath74 , @xmath75 , @xmath76@xmath77 , @xmath78 and @xmath79 . assuming a sa recovery time of @xmath80ps , this corresponds to a gain recovery of @xmath81ps , a bandwith fwhm of @xmath82ghz , i.e. @xmath83 nm around @xmath6 nm in good agreement with the vcsel resonance linewidth measurement with a tunable laser and a time delay of @xmath84ns .
importantly , we stress that the existence of pml regimes below the lasing threshold is mostly independent of the phase - amplitude couplings which explains while in the past we choose @xmath85 for the sake of simplicity .
however , we set here the more realistic values of @xmath86 and @xmath87 .
the lasing threshold is determined by the pump level @xmath65 for which the off solution @xmath88 becomes linearly unstable ; in our case it is given by @xmath89 . above threshold ,
the continuous wave ( cw ) solution bifurcates supercritically from the off state , but pml is found below threshold where the pulsating branches emerge as saddle - node bifurcation of limit cycles , see @xcite for more details . and @xmath90 .
the dashed green lines ( not to scale ) represent the modulation of the gain .
[ theo_modj_trace ] ] we set the modulation period very close to the delay value as @xmath91 , the choice of such a precise value will become clear in a moment .
the dc bias current is below threshold @xmath92 but also below the minimal current for which ls exists that we denote @xmath93 . as such
, we are in a situation similar to the experiment .
we depict our results in fig .
[ theo_modj_trace ] where we evidence the existence of bistability between different temporal patterns composed of either one or three pulses .
one clearly see that the pulses organize in the region where the gain is maximal .
an inspection of the gain and of the saturable absorption helps to clarify the mechanism of formation of these pulses . during a fraction of the round - trip ,
the gain is above @xmath94 while it remains still below @xmath95 . as such one or
several pulses can be generated .
the pulses do interact via the gain dynamics which is apparent in the fact that they do not have exactly the same heights .
gain dynamics is known to induce repulsive interaction @xcite between pulses .
intuitively , pulses will move toward higher gain , which corresponds to maximizing the distance between them , hence explaining the repulsive nature of the interaction .
however , here they can not move away from each other since the gain varies in time an go below the minimal value at which they can exists .
as such , the equilibrium distance between pulses comes for the complex interplay between the repulsive interaction between pulses and the presence of the modulation that acts as an external potential . over two round - trips with @xmath96 and @xmath90 .
the dashed green lines ( not to scale ) represent the modulation of the gain .
[ theo_modj_gq ] ] a rich multistability diagram for the number of pulse as a function of the gain modulation @xmath66 can be appreciated in fig .
[ theo_modj_diag ] , in agreement with experimental results of fig .
[ address]f ) . for increasing modulation starting from the off solution
, we find in fig .
[ theo_modj_diag ] that small amplitude broad peaks appear for @xmath97 .
this corresponds to the laser reaching exactly threshold at some instant since @xmath92 and such low and broad intensity pulses ( not shown ) can be interpreted as locally amplified spontaneous emission .
the first real ls appears at @xmath98 . upon increasing further the modulation amplitude , the pulses energy slowly increases which explains the slope of the plateau ( the experiment considering only the pulse number ) while the jumps correspond to the apparition of another pulse .
notice in addition that the pulses within a group may have slightly different intensities .
such an information is not apparent in fig .
[ theo_modj_diag ] as we compute the average intensity over a round - trip . upon decreasing the bias ,
a strong hysteresis is apparent as in fig .
[ address]f ) .
interestingly , one is able to reach much lower values for the modulation amplitude .
the value of @xmath99 corresponds to a maximal gain of @xmath100 which is very close to the value of @xmath94 , i.e. the minimal value of the current in dc allowing for the existence of a single ls .
the diagram in fig .
[ theo_modj_diag ] is normalized in such a way that the integrated intensity of a single pulse at the minimal bias current is unity .
we remark that if we have bistability between 1 and 4 pulses , say for instance at @xmath101 , this certainly implies that we can also have 2 and 3 pulses .
as such the diagram of fig .
[ theo_modj_diag ] only gives the extreme envelope of a more complex multi - stable diagram . .
we represent the ( normalized ) average intensity over a round - trip . the blue circles ( resp .
red squares ) correspond to an upward ( resp .
downward ) scan of the modulation amplitude signaling a strong multi - stability .
each jump corresponds to the appearance of the disappearance of an additional ls .
[ theo_modj_diag ] ] which corresponds indeed to the effective round - trip in the cavity dressed by the active medium response time .
one notice the sharpness of such a resonance tongue whose width is typically proportional to the pulsewidth .
[ theo_modj_scantau ] ] experimental results also shown a strong sensitivity of dynamics with respect to the modulation period .
we notice that the period of the pml regimes scales as @xmath102 , see @xcite for instance .
the pulsewidth is proportional to the inverse of the filter bandwidth and there is a deep relation between pulsewidth and deviation of the pulse train period from the value given by the time of flight .
one can intuitively consider that @xmath71 is the `` inertia '' of the filter .
as such , if a short pulse , say a dirac delta , is re - injected after a time of flight @xmath23 , the filter needs a typical time @xmath103 to filter and re - emit it . in our case
, we found that for @xmath92 the fundamental period is @xmath91 which explains why we choose this value in our analysis .
we depict in fig .
[ theo_modj_scantau ] the sensitivity to the modulation period for a low amplitude modulation of @xmath98 .
one can verify from fig .
[ theo_modj_diag ] that in this case we can have either 1 or 4 pulses . in order to have a simple picture
, we start from an initial condition with as many pulses as possible , i.e. the harmonic mode - locking of maximal order found just threshold at @xmath104 in the absence of modulation @xmath105
. then we suddenly decrease @xmath65 toward @xmath92 and set the modulation to @xmath98 .
such a method allows us to find the maximal number of packed pulse that the system can support as a function of the detuning of the modulation frequency .
we see in fig .
[ theo_modj_scantau ] that indeed a precise tuning is necessary as the full width of the resonance tongue is @xmath106 , i.e. of the same order of magnitude than the pulsewidth .
saying that the resonance tongue if of the order of the pulsewidth yields a bandwidth of @xmath107ps/@xmath108khz in good qualitative agreement with the experimental results .
we have shown that electrically biased broad - area vcsels with optical feedback from a rsam can be operated in a regime where the passively mode - locked pulses can be addressed and controlled individually when the compound system is operated below threshold .
the strong multistability between the off solution and a large variety of pulsating solutions with different number and arrangements of pulses per round - trip , demonstrate that the mode - locked pulses are mutually independent .
we show how a modulation of the bias current allows controlling the number of the pulses travelling within the cavity , paving the way an arbitrary pattern generator of picosecond pulses .
j.j . acknowledges financial support from the ramon y cajal fellowship and the cnrs for supporting a visit at the inln where part of his work was developed as well as useful discussions with h. wenzel .
j.j . and s.b .
acknowledge financial support from project ranger ( tec2012 - 38864-c03 - 01 ) and from the direcci general de recerca , desenvolupament tecnolgic i innovaci de la conselleria dinnovaci , interior i justcia del govern de les illes balears co - funded by the european union feder funds .
m.m . and m.g .
acknowledge funding of rgion paca with the projet volet gnral 2011 gedepulse and anr project optiroc .
u. keller , k. j. weingarten , f. x. krtner , d. kopf , b. braun , i. d. jung , r. fluck , c. hnninger , n. matuschek , and j. aus der au , `` semiconductor saturable absorber mirrors ( sesam s ) for femtosecond to nanosecond pulse generation in solid - state lasers , '' _ selected topics in quantum electronics , ieee journal of _ , vol . 2 ,
pp . 435453 , 1996 .
r. hring , r. paschotta , a. aschwanden , e. gini , f. morier - genoud , and u. keller , `` high - power passively mode - locked semiconductor lasers , '' _ quantum electronics , ieee journal of _ , vol .
38 , pp . 12681275 , 2002 .
e. a. avrutin , j. h. marsh , and e. l. portnoi , `` monolithic and multi - gigahertz mode - locked semiconductor lasers : constructions , experiments , models and applications , '' _ iee proc .- optoelectron .
147 , pp . 251278 , 2000 .
p. stolarz , j. javaloyes , g. mezosi , l. hou , c. ironside , m. sorel , a. bryce , and s. balle , `` spectral dynamical behavior in passively mode - locked semiconductor lasers ( invited ) , '' _ photonics journal , ieee _ , vol . 3 , no . 6 ,
1067 1082 , dec .
g. tandoi , j. javaloyes , e. avrutin , c. ironside , and j. marsh , `` subpicosecond colliding pulse mode locking at 126 ghz in monolithic gaas / algaas quantum well lasers : experiments and theory , '' _ selected topics in quantum electronics , ieee journal of _ , vol . 19 , no . 4 , pp . 11006081100608 , 2013 .
j. javaloyes and s. balle , `` anticolliding design for monolithic passively mode - locked semiconductor lasers , '' _ opt .
_ , vol .
36 , no .
22 , pp . 44074409 , nov 2011 .
[ online ] .
available : http://ol.osa.org/abstract.cfm?uri=ol-36-22-4407 m. xia , m. thompson , r. penty , and i. white , `` external - cavity mode - locked quantum - dot laser diodes for low repetition rate , sub - picosecond pulse generation , '' _ selected topics in quantum electronics , ieee journal of _ , vol .
17 , no . 5 , pp . 12641271 , 2011 .
m. cataluna , y. ding , d. nikitichev , k. fedorova , and e. rafailov , `` high - power versatile picosecond pulse generation from mode - locked quantum - dot laser diodes , '' _ selected topics in quantum electronics , ieee journal of _ , vol .
17 , no . 5 , pp . 13021310 , sept 2011 .
j. balzer , t. schlauch , a. klehr , g. erbert , g. trankle , and m. hofmann , `` high peak power pulses from dispersion optimised modelocked semiconductor laser , '' _ electronics letters _
49 , no . 13 , pp . 838839 , june 2013 .
m. marconi , j. javaloyes , s. balle , and m. giudici , `` how lasing localized structures evolve out of passive mode locking , '' _ phys .
112 , p. 223901
, jun 2014 . [ online ] .
available : http://link.aps.org/doi/10.1103/physrevlett.112.223901 f. leo , s. coen , p. kockaert , s. gorza , p. emplit , and m. haelterman , `` temporal cavity solitons in one - dimensional kerr media as bits in an all - optical buffer , '' _ nat photon _
, vol . 4 , no . 7 ,
. 471476 , jul 2010 .
[ online ] .
available : http://dx.doi.org/10.1038/nphoton.2010.120 m. grabherr , r. jager , m. miller , c. thalmaier , j. herlein , r. michalzik , and k. ebeling , `` bottom - emitting vcsel s for high - cw optical output power , '' _ photonics technology letters , ieee _ , vol .
10 , no . 8 , pp . 10611063 , aug 1998 .
m. marconi , j. javaloyes , s. balle , and m. giudici , `` passive mode - locking and tilted waves in broad - area vertical cavity surface emitting lasers , '' _ selected topics in quantum electronics , ieee journal of _ , vol .
pp , no .
11 , 2014 .
p. coullet , c. riera , and c. tresser , `` stable static localized structures in one dimension , '' _ phys .
_ , vol .
30693072 , apr 2000 .
[ online ] .
available : http://link.aps.org/doi/10.1103/physrevlett.84.3069 f. t. arecchi , g. giacomelli , a. lapucci , and r. meucci , `` two - dimensional representation of a delayed dynamical system , '' _ phys .
a _ , vol .
r4225r4228 , apr 1992 .
[ online ] .
available : http://link.aps.org/doi/10.1103/physreva.45.r4225 g. giacomelli and a. politi , `` relationship between delayed and spatially extended dynamical systems , '' _ phys .
_ , vol .
26862689 , apr 1996 .
[ online ] .
available : http://link.aps.org/doi/10.1103/physrevlett.76.2686 m. butkus , e. a. viktorov , t. erneux , c. j. hamilton , g. maker , g. p. a. malcolm , and e. u. rafailov , `` 85.7 mhz repetition rate mode - locked semiconductor disk laser : fundamental and soliton bound states , '' _ opt . express _
21 , no . 21 , pp .
2552625531 , oct 2013 .
[ online ] .
available : http://www.opticsexpress.org/abstract.cfm?uri=oe-21-21-25526 a. g. vladimirov and d. turaev , `` model for passive mode locking in semiconductor lasers , '' _ phys .
a _ , vol .
72 , p. 033808
, sep 2005 .
[ online ] .
available : http://link.aps.org/doi/10.1103/physreva.72.033808 mathias marconi was born in nice , france , in 1988 . from 2006 to 2011 , he was a student at universit de nice sophia antipolis ( uns ) , france . in 2011 , he was an exchange student at strathclyde university , glasgow , u.k .
the same year he obtained the master degree in optics from uns .
he obtained the ph.d .
degree at the institut non - linaire de nice , valbonne , france in 2014 .
his research interests include semiconductor laser dynamics and pattern formation in out of equilibrium systems .
he is a member of the european physical society .
julien javaloyes ( m11 ) was born in antibes , france in 1977 .
he obtained his m.sc . in physics at the ens lyon the phd in physics at the institut non linaire de nice / universit de nice sophia - antipolis working on recoil induced instabilities and self - organization processes in cold atoms .
he worked on delay induced dynamics in coupled semiconductor lasers , vcsel polarization dynamics and monolithic mode - locked semiconductor lasers .
he joined in 2010 the physics department of the universitat de les illes balears as a ramn y cajal fellow .
his research interests include laser dynamics and bifurcation analysis .
patrice camelin made a bts gnie optique photonique before continuing his studies at universit de bourgogne ( ub ) were he obtained his license degree in physics and his master degree in physique lasers et matriaux .
he is currently a phd student at the university of nice sophia - antipolis .
his research interests include laser dynamics , passive mode - locking and spatially extended systems .
daniel chaparro gonzlez holds the licenciado degree by the university of seville .
he studied the postgraduate in germany in the group of non - linear photonics of the institute of applied physics of the university of mnster , where he worked on laser applications for optical tweezers and manipulation and structuration of fluids .
he wrote a thesis entitled fabrication of polymer diffraction gratings by means of optically induced dielectrophoresis with which he obtained the master of science degree by the university of mnster .
nowadays , he is a phd student of the group of non - linear waves of the university of the balearic islands working in theoretical and experimental aspects of the use of semiconductor lasers to perform remote measurements of distances .
salvador balle ( m92 ) was born in manacor , mallorca .
he graduated in physics at the universitat autnoma de barcelona , where he obtained a phd in physics on the electronic structure of strongly correlated fermi liquids .
after postdoctoral stages in palma de mallorca and philadelphia where he became interested in stochastic processes and laser dynamics , he joined in 1994 the physics department of the universitat de les illes balears , where he is professor of optics since 2006 .
his research interests include laser dynamics , semiconductor optical response modeling , multiple phase fluid dynamics and laser ablation .
massimo giudici ( m09 ) received the laurea in fisica from university of milan in 1995 and ph.d from universit de nice sophia - antipolis in 1999 .
he is at present full professor at universit de nice sophia - antipolis and deputy director of the laboratory institut non linaire de nice , where he carries out his research activity .
giudici s research interests revolve around the spatio - temporal dynamics of semiconductor lasers . in particular
, he is actively working in the field of dissipative solitons in these lasers . his most important contributions concerned cavity solitons in vcsels , longitudinal modes dynamics , excitability and stochastic resonances in semiconductor lasers and the analysis of lasers with optical feedback . | we show experimentally and theoretically that localized pulses can be generated from an electrically biased @xmath0 m multi - transverse mode vertical - cavity surface - emitting laser .
the device is passively mode - locked using optical feedback from a distant resonant saturable absorber mirror and it is operated below threshold .
we observe multistability between the off solution and a large variety of pulsating solutions with different number and arrangements of pulses per round - trip , thus indicating that the mode - locked pulses are localized , i.e. mutually independent .
we show that a modulation of the bias current allows controlling the number of the pulses travelling within the cavity , thus suggesting that our system can be operated as an arbitrary pattern generator of 10 ps pulses and 1 w peak power .
mode - locking , broad - area lasers , vcsels |
let @xmath0 be a reiamannian 2-sphere . denote the area of @xmath0 by @xmath6 .
in this paper we consider the problem of slicing @xmath0 by short curves .
we start with the following isoperimetric problem : when is it possible to subdivide @xmath0 into two regions of relatively large area by a short simple closed curve ?
papasoglu @xcite used besicovitch inequality to show that there exists a simple closed curve of length @xmath7 subdividing @xmath0 into two regions of area @xmath8 .
a similar result was independently proved by balacheff and sabourau @xcite using a variation of gromov s filling argument .
on the other hand , consider the 3-legged starfish example on figure [ starfish ] . as it has been observed in @xcite , for any @xmath9
, if the tentacles are sufficiently thin and long , the length of the shortest simple closed curve subdividing @xmath0 into two regions of area @xmath10 can be arbitrarily large .
a. nabutovsky asked ( @xcite ) the following question : what is the maximal value of @xmath11 $ ] such that for some @xmath12 each riemannian 2-sphere of area @xmath13 can be subdivided into two discs of area @xmath14 by a simple closed curve of length @xmath15 ? our first result provides an answer for this question .
[ 1/3 area ] there exists a simple closed curve @xmath16 of length @xmath17 subdividing @xmath0 into two subdiscs of area @xmath18 . to prove theorem [ 1/3 area ] we obtain the following result of independent interest .
[ tree ] there exists a map @xmath19 from @xmath0 into a trivalent tree @xmath20 , such that fibers of @xmath19 have length @xmath17 and controlled topology : preimage of every interior point is a simple closed curve , preimage of every terminal vertex is a point and preimage of every vertex of degree @xmath21 is homeomorphic to the greek letter @xmath22 .
theorem [ tree ] follows from a more general theorem [ tree1 ] in section 3 for spheres with @xmath1 holes . using different methods
guth @xcite proved existence of a map from @xmath0 into a trivalent tree with lengths of fibers bounded in terms of hypersphericity of @xmath0 .
it follows from theorem 0.3 in @xcite that there exists such a map with fibers of length @xmath23 .
the second main result of this paper is about slicing @xmath0 by @xmath24cycles .
if instead of simple closed curves we allow subdividison by 1-cycles , we show that @xmath0 can be subdivided into two regions with arbitrary prescribed ratio of areas by a 1-cycle of length @xmath25 .
in fact , we prove the following [ morse ] there exists a morse function @xmath2 with fibers of length @xmath25 .
this improves the result of balacheff and sabourau @xcite that there exists a sweep - out of @xmath0 by @xmath24cycles of length @xmath26 .
theorem [ morse ] follows from a more general result in section 4 , where we prove existence of a morse function on a sphere with @xmath27 holes , such that the function is constant on each connected component of the boundary and the length of fibers are bounded in terms of area and boundary length .
alvarez paiva , balacheff and tzanev @xcite show that existence of a morse function on a riemannian 2-sphere with bounded fibers yields a length - area inequality for the shortest periodic geodesic on a finsler 2-sphere ( for both reversible and non - reversible metrics ) .
note that arguments used by croke @xcite to prove the length - area bound for the shortest closed geodesic on a riemannian 2-sphere ( see also @xcite for the best known constant ) can not be directly generalized to the finsler case because co - area inequality fails for non - reversible finsler metrics .
hence , theorem [ morse ] yields a better constant for theorem vi in @xcite .
[ finsler ] let @xmath0 be a finsler two - sphere with holmes - thompson area @xmath28 .
then @xmath0 carries a closed geodesic of length @xmath29 .
the reason for constants @xmath30 and @xmath31 in our theorems is the following .
we obtain the desired slicing of the sphere by repeatedly using the result of papasoglu @xcite to subdivide the sphere into smaller regions by a curve of length at most @xmath32 times the square root of the area of the region . at each step
the area of the region reduces at least by a factor of @xmath33 .
we then assemble these subdividing curves into one foliation with lengths bounded by the geometric progression @xmath34 in the proof of theorem [ morse ] some subdividng curves are used twice so an additional factor of @xmath35 appears .
after the first subdivision happens our regions are no longer spheres , but rather spheres with a finite number of holes
. it may not be possible to find a short simple closed curve subdividing it into two parts of area @xmath36 of its area .
instead we may have to use a collection of arcs with endpoints on the boundary subdividng the region into many pieces , each of small area .
the issue is then how to assemble all of these subdividing curves into one foliation .
the main technical result of this paper ( proved in the next section , see proposition [ subdivision ] ) is that we can always choose these subdividing arcs in such a way that they belong to a single connected component of the boundary of a certain subregion @xmath37 with area of @xmath37 between @xmath38 and @xmath33 of the area of the region .
this result make assembling curves into one foliation a straightforward procedure .
after the first version of this article appeared on the web , f. balacheff in @xcite improved constant @xmath30 in theorems 1 and 2 to @xmath39 , and constant @xmath40 in theorem 4 to @xmath41 , but not constant @xmath31 in theorem 3 .
* acknowledgements . *
the author is grateful to alexander nabutovsky for reading the first draft of this paper and to juan carlos alvarez paiva , florent balacheff , gregory chambers and regina rotman for valuable discussions .
the author would like to thank the referee for helpful comments .
the author acknowledges the support by natural sciences and engineering research council ( nserc ) cgs graduate scholarship , and by ontario graduate scholarship .
let @xmath0 be a riemannian 2-sphere
. for @xmath42 let @xmath43 denote the set of simple closed curves on @xmath0 that divide it into subdiscs of area @xmath44 . define @xmath45 and @xmath46 , where the supremum is taken over all metrics on @xmath47 of area @xmath13 . by definition @xmath12 is increasing and by proposition [ semicontinuity ] below it is lower semicontinuous . for any @xmath9 it follows from the example in the introduction ( see figure [ starfish ] ) that @xmath48 .
for @xmath49 we have the following result of papasoglu . [ papasoglu ] ( papasoglu , @xcite )
@xmath50 we will need to generalize theorem [ papasoglu ] to spheres with finitely many holes and allow a larger class of subdividing curves than just simple closed curves .
when the surface has boundary we will allow the subdividing curve @xmath16 to consist of several arcs with endpoints on the boundary . in this case we will define a distinguished connected component @xmath37 of @xmath51 and require that @xmath16 is contained in a connected component of @xmath52 .
this is a technical condition that will make it easer to repeatedly cut the surface into smaller pieces and concatenate the subdivding curves to obtain a slicing of @xmath0 .
let @xmath53 be a complete riemannian 2-surface with boundary homeomorphic to a sphere with @xmath27 holes .
let @xmath16 be a simple closed curve in the interior of @xmath53 or a union of finitely many arcs @xmath54 , where @xmath55 are arcs with endpoints on @xmath56 that do not pairwise intersect and have no self - intersections .
let @xmath57 be the set of connected components of @xmath58 .
let @xmath59 denote the set of all such @xmath16 on @xmath53 that in addition satisfy 1 .
@xmath60 2 .
@xmath16 is contained in a connected component of @xmath52 in particular , ( 1 ) implies that the area of every connected component of @xmath58 is bounded from above by @xmath61 .
define @xmath62 and @xmath63 , where the supremum is taken over all metrics on a sphere with @xmath27 holes that have area @xmath13 .
we have the following useful fact . [ semicontinuous ]
@xmath64 and @xmath12 are lower semi - continuous for @xmath65 $ ] .
we prove the result for @xmath64 and for @xmath12 it will follow as a special case from the argument below .
let @xmath66 be an increasing sequence converging to @xmath67 for some @xmath68 .
fix @xmath69 .
let @xmath53 be a complete riemannian surface of area @xmath13 diffeomorphic to a sphere with @xmath27 holes .
we would like to show that for some @xmath70 there exists @xmath71 with @xmath72 .
we can find @xmath73 small enough so that it satisfies the following requirements : 1 .
the area of the @xmath74-tubular neighbourhood of @xmath75 satisfies @xmath76 .
any ball @xmath77 around @xmath78 is bilipschitz diffeomorphic to the euclidean disc of radius @xmath74 with lipschitz constant between @xmath79 and @xmath80 . in particular , @xmath81 and @xmath82 .
3 . @xmath83 for such a @xmath74 choose @xmath70 so that @xmath84 .
let @xmath85 be the subdividing arcs of length @xmath86 . in the following argument it will be more convenient to consider @xmath87 , the connected component of @xmath52 that contains @xmath16 ( recall that by definition of @xmath64 @xmath16 lies in a single boundary component of @xmath52 ) .
if @xmath16 is a closed curve then @xmath88 . if @xmath16 is a union of arcs then @xmath87 is a closed curve made out of arcs of @xmath16 and arcs of @xmath75 .
let @xmath89 denote the element of @xmath90 of smaller area and @xmath91 denote the element of larger area .
we can assume that @xmath92 for otherwise we are done .
therefore , we have @xmath93 and @xmath94 .
let @xmath95 denote @xmath96 . by our choice of @xmath74
we have that the area of a ball @xmath97 for @xmath98 and @xmath99 . by fubini s theorem
we obtain @xmath100 .
hence , for some @xmath101 we have @xmath102 .
since @xmath103 is non - empty we can always find such a ball so that @xmath104 is non - empty .
we will now construct a new curve @xmath105 that coincides with @xmath87 outside of @xmath106 and divides @xmath53 into regions @xmath107 and @xmath108 so that one of them is connected and each of them has area @xmath14 .
moreover , @xmath109 .
this implies the desired inequality @xmath110 .
we construct @xmath105 by cutting @xmath87 at the points of intersection with @xmath111 and attaching arcs of @xmath111 .
we do it in such a way that @xmath77 is now entirely contained in the smaller of two regions .
this increases the area of the smaller region by at least @xmath112 and increases the length of the subdividing curve by at most @xmath113 .
the procedure is illustrated in figure [ semicontinuity ] .
let @xmath114 be connected components of @xmath115 and @xmath116 denote connected components of @xmath117 .
first we erase all arcs of @xmath87 that are in @xmath77 . for each @xmath118
we erase the arc @xmath119 from @xmath111 . for each @xmath120 ,
@xmath121 , we add a copy of @xmath120 and perturb it so that the new curve @xmath105 does not intersect @xmath77 in the neighbourhood of @xmath120 .
this does not increase the number of connected components of either region .
the only example that i know where upper semicontinuity of @xmath12 fails is the three - legged starfish on figure [ starfish ] ( showing @xmath48 for @xmath9 , while @xmath122 by theorem [ 1/3 area ] ) .
the following question seems natural : * question * is @xmath12 continuous for @xmath123 ?
it can be easily shown from the definition that @xmath124 . under the additional assumption
@xmath125 we are able to prove that they are equal . [
subdivision ] suppose @xmath125 , then @xmath126 . in the proof of proposition [ subdivision
] we will use the following topological fact .
[ 2-gon ] let @xmath0 be a submanifold ( with boundary ) of @xmath47 and let @xmath16 be a simple closed curve in @xmath47 .
suppose the intersection of @xmath16 and @xmath75 is non - empty and transversal .
if @xmath28 denotes a connected component of @xmath127 then there exists an arc @xmath128 and an arc @xmath129 , such that @xmath130 bounds a disc @xmath131 and @xmath132 is connected .
] the proof of this lemma is the main technical part of the paper .
the proof is illustrated on figure [ 2gon_fig ] .
the reader can draw several pictures like on figure [ 2gon_fig ] and convince him or herself that the statement is correct .
the strategy of the proof is to find some disc @xmath133 , with possibly a large number of components of @xmath134 .
we then show that there exists a subdisc of @xmath133 whose intersection with @xmath0 has a smaller number of connected components .
let @xmath135 and @xmath136 be the connected component of @xmath75 that contains @xmath137 .
define an interval @xmath138 to be a connected component of @xmath139 that has @xmath137 as an enpoint ( note that it is unique ) .
we denote the other endpoint of @xmath140 by @xmath141 .
let @xmath142 be a tangent vector to @xmath16 at @xmath137 pointing inside @xmath0 .
let @xmath143 be an arc of @xmath16 that starts at @xmath137 in the direction @xmath142 and ends at @xmath141 .
observe that the curve @xmath144 is a simple closed curve enclosing a disc @xmath145 that contains a non - empty subset of @xmath146 .
if @xmath147 has one connected component then we are done .
assume it has more than one .
let @xmath37 be a component of @xmath148 with @xmath149 and let @xmath150 be a different component . define a point @xmath151 by @xmath152 it follows from the definition that @xmath153 .
we can find a point @xmath154 and an arc @xmath155 from @xmath156 to @xmath157 .
the interior of @xmath158 is contained in the interior of @xmath159 .
it follows that @xmath160 .
denote the arc of @xmath143 between @xmath156 and @xmath157 by @xmath161 .
we have that @xmath162 separates @xmath47 into a disc @xmath163 that contains @xmath150 and its complement that contains @xmath37 .
set @xmath164 .
it is non - empty and has fewer connected components than @xmath148 .
we iterate this procedure until we are left with just one connected component .
we now prove proposition [ subdivision ] .
the direction @xmath124 is simple . given a riemannian 2-sphere we can make @xmath27 holes in it of small area and small boundary length .
we subdivide the resulting sphere with holes and use the fact that the connected component of @xmath52 that contains @xmath16 will have length close to @xmath64 .
here we did not use that @xmath165 . to prove the other direction we proceed as follows .
let @xmath166 be a small positive constant .
given a sphere with @xmath27 holes @xmath53 of area @xmath13 we attach @xmath27 discs of total area @xmath166 to the boundary of @xmath53 .
we obtain a riemannian 2-sphere @xmath0 .
let @xmath167 be a simple closed curve of length @xmath168 subdividing @xmath0 into two subdiscs @xmath28 and @xmath169 of area between @xmath170 and @xmath171 .
suppose @xmath28 is a subdisc of area @xmath172 .
if @xmath167 is disjoint from @xmath56 we conclude that @xmath173 .
suppose @xmath167 intersects @xmath56 .
by lemma [ 2-gon ] there exists an arc @xmath143 of @xmath167 and an arc @xmath140 of @xmath56 , such that @xmath174 is connected .
we consider two possibilities .
first , suppose @xmath175 .
let @xmath176 be the intersection of @xmath143 with the interior of @xmath53 .
@xmath37 is a connected component of @xmath177 of area between @xmath178 and @xmath179 .
the rest of connected components of @xmath177 have area less than @xmath179 .
so @xmath180 .
alternatively , suppose @xmath181 . in this case we can define a new curve @xmath176 , such that the number of connected components of @xmath177 is smaller than the number of connected components of @xmath182 .
we do this by replacing the arc @xmath143 of @xmath167 by @xmath183 .
note that we can slightly perturb the part of the new curve that coincides with @xmath140 so that it is entirely in @xmath184 , in particular , the intersection of @xmath176 with @xmath56 is transversal and the length of the intersection of @xmath176 with the interior of @xmath53 is smaller than that of @xmath167 . as a result we transferred the area of @xmath37 from @xmath28 to @xmath169 .
since @xmath185 , @xmath125 and @xmath186 we obtain that the area of each of the two discs @xmath187 is at least @xmath170 . in this way
we can continue reducing the number of connected components of @xmath188 until one of the subdiscs of @xmath189 contains only one connected component of @xmath53 or until we encounter the first possibility above .
by proposition [ semicontinuous ] we conclude that @xmath190 .
a map @xmath19 from @xmath53 to a trivalent tree @xmath20 is called a @xmath20-map if the topology of fibers of @xmath19 is controlled in the following way : the preimage of any point in an edge of @xmath20 is a circle , there exist @xmath27 terminal vertices @xmath191 , such that @xmath192 is a connected component of @xmath56 , the preimage of any other terminal point of @xmath20 is a point , and the preimage of a trivalent vertex of @xmath20 is homeomorphic to the greek letter @xmath22 .
[ tree1 ] for @xmath193 $ ] and any @xmath69 there exists a @xmath194map @xmath19 from @xmath195 , @xmath196 , so that each fiber of the map has length less than @xmath197 .
theorem [ tree ] follows by taking @xmath198 and applying theorem [ papasoglu ] . in the proof
we will repeatedly use the following simple fact , so it is convenient to state it as a separate lemma .
[ concatenation ] let @xmath37 and @xmath150 be two closed smooth submanifolds ( with boundary ) of @xmath195 , such that @xmath199 is a connected arc .
let @xmath200 denote the connected component of @xmath201 that contains @xmath202 .
suppose @xmath203 and that each @xmath204 admits a @xmath194map with fibers of length @xmath205 , then @xmath206 admits a @xmath194map with fibers of length @xmath207 . ] from the assumption that @xmath204 admits a @xmath194map it follows that there exists an embedded cylinder @xmath208 and a map @xmath209 $ ] with fiber @xmath210 and the length of all fibers @xmath205 .
the boundary @xmath211 with @xmath212 contained in the interior of @xmath204 .
let @xmath143 and @xmath161 be the endpoints of @xmath202 . for a sufficiently small @xmath73 we perform a surgery on the the closed curves in the foliation @xmath213 .
the surgery happens in the @xmath214neighbourhood of @xmath215 and is depicted on figure [ concat_fig ] . as a result of the surgery we obtain three families of curves .
the outer " family converging to @xmath216 , and two inner " families each converging to @xmath217 or @xmath218 .
these three families are separated by a @xmath22 graph ( drawn in red on figure [ concat_fig ] ) .
this surgery defines the desired @xmath194map .
we will need first a version of theorem [ tree1 ] for very small balls .
[ small length ] for any @xmath219 there exists @xmath220 , such that for every disc @xmath221 with @xmath222 there exists a diffeomorphism @xmath19 from @xmath223 to the standard closed disc @xmath224 so that the preimage of each concentric circle @xmath225 , has length @xmath226 . for @xmath73 sufficiently small every ball @xmath227 of radius @xmath74
is @xmath228bilipschitz diffeomorphic to a disc in the closed upper half - plane @xmath229 .
let @xmath230 denote the image of @xmath223 under such a diffeomorphism .
( here we are assuming that @xmath231 and @xmath223 is contained within a ball of radius @xmath74 ) .
after a small perturbation we may assume that the projection @xmath232 of @xmath233 onto @xmath234 coordinate is a morse function .
define a 1-dimensional simplicial complex @xmath235 as follows .
let @xmath236 be a regular value of the projection function @xmath237 restricted to the boundary of @xmath230 .
@xmath238 is a finite union of disjoint closed intervals @xmath239 .
we set @xmath240 to be the midpoints of @xmath241s .
if @xmath236 and @xmath242 are two consecutive critical values of @xmath232 , it follows that @xmath243 is a collection of disjoint simple arcs as on figure [ fig_small length ] . at a critical point
@xmath233 locally looks like the graph of a function @xmath244 . we connect the endpoints of the intervals of @xmath20 by a horizontal arc tangent to the critical point and contained in @xmath230 .
note that @xmath230 retracts onto @xmath20 , so in particular @xmath20 must be connected and simply connected , hence a tree .
we contract @xmath223 along the edges of @xmath20 in the obvious way . as a result
we obtain a contraction of @xmath245 inside @xmath223 to a point through curves of length @xmath226 . after a small perturbation we can assume that this homotopy realizes the desired diffeomorphism .
for details we refer the reader to @xcite , where it is shown that if there exists a homotopy of the boundary of @xmath223 to a point through curves of length @xmath246 , then there exists a diffeomorphism from @xmath223 to @xmath247 so that preimages of concentric circles have length @xmath246 . in @xmath230 ]
[ small area_s^2 ] for any @xmath219 there exists @xmath249 , such that for every disc @xmath221 with @xmath250 there exists a @xmath194map @xmath19 from @xmath223 with fibers of length less than @xmath251 .
the proof is similar to that of lemma 2.2 in @xcite .
choose @xmath252 , where @xmath253 is as in lemma [ small length ] .
the proof is by induction on @xmath254 . for @xmath255
the result follows by lemma [ small length ] .
assume the lemma to be true for all subdiscs with @xmath256 and consider the case when this quantity equals @xmath257 .
subdivide @xmath245 into 4 arcs of equal length . by besicovitch lemma
we can find an arc @xmath202 of length @xmath258 connecting two opposite arcs .
@xmath202 subdivides @xmath223 into two subdiscs @xmath259 and @xmath260 of area @xmath261 and boundary length @xmath262 . hence , by inductive assumption @xmath259 and @xmath260 admit @xmath194maps with fibers of length @xmath263 . by inductive assumption
each disc admits a @xmath194map with fibers of length @xmath264 .
we also have @xmath265 so the result follows by lemma [ concatenation ] .
we need to generalize this result about small discs to other small submanifolds of @xmath195 .
[ small area_mk ] for any @xmath219 there exists @xmath249 , such that for every submanifold with boundary @xmath266 , with @xmath267 there exists a @xmath194map @xmath19 from @xmath53 with fibers of length @xmath268 .
let @xmath53 be a closed submanifold ( with boundary ) of @xmath195 and let @xmath269 be a connected component of @xmath56 .
if @xmath270 then there exists an open ball @xmath271 of radius @xmath272 whose interior does not intersect the boundary of @xmath53 ( and , in particular , it does not intersect the boundary of @xmath195 ) . as @xmath273 the area of @xmath274 approaches the area of a euclidean disc of the same diameter .
since @xmath195 is compact , this happens uniformly for all balls of radius @xmath272 disjoint from the boundary . for @xmath275
sufficiently small we may conclude that @xmath276 .
hence , for a sufficiently small @xmath28 , if @xmath277 then the distance @xmath278 .
we attach @xmath279 arcs @xmath280 to the boundary of @xmath56 of total length less than @xmath281 and so that @xmath282 is connected and its complement in @xmath53 is homeomorphic to a disc .
denote this disc by @xmath223 .
consider the normal @xmath214neighbourhood @xmath283 of @xmath245 in @xmath53 ( see figure [ fig_small area ] ) for some small @xmath74 .
by lemma [ small area_s^2 ] the complement of @xmath283 in @xmath53 admits a @xmath20 map with fibers of length @xmath268 .
let @xmath202 be a short closed curve in @xmath283 which separates one connected component of @xmath56 from other connected components as on figure [ fig_small area ] .
curve @xmath202 separates @xmath283 into two regions .
let @xmath169 denote the region that contains only one connected component of @xmath56 .
by lemma [ concatenation ] we can extend the @xmath194map to @xmath284 . by chopping off connected components of @xmath56 and applying lemma [ concatenation ]
repeatedly we obtain the desired @xmath194map . ]
the proof of theorem [ tree1 ] proceeds inductively by cutting @xmath0 into smaller pieces until their size is small enough so that lemma [ small area_mk ] can be applied .
we assemble @xmath194maps on these smaller regions to obtain one map from @xmath0 with the desired bound on lengths of fibers . fix @xmath285 and let @xmath28 be as in lemma [ small area_mk ] .
let @xmath286 .
we claim that for every @xmath266 with @xmath287 there exists a @xmath194map with fibers of length @xmath288 when @xmath289 the inequality ( [ inductive statement ] ) is true by lemma [ small area_mk ] .
we assume it to be true for @xmath290 and prove it for @xmath291 . by proposition [ subdivision ]
there exists @xmath292 of length @xmath293 .
we have two possibilities . *
* @xmath16 is a simple closed curve . in this case @xmath16 separates @xmath53 into two regions @xmath294 and @xmath295 of area @xmath296 .
the number of connected components of @xmath297 is at most @xmath298 . by inductive assumption @xmath294 and
@xmath295 admit @xmath194maps into trees @xmath299 and @xmath300 respectively .
we construct @xmath20 by identifying the terminal vertex @xmath301 of @xmath299 with the terminal vertex @xmath302 of @xmath300 .
@xmath194map @xmath19 is defined by setting it equal to @xmath303 when restricted to @xmath304 .
a simple calculation shows that lengths of fibers of @xmath19 satisfy the desired bound . * case 2 .
* @xmath16 is a collection of arcs with endpoints on @xmath56 .
let @xmath305 be connected components of @xmath56 that intersect arcs of @xmath16 .
for a sufficiently small @xmath74 the normal neighbourhood @xmath306 is foliated by closed curves of length very close to @xmath307 , which are transverse to the arcs of @xmath16 .
in particular , each @xmath306 admits a @xmath194map with fibers of length @xmath308 .
let @xmath37 be the connected component of @xmath51 with area satisfying @xmath309 ( recall that it exists by definition of @xmath64 ) .
let @xmath169 denote @xmath310 and @xmath311 denote @xmath312 .
( see figure [ fig_horseshoe ] ) .
] let @xmath313 be the connected component of @xmath314 that contains @xmath315 .
we say that an arc @xmath202 of @xmath16 is a horseshoe if @xmath202 is in @xmath311 and the endpoints of @xmath202 lie on the same connected component of @xmath316 .
we will use the following simple observation .
[ horseshoe ] if @xmath16 contains no horseshoes , then there exists a connected component @xmath317 of @xmath316 , such that @xmath318 is connected .
let @xmath317 be a connected component of @xmath316 and suppose @xmath318 contains more than one interval .
let @xmath319 be an interval of @xmath320 and let @xmath236 and @xmath242 denote the endpoints of @xmath319 .
@xmath321 consists of two arcs , call them @xmath136 and @xmath322 .
let @xmath323 be a connected component of @xmath316 that intersects @xmath136 .
we claim that @xmath322 does not intersect @xmath324 . for
suppose it does , consider then a subarc @xmath325 of @xmath136 from @xmath236 until the fist point of intersection with @xmath324 and denote this point by @xmath143 .
similarly , denote by @xmath326 the subarc of @xmath322 from @xmath236 until the fist point of intersection with @xmath324 and call it @xmath161 .
let @xmath327 be an arc of @xmath324 from @xmath143 to @xmath161 , then @xmath328 is a closed curve separating @xmath311 into two connected components .
moreover , the point @xmath242 and points of @xmath329 belong to different connected components of @xmath330 .
this is a contradiction since @xmath169 is connected .
suppose the intersection @xmath331 is not connected .
we can then find a subarc @xmath332 that intersects @xmath324 only at the endpoints .
the number of connected components of @xmath316 that intersect @xmath333 is strictly smaller than the number of components that intersect @xmath136 . proceeding in this way we can find an arc of @xmath313 that has endpoints on @xmath334 , and its interior intersects only one connected component @xmath335 of @xmath316 .
it follows that @xmath336 is connected .
we can now construct the desired @xmath194map . by inductive assumption
there exists a @xmath194map on @xmath169 with fibers @xmath337 .
note that @xmath338 and the number of connected components @xmath339 of @xmath314 satisfies @xmath340 .
suppose first that @xmath341 is a horseshoe .
it separates @xmath311 into two connected components .
let @xmath91 denote the connected component that does not contain @xmath169 .
since @xmath342 , it admits a @xmath194map with the desired bound on length of fibiers . by lemma [ concatenation ]
we can extend the @xmath194map to @xmath343 .
inductively we extend the @xmath194map to a subset @xmath344 , such that @xmath345 and @xmath346 does not contain any horseshoes . by lemma [ horseshoe ] @xmath347
is connected for some @xmath348 . by lemma [ concatenation ]
we can extend the @xmath194map to @xmath349 .
we iterate this procedure until we extended the @xmath194map to a set @xmath350 , which contains @xmath306 for all @xmath348 .
the last step is to extend the @xmath194map to the whole of @xmath53 exactly as we did it in case 1 .
to finish the proof of theorem [ tree1 ] recall that @xmath351 as @xmath352 .
since @xmath28 can be chosen arbitrarily small theorem [ tree1 ] follows from ( [ inductive statement ] ) .
now we can prove theorem [ 1/3 area ] .
let @xmath0 be a riemannian 2-sphere and @xmath353 a @xmath194map with fibers of length @xmath17 .
if there are no trivalent vertices in @xmath20 we can find a short closed curve that separates @xmath0 into two halves of equal area . for each trivalent vertex @xmath354 let @xmath355
. denote the closed curve @xmath356 by @xmath357 .
assume that for every @xmath358 ( @xmath359 ) @xmath357 separates @xmath0 into two discs , s.t .
the area of the smaller disc is strictly smaller than @xmath360 .
let @xmath358 be such that the area of the smaller disc is maximized among all such curves .
let @xmath361 denote the smaller and @xmath362 denote the larger of the subdiscs of @xmath363 .
let @xmath364 denote the edge of @xmath20 adjacent to @xmath365 , such that @xmath366 .
we observe that the other two edges adjacent to @xmath365 are contained in @xmath367 , for otherwise it would follow that the area of the smaller disc is not maximal for @xmath357 .
we conclude that for some @xmath368 @xmath369 subdivides @xmath0 into two discs of area @xmath18 for otherwise it would again contradict our choice of @xmath357 .
in this section we prove the existence of a morse function @xmath19 from @xmath0 to @xmath370 with short preimages .
[ m - function def ] let @xmath0 be a manifold with boundary .
a function @xmath2 is called an @xmath371function if it is morse on the interior of @xmath0 , constant on each boundary component and maps boundary components to distinct points disjoint from the critical values of @xmath19 .
the proof of theorem [ morse1 ] proceeds along the same lines as the proof of theorem [ tree1 ] .
there are two main differences .
the fist difference is that we would like the function to be smooth ( in particular , curves in the foliation @xmath369 are not allowed to have corners ) and have singularities of morse type .
this is accomplished by a simple surgery described in lemma [ theta ] .
the second difference is that we would like to bound the length of the whole level set , not just of the individual connected components .
we will need one technical lemma similar to concatenation lemma [ concatenation ] .
let @xmath374 denote a union of three non - intersecting simple arcs @xmath375 with common endpoints .
@xmath22 subdivides @xmath195 into three regions @xmath376 , @xmath377 , @xmath378 , choosing the indices so that @xmath379 .
let @xmath380 denote a curve obtained by pushing @xmath381 inside @xmath382 by a small perturbation , so that it is contained in a small normal neighbourhood of @xmath22 , smooth and has length @xmath383 .
let @xmath384 denote the neighbourhood of @xmath22 bounded by curves @xmath380 ( see figure [ fig_theta ] ) .
let @xmath236 and @xmath242 denote the endpoints of @xmath375 .
since @xmath0 is orientable , the normal tubular neighbourhood of @xmath381 is homeomorphic to the cylinder .
let @xmath380 be the boundary component of the tubular neighbourhood that does not intersect @xmath399 , @xmath400 .
after a small perturbation in the neighbourhood of @xmath236 and @xmath242 we can assume that @xmath380 is smooth and there is a diffeomorphism @xmath401 from @xmath402 \times s^1 $ ] onto the region between @xmath381 and @xmath380 with @xmath403 .
let @xmath269 be the midpoint of @xmath404 .
we perform a straightforward surgery to the curves @xmath405 depicted on figure [ fig_theta2 ] . in the neighbourhood of
@xmath269 we can choose a coordinate chart so that curves in the new foliation are given by @xmath406 .
[ m - small area disc ] for any @xmath219 there exists @xmath249 , such that for every disc @xmath407 with @xmath250 there exists an @xmath371map @xmath19 from @xmath223 with fibers of length less than @xmath251 . as in the proof of lemma [ small area_s^2 ]
, we proceed by induction on @xmath254 .
however , the inductive assumption is now different .
we would like to show that for every @xmath69 a subdisc @xmath407 admits an @xmath371map with fibers of length @xmath408 assume the lemma to be true for all subdiscs with @xmath409 .
we take small tubular neigbourhood of @xmath245 in @xmath223 and foliate it by closed curves of length @xmath410 .
we subdivide the innermost curve @xmath16 into 4 arcs of equal length . by besicovitch lemma we can find an arc @xmath202 of length @xmath411 connecting two opposite subarcs of @xmath16 .
let @xmath412 be the neighbourhood of @xmath413 as in lemma [ theta ] .
note that @xmath414 has 3 connected components : one of them is @xmath245 and denote the other two by @xmath136 and @xmath322 , bounding subdiscs @xmath259 and @xmath260 respectively .
assume that @xmath259 is a disc of smaller area , hence @xmath415 .
since @xmath420 by inductive assumption @xmath259 admits an @xmath371map @xmath421 with fibers of length @xmath422 . after appropriately scaling @xmath421 on a small neigbourhood of @xmath136 and multiplying by @xmath423 if necessary we can assume that @xmath424 is the minimum point of @xmath421 .
furthermore , we scale and shift @xmath421 so that @xmath425 and @xmath426 is contained between @xmath13 and @xmath427 we now extend the @xmath371map to @xmath428 with fibers of length @xmath429 @xmath430 by inductive assumption @xmath260 admits an @xmath371function @xmath431 with fibers of length @xmath432 .
we modify @xmath431 so that it takes on its minimum at @xmath433 .
we can now extend @xmath19 to the whole of @xmath223 .
[ m - small area mk ] for any @xmath219 there exists @xmath249 , such that for every submanifold with boundary @xmath266 , with @xmath267 there exists an @xmath371map @xmath19 from @xmath53 with fibers of length @xmath436 . as in the proof of lemma [ small area_mk ]
we connect all components of @xmath56 with @xmath437 closed curves @xmath55 of total length @xmath438 and denote the union of @xmath55s and @xmath56 by @xmath313 .
denote the normal @xmath214neighbourhood of @xmath313 in @xmath53 by @xmath412 .
after a small perturbation we can assume that the boundary of @xmath412 is smooth .
the complement @xmath223 of @xmath412 in @xmath53 is a disc and so by lemma [ m - small area disc ] it admits an @xmath371map with fibers of length @xmath268 .
next we proceed as in the proof of lemma [ small area_mk ] extending the domain of the @xmath371function over connected components of @xmath56 one by one .
let @xmath439 be a collection of nested simple closed curves in @xmath412 and @xmath304 denote the subset of @xmath412 between @xmath440 and @xmath441 .
we require that @xmath442 , @xmath443 and for each @xmath348 @xmath304 contains exactly one connected component of @xmath56 .
we can then apply lemma [ theta ] to extend the @xmath371function from @xmath223 to @xmath294 and inductively to the whole of @xmath53 .
let @xmath294 and @xmath295 denote the two components of @xmath448 . in this case
there is an @xmath371function @xmath421 from @xmath294 onto @xmath449 $ ] and an @xmath371function @xmath431 onto @xmath450 $ ] .
hence , we obtain an @xmath371function @xmath451 $ ] satisfying the same bound on the length of the fibers . using the inductive assumption
we calculate that this length bound is exactly what we want . as in the proof theorem [ tree1 ]
we add a small collar around boundary components of @xmath195 that intersect @xmath16 .
we apply the inductive assumption to @xmath227 and extend this map to regions separated from @xmath169 by horseshoes " .
then we extend the map to collars whose intersection with @xmath314 is a connected arc .
we iterate this procedure until we extended the map to all of @xmath195 . by appropriately scaling the @xmath371functions obtained for each region we ensure the correct bound on the lengths of fibers . | we show that for every complete riemannian surface @xmath0 diffeomorphic to a sphere with @xmath1 holes there exists a morse function @xmath2 , which is constant on each connected component of the boundary of @xmath0 and has fibers of length no more than @xmath3 .
we also show that on every 2-sphere there exists a simple closed curve of length @xmath4 subdividing the sphere into two discs of area @xmath5 .
[ theorem]definition [ theorem]lemma [ theorem]proposition [ theorem]corollary |
double - well lattices @xcite provide an interesting , experimentally accessible tool to investigate quantum effects for atom numbers even down to less than six atoms @xcite .
three - well lattices and four - well lattices could be realised either by using subwavelength lattices @xcite or by adjusting the technique used in refs .
@xcite by using more lasers . the number of particles in one of the two wells of a double - well lattice can easily be measured by averaging over all double wells @xcite . for a single three - well potential , transport and interaction blockade of cold bosonic atoms
have been discussed in ref .
@xcite . in a single three - well potential , few particle dynamics could be observed by using the single - site addressability of refs .
@xcite .
one of the fascinating aspects of quantum mechanics are non - classical quantum states like the noon - states @xcite : @xmath0 where @xmath1 refers , e.g. , to @xmath2 particles being on the left and @xmath3 particles on the right of some barrier but noon - states can also be investigated in phase - space .
these states have also been called `` schrdinger cat''-states , e.g. , in refs . @xcite . like the spin - squeezed states of refs .
@xcite , such states are relevant to improve interferometric measurements @xcite . suggestions on how such interesting many - particle quantum superpositions might be obtained can be found , e.g. , in refs .
@xcite and references therein .
the focus of the present paper lies on the numeric creation of spatial two - particle noon - states in three - well potentials with a focus on experimental signatures to verify that indeed noon - states rather than statistical mixtures have been created . to achieve the noon - states via controlled quantum dynamics
, we suggest to periodically shake the three - well potential .
this approach to enhance tunnelling in a controlled way is based on photon - assisted tunnelling @xcite which was realised experimentally for bose - einstein condensates in optical lattices @xcite ( for the interpretation of the experimental data see also @xcite ) .
recent research in periodically driven systems include nonlinear landau - zener processes @xcite and transport of bound pairs @xcite .
the paper is organised as follows : in sec .
[ sec : model ] we introduce the hamiltonian used to model the periodically shaken three - well potential as well as our main aim , the two - particle noon - states .
section [ sec : simplified ] shows that for some parameters , the quantum dynamics of the periodically shaken system can be understood within a simplified model originally developed for double wells @xcite which we adopt for the present situation . in sec .
[ sec : how ] we introduce the two - particle noon - states and explain how they can be distinguished from statistical mixtures .
section [ sec : noon ] shows that periodic shaking can considerably increase the fidelity , i.e. , the quality of the quantum superpositions . in order to show that the quantum superpositions can indeed be distinguished from statistical mixtures , last but not least decoherence effects are included
for the description of ultracold atoms in a three - well potential , the three - site version of a many - particle hamiltonian @xcite originally developed in nuclear physics @xcite is used : @xmath4\left(\hat{c}_1^{\dag}\hat{c}_1^{\phantom\dag}-\hat{c}_3^{\dag}\hat{c}_3^{\phantom\dag}\right)\end{aligned}\ ] ] the operator @xmath5 annihilates ( creates ) a boson in well @xmath6 ; @xmath7 is the hopping matrix element , @xmath8 is the tilt between well 1 and well 3 and @xmath9 is the driving amplitude .
the interaction between a pair of particles in the same well is denoted by @xmath10 .
quantum dynamics beyond such models has been investigated , e.g. , by refs .
@xcite ) can indeed be valid to describe the dynamics of few particles in few well systems . ] .
a systematic sketch of the three - well potential is shown in fig .
[ fig : sketch ] . the three - well potential with initially two atoms in the middle well .
this situation is an experimentally realistic initial condition @xcite .
experimental investigations would , e.g. , be possible in three - well lattices @xcite ( cf . the double - well lattice potential with less than six atoms per lattice site of ref .
periodic shaking is currently established as an experimental tool to control the quantum dynamics @xcite . ]
a fock - basis is useful to describe @xmath11 atoms in the three - well potential : @xmath12 with @xmath13 particles in the right well , @xmath3 particles in the middle well and @xmath14 particles in the left well .
the aim of this paper is not to produce a single - particle quantum superposition but a two - particle noon - state : @xmath15 ( @xmath16 ) , the fidelity , @xmath17 or more precise the fidelity maximised over all possible angles @xmath18 is used : @xmath19 in order to be a useful flag to indicate a noon - state , the fidelity has to be larger than @xmath20 ; below this value , the state might not be a superposition at all as , e.g. , the state @xmath21 would also result in a fidelity of @xmath22 .
figure [ fig : exactdiagonal ] displays an example for which the fidelity is well above @xmath20 .
the data is plotted as a function of the dimensionless time , @xmath23 figure [ fig : ohnetreiben ] shows that for the timescales of fig .
[ fig : exactdiagonal ] and a large parameter range , even without shaking , the fidelity can reach values well above 50% ( fig .
[ fig : ohnetreiben ] ) but it stays again below 90% ( fig .
[ fig : ohnetreiben ] ) .
thus , so far there does not seem to be an advantage of using periodic shaking .
however , before showing that periodic shaking can lead to fidelities well above 95% , we show that it is possible to quantitatively understand the behaviour displayed in fig .
[ fig : exactdiagonal ] .
in order to understand the behaviour under periodic shaking , a simplified version of the hamiltonian ( [ eq : h ] ) can be used .
this approach introduced below will be useful for large shaking frequencies and not too long timescales @xcite . extending the simplified model of ref .
@xcite to a three - well potential starts with writing the schrdinger equation corresponding to the hamiltonian ( [ eq : h ] ) in the fock basis ( [ eq : fock ] ) . using the short notation of eq .
( [ eq : fochshort ] ) , we then introduce the ansatz motivated by the interaction picture @xmath24 \;,\ ] ] with @xmath25 .
the tunnelling between fock states is then included in a system of first - order one - dimensional differential equations for the @xmath26 ( cf . refs .
the time - dependence enters entirely via the exponents introduced in the ansatz ( [ eq : ansatz ] ) ; it includes terms linear in time ( either proportional to the interaction or to the tilt ) as well as a term proportional to the driving amplitude for which the time - dependence is a @xmath27 .
this last term can be transferred into a sum of exponentials with linear time - dependence via @xcite @xmath28 where the @xmath29 are bessel functions .
this leads to a series of relevant frequencies which for two particles read : @xmath30 the model is still equivalent to the full schrdinger equation corresponding to eq .
( [ eq : h ] ) ; it now reads : @xmath31 where @xmath32 in general , for given parameters of tilt , interaction and shaking frequencies , there will be only one term with slow time - dependence . following the reasoning of the rotating wave approximation @xcite ,
this term should dominate the dynamics for not too large timescales . defining @xmath33 where @xmath34 is the integer for which @xmath35 reaches its minimum and repeating the same reasoning for @xmath36 and @xmath37 leads to a much simpler version of the matrix in eq .
( [ eq : dglsa ] ) which surprisingly can even be solved via exact diagonalisation ) can , in general , only be solved by numerically integrating the time - dependent schrdinger equation . within this paper , we do this by using the shampine - gordon routine @xcite ; exact diagonalisation is a much simpler approach . ]
( see ref .
@xcite where even some analytical results could be derived ) .
the derivation of our simplified model explains [ cf .
( [ eq : simplified ] ) ] why the two curves in fig . [ fig : exactdiagonal ] agree well except for the deviation for larger times .
the main idea is to be able to distinguish quantum superpositions from statistical mixtures via the behaviour concerning the return to the initial state .
why a tilted three - well potential is useful for considerations like that can already be motivated on the single - particle level : on the one hand , despite the tilt the tunnelling processes @xmath38 do not involve any change of the potential energy . on the other hand , the individual tunnelling processes @xmath39 ) would thus be suppressed .
this offers a starting point to distinguish the time - evolution of quantum superpositions from statistical mixtures : on the one hand , tunnelling of the quantum superposition @xmath40 back to the initial state @xmath41 is allowed energetically . on the other hand , for a statistical mixture like @xmath42 the tunnelling to the state @xmath41
is suppressed for energetic reasons .
there is , however , an even stronger reason why the return to the initial state can verify interesting quantum superpositions : while experimentally measuring the fidelity ( [ eq : fidelity ] ) would either be impossible or at least be a considerable challenge for future experiments , measuring the number of atoms in one of the wells is comparatively straight - forward both in an optical superlattice @xcite and in a single three - well potential @xcite .
thus , for the two - particle noon - state @xmath43 one could experimentally verify , that with nearly 50% probability both particles are in the left well and with 50% probability both particles are in the right well . in order to distinguish this situation from the statistical mixture ,
@xmath44 we will be searching for parameters such that the probability to return to the initial state is high .
as the initial state is the fock state with two particles in the middle well , this is an unambiguous experimentally verifiable signature .
it remains to be shown that the return to the initial state at time @xmath45 in combination with the 50/50 probability to find both particles either in the left or in the right well at time @xmath46 really is unambiguous . using the time - evolution operator @xmath47 ,
one has : @xmath48 let us consider one particular ideal set of parameters for which the wave - function @xmath49 reaches the perfect cat - state ( [ eq : cat ] ) at @xmath50 followed by a probability to return to the initial state at @xmath51 of 100% : @xmath52 within this ideal set of parameters , an inaccurate preparation of the wave - function could lead to a return - probability of just @xmath53 : @xmath54
the fact that @xmath55 implies : @xmath56 thus , for our example of a perfect return to the initial state , a fidelity of @xmath53 at @xmath50 leads to a probability of @xmath53 to return to the initial state at @xmath51 .
in particular , if a noon - state has a 100% return probability to the initial state at a given time , the corresponding statistical mixture ( [ eq : corrmix ] ) will only manage a return to the initial state with 50% probability at the same point of time and @xmath57 . ] . while we will investigate one particular source of decoherence particle losses in the next section , other sources of decoherence
would also lead to statistical mixtures and thus to a different dynamics .
given the level at which today s experiments with ultra - cold atoms are performed , we can assume that the particles do mostly behave as predicted by the schrdinger equation without decoherence .
if the return to the initial state is less than the expected 100% ( but still well above 50% ) , this does not necessarily imply that the noon - state was partially destroyed via decoherence : experimental uncertainties which would average over different solutions of the schrdinger equation could produce a similar effect .
nevertheless , such a high return probability can still be a signature of the noon - state : it has to be checked if the variations of the parameters still lead to high fidelities at @xmath50 . in an experiment
this would lead to probabilities of the order of 50/50 to find the two particles either in the left or in the right well .
to obtain better results than displayed in figs .
[ fig : exactdiagonal ] and [ fig : ohnetreiben ] , three requirements were used for the numeric optimisation of the parameters : * a fidelity well above 90% . * a high probability to return to the initial state .
* not too large timescales .
there are parameter - sets with and without tilt that fulfil these conditions .
one example is shown in figs .
[ fig : fidgut ] and [ fig : dekohaerenz ] .
the approach via the simplified model ( [ eq : simplified ] ) only works if more than one frequency is included in eqs .
( [ eq : vielefrequenzen])-([eq : vielefrequenzen2 ] ) ( see fig . [
fig : fidgut ] ) . at first glance
, the standard deviation displayed in fig . [ fig : dekohaerenz ] might render it difficult to distinguish the behaviour including decoherence from the ideal quantum case .
however , for the accuracy with which the average can be measured , the standard error is relevant which is smaller by a factor of @xmath58 for @xmath59 measurements . taking @xmath60 this would thus reduce the error by a factor of 10 .
our theory so far ignores decoherence . for the present system , atom losses via scattering with back - ground atoms
are an important decoherence mechanism ( cf .
@xcite ) . to include this into our model
, we use a method called _ piecewise deterministic processes _ , pdp in ref .
@xcite , which is based on refs @xcite .
the main idea is to use quantum dynamics for the time - evolution in combination with atom losses for which the points of time of the atom losses are chosen via random numbers @xcite .
the results displayed in the middle panel of fig . [ fig : dekohaerenz ] are averaged over many sets of random numbers .
atoms are lost with a loss - rate @xmath61 ( cf .
[ sec : lindblad ] ) , @xmath62 corresponding to an exponential decay of the average number of particles , which for the two particles reads : @xmath63 the middle panel of fig .
[ fig : dekohaerenz ] uses @xmath64 with @xmath65 ; for the timescales of the double - well lattice experiment @xcite , the fidelity reaches its maximum at @xmath66ms and thus @xmath67khz . in order to be able to observe the return to the initial state with 1% accuracy , we require the average number of particles ( [ eq : nmittel ] ) to be at least 99% of the initial value up to a timescale for which pure quantum evolution predicts a nearly perfect return to the initial state ( @xmath68ms ) .
this results in a lifetime of the particles of @xmath69s .
this realistic lifetime of a single particle in the lattice lies two orders of magnitude below the lifetime required to produce noon - states of , say , 100 particles according to the scheme of ref .
@xcite .
the lower panel of fig .
[ fig : dekohaerenz ] shows what would happen if the quantum superposition was turned into a statistical mixture at the time for which the fidelity reaches its maximum : although now there still are two particles involved in the simulation , the average occupancy of the middle well would remain a factor of two below the values reached for pure quantum evolution . in order to reproduce the generation of noon - states shown in figs .
[ fig : fidgut ] and [ fig : dekohaerenz ] , an experiment would require a control of both interaction ( @xmath70 ) and driving amplitude @xmath71 of the order of several per cent , the tilt @xmath8 should be zero to about @xmath72 in order to reach fidelities above 0.95 . in an experiment
, there might be problems in realising a periodic shaking which is exactly zero for times @xmath73 and perfectly sinusoidal for times @xmath74 . to show that this is not necessarily a problem , fig .
[ fig : mean ] shows what happens if the shaking amplitude is switched on and off with @xmath75 as the characteristic timescale : @xmath76\nonumber\\ & \times & \left[1-\exp\left({\textstyle-\frac{2\pi(t - t)}{t_{\rm settl}}}\right)\right]\label{eq : settl}\;,\nonumber\\%\quad & & 0\le t\le t\,;\end{aligned}\ ] ] the shaking now includes a phase @xmath77 in @xmath78 .
plotted is the mean deviation from the case without any increase of the amplitude : @xmath79 \equiv \frac
1t\int_0^t \left|\langle n_2\rangle -\langle n_2\rangle_{\rm id}\right|^2 dt\ ] ] where @xmath80 is the model ( [ eq : h ] ) .
to conclude , we have shown via numerical investigations that high - fidelity - two - particle noon - states can emerge in three - well potentials .
there is , in particular , an easily observable difference between statistical mixtures and perfect noon - states : only the latter would return to the initial states . for realistic conditions , decoherence would be present in such experiments , which would destroy the quantum superpositions . here , the proposed scheme might offer the possibility to measure the timescales of decoherence .
three - well lattices could be used to realise such experiments - produced either via a superposition of lasers similar to the double - well lattices of refs .
@xcite or by using subwavelength lattices @xcite . the number of atoms per three - well lattice can be controlled precisely via the mott - insulator used to load the lattice @xcite .
such a setup would allow to measure the average number of particles of one of the three lattice sites in a single experiment by averaging over all three - well potentials .
furthermore , by focusing on one single three - well potential within the three - well lattice via techniques similar to refs .
@xcite and repeating the experiments many times , even more precise measurements could be possible .
we thank m. holthaus for his support and h .- p .
breuer , m. esmann and f. schreck for discussions .
bg acknowledges funding by the studienstiftung des deutschen volkes .
nt acknowledges funding by the university of oldenburg .
cw acknowledges support by c. & c. hettich in initiating this project .
a derivation of the lindblad master equation can be found , e.g. , in ref .
often , @xmath81 is the annihilation ( creation ) operator of a photon in mode @xmath82 but atom losses can be described by the same approach : @xmath83 -\frac{\kappa_-}2\left(a^{\dag}a\hat{\varrho}+\hat{\varrho } a^{\dag}a - 2a\hat{\varrho } a^{\dag}\right)\nonumber\\ & & -\frac{\kappa_+}2\left(aa^{\dag}\hat{\varrho}+\hat{\varrho}a a^{\dag } - 2a^{\dag}\hat{\varrho } a\right)\;,\end{aligned}\ ] ] where @xmath84 is the density matrix .
we assume that particles only are lost , thus @xmath85 and @xmath86 and thus derive a rate equation @xcite by using the fact that the photon number distribution @xmath87 is related to the density matrix via @xmath88 the rate - equation thus reads : @xmath89 using this rate - equation , it is straight - forward @xcite to derive the exponential decay of the numbers of particles given in eq .
( [ eq : nvont ] ) . as explained at the beginning of sec .
[ sec : noon ] , we use piecewise deterministic processes @xcite ( cf .
@xcite ) to model the atom losses ; in between loss events , the hamiltonian dynamics of eq .
( [ eq : h ] ) is used .
further theoretical investigations of atom losses can be found , e.g. , in ref .
@xcite .
57ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop _ _
( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) | few - particle dynamics in a three - well potential are investigated numerically .
it is shown that periodically shaking the potential can considerably increase the fidelity of emerging spatial quantum superpositions .
such noon - states are important for quantum interferometry .
if the two particles initially sit in the middle well , the probability to return to this state can distinguish pure quantum dynamics from statistical mixtures .
the numeric implementation of decoherence via particle losses shows clear differences from the pure quantum behaviour .
a three - well lattice could be an ideal system for experimental realisations . |
the first galaxies are believed to have formed within the first 300 - 400 myr of the universe and great strides have been made towards identifying objects within this era . since the installation of the wide field camera 3 ( wfc3 ) instrument on the hubble space telescope ( hst ) , an increasing number of candidates have been identified by means of their photometric properties , with @xmath5700 probable galaxies identified at @xmath217 - 8 ( @xcite : see also @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) and another 10 - 15 candidates identified even further out at @xmath219 - 11 ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
one of the most interesting questions to investigate with these large samples is the build - up and evolution of galaxies .
while these issues have long been explored in the context of fainter galaxies through the evolution of the @xmath22 lf , less progress has been made in the study of the most luminous galaxies due to the large volumes that must be probed to effectively quantify their evolution .
the entire enterprise of finding especially bright galaxies at @xmath9 has been limited by the availability of sufficiently deep , multi - wavelength near - infrared data over wide areas of the sky .
the most noteworthy such data sets are the ukidss uds program ( lawrence et al .
2007 ) , the ultravista program ( mccracken et al . 2012 ) , the 902-orbit candels program from the hubble space telescope ( grogin et al . 2011
; koekemoer et al . 2011 ) , the borg / hippies pure - parallel data set ( trenti et al . 2011 ; yan et al . 2011 ; bradley et al .
2012 ; schmidt et al .
2014 ; trenti 2014 ) , and the zfourge data set ( tilvi et al . 2013 ; i. labb et al .
2016 , in prep ) of these surveys , arguably the program with the best prospects for probing the bright end of the @xmath4 population would be the wide - area candels program.1 deg@xmath23 ) uds and ultravista programs have great potential to find large numbers of bright @xmath24 sources as demonstrated by the recent bowler et al .
( 2014 ) results ( see also bowler et al .
2015 ) , but may not yet probe deep enough to sample the @xmath25 galaxy population . ] the challenge with candels has been that it is only covered with particularly deep near - infrared observations from 1.2@xmath8 m to 1.6@xmath8 m , but lacks hst - depth @xmath3-band observations at 1.05@xmath8 m over the majority of the area .
deep observations at 1.05@xmath8 m are needed for the determination of photometric redshifts for galaxies in the redshift range @xmath26 to @xmath27 . while this can be partially compensated for by the availability of moderate - depth @xmath28 m observations from various ground - based programs over the candels program , e.g. ,
hugs @xcite , ultravista @xcite , and zfourge ( i. labb et al .
2016 , _ in prep _ ) , such observations are not available over the entire program , making it difficult to consider a search for bright @xmath4 galaxies over the full area . fortunately , there appears to be one attractive , alternate means for making use of the full candels area to search for bright @xmath4 galaxies . this is to exploit the availability of uniformly deep spitzer / irac observations over the full area ( e.g. , ashby et al .
2013 ) and redshift information present in the [ 3.6]@xmath6[4.5 ] colors of @xmath29 - 8 galaxies .
as demonstrated by many authors ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , the [ 3.6]@xmath6[4.5 ] colors appear to depend on redshift in a particularly well - defined way , a dependence which appears to arise from very strong nebular emission lines , such as h@xmath13 and [ oiii]@xmath305007 , which pass through the irac bands at particular redshifts .
for example , while @xmath216.8 galaxies have very blue [ 3.6]@xmath6[4.5 ] colors likely due to contamination of the [ 3.6 ] filter by [ oiii]+h@xmath7 lines ( and no similar contamination of the [ 4.5 ] band ) , @xmath317 galaxies exhibit much redder [ 3.6]@xmath6[4.5 ] colors , as only the 4.5@xmath8 m band is contaminated by the especially strong [ oiii]+h@xmath7 lines ( @xcite ; @xcite ; @xcite ) . here , we make use of the redshift information in the spitzer / irac observations and apply a consistent set of selection criteria to search for bright @xmath32 galaxies over all 5 candels fields .
a full analysis of the hst + ground - based observations is made in preselecting candidate @xmath24 galaxies , for further consideration with the available spitzer / irac data .
the identification of such bright sources allows us to better map out the bright end of the uv luminosity function ( lf ) at @xmath4 and constrain quantities like the characteristic luminosity @xmath33 or the functional form of the lf at @xmath4 .
@xcite only observe a modest ( @[email protected] mag ) brightening in the characteristic luminosity @xmath33 or bright end cut - off from @xmath32 to @xmath29 taking advantage of the full candels + xdf + hudf09-ps search area ( @xmath11000 arcmin@xmath23 ) .
@xcite also report evidence for a limited evolution in the characteristic luminosity with cosmic time , based on a wider - area search for @xmath34 - 7 galaxies found over the @xmath11.7 deg@xmath23 ultravista+uds area .
limited evolution was also reported by finkelstein et al .
( 2015 ) in subsequent work , but utilizing a @xmath13 - 15@xmath35 smaller area than bouwens et al .
( 2015 ) or bowler et al .
( 2015 ) had used .
this paper is organised as follows : 2 presents our @xmath215 - 8 catalogs and data sets , as well as methodology for performing photometry . 3 describes the selection criteria we define for our samples and methodology .
4 presents the results of our investigation and discusses the constraints added by @xmath3-band observations and keck / mosfire spectroscopy . in 5
, we use the present search results to set a constraint on the bright end of the @xmath4 lf . finally ,
6 includes a summary of our paper and a prospective . throughout this paper , we refer to the hst f606w , f814w , f105w , f125w , f140w , and f160w bands as _ _
v__@xmath36 , _ _
i__@xmath37 , _ _
y__@xmath38 , _
_ j__@xmath39 , _ _ jh__@xmath40 and _ _ h__@xmath18 , respectively , for simplicity .
we also assume _
_ h__@xmath41 = 70 km / s / mpc , @xmath42 0.3 , and @xmath43 0.7 .
all magnitudes are in the ab system @xcite .
cccccc candels gs deep & 64.5 & 27.8 & 27.5 & 26.1 & 25.9 + candels gs wide & 34.2 & 27.1 & 26.8 & 26.1 & 25.9 + ers & 40.5 & 27.6 & 27.4 & 26.1 & 25.9 + gs other & 31.8 + candels gn deep & 62.9 & 27.7 & 27.5 & 26.1 & 25.9 + candels gn wide & 60.9 & 26.8 & 26.7 & 26.1 & 25.9 + gn other & 34.0 + candels uds & 191.2 & 26.6 & 26.8 & 25.5 & 25.3 + candels cosmos & 183.9 & 26.6 & 26.8 & 25.4 & 25.2 + candels egs & 192.4 & 26.6 & 26.9 & 25.5 & 25.3 + total & 896.3
the sample of @xmath32 galaxies we identify in this paper is based on hst + ground - based observations that were acquired over 5 candels and ers fields ( @xcite ; @xcite ; windhorst et al .
2011 ) .
the near - ir hst observations over the candels fields range in depth from @xmath14 orbits over the @xmath1130 arcmin@xmath23 candels deep components in goods - north ( gn ) and goods - south ( gs ) to @xmath44 orbit depth over the @xmath1550 arcmin@xmath23 candels wide component in the gn , gs , uds , cosmos , and egs fields . over the gn and gs fields , the near - ir imaging observations are available in the @xmath45 , @xmath46 , and @xmath47 bands , while in the uds , cosmos , and egs fields , the near - ir observations are available in the @xmath46 and @xmath47 bands .
these fields also feature observations at optical wavelengths with the hst acs camera in the @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath52 bands for candels - gn+gs ( with 3 - 10 + orbits per band ) , as well as @xmath49 and @xmath51 observations ( @xmath12-orbit depth ) for the candels - uds+cosmos+egs fields .
in addition to the hst observations , these fields also have very deep ground - based observations from cfht , subaru suprime - cam , vlt hawk - i , and vista / vircam over the latter fields .
optical data are available in candels - cosmos field in the _ u _ , _ g _ , _ r _ , _ i _ , _ y _ and _ z _ bands as part of the cfht legacy survey , and also in the _ b _ , _ g _ , _ v _ , _ r _ , _ i _ and _ z _ bands from subaru observations over the same field ( @xcite ) . the candels - egs field is observed in the same bands as the cosmos field , as part of the cfht legacy survey , whilst the candels - uds field is observed by subaru as part of the subary xmm - newton deep field ( sxdf ) program ( furusawa et al .
2008 ) . for extended sources ,
these optical observations reach similar or greater depths to the available hst data over these fields ( i.e. , 26 mag to 28 mag at @xmath53 in @xmath54-diameter apertures : see bouwens et al .
2015 ) and allow us to exclude any potential lower redshift contaminants from our samples .
importantly , our ground - based observations also include moderately - deep ( @xmath126 mag at @xmath53 [ 1.2@xmath55-diameter apertures ] ) y - band observations which we use to constrain the nature of our selected @xmath4 candidates ( where hst observations are unavailable ) .
these observations are available over the candels - uds+cosmos fields through hawk - i and vista as part of the hugs ( @xcite ) and ultravista ( @xcite ) programs , respectively .
a more detailed description of the observations we utilize in constructing our source catalogs , as well as our procedure for constructing these catalogs is provided in @xcite ( see table 1 , figure 2 , and 3 from @xcite ) .
hst photometry was performed running the source extractor software @xcite in dual - image mode , taking the detection images to be the square root of the @xmath56 image @xcite and psf - matching the observations to the h@xmath18-band psf .
the colors and total magnitudes were measured with kron - like ( 1980 ) apertures and kron factors of 1.6 and 2.5 respectively .
photometry on sources in the ground - based data is performed after the contamination from foreground sources is removed , using an automated cleaning procedure ( @xcite ; @xcite ) .
the positions and two - dimensional spatial profiles of the foreground sources are assumed to match that seen in the high - spatial resolution hst images , after psf - matching to the ground - based observations .
the total flux in each source is then varied to obtain a good match to the light in the ground - based images .
light from the foreground sources is subsequently subtracted from the images , before doing photometry on the sources of interest .
flux measurements for individual sources are then performed in 1.2@xmath55-diameter circular apertures due to the objects being inherently unresolved in the ground - based observations .
these flux measurements are then corrected to total , based on the model flux profiles computed for individual sources based on the observed psfs .
the procedure we employ here to derive fluxes is very similar to that employed in skelton et al .
( 2014 : see also galametz et al .
2013 and guo et al .
2013 who have also adopted a similar procedure for their ground - based photometry ) .
the detailed information we have on @xmath34 - 9 galaxy candidates over the candels fields from hst is nicely complemented in the mid - ir by the spitzer extended deep survey ( seds , pi : fazio ) program @xcite , which ranges in depth from 12 hours to @xmath57100 hours per pointing , though 12 hours is the typical exposure time .
the seds program provides us with flux information at @xmath58 m and @xmath59 m , which can be useful for probing @xmath34 - 9 galaxies in the rest - frame optical , quantifying the flux in various nebular emission lines , and estimating the redshift . over the goods - north and goods - south fields ,
we make use of spitzer / irac reductions , which include essentially all the spitzer / irac observations obtained to the present ( labb et al .
2015 : but see also ashby et al .
2015 ) , with 50 - 200 hours of observations per pixel in both bands ( and typically @xmath1100 hours ) .
our procedure for performing photometry on the irac data is essentially identical to that used on the ground - based observations , except that we utilize @xmath60-diameter circular apertures for measuring fluxes .
these fluxes are then corrected to total based on the model profile of the individual sources + the psf .
depending on the size of the source , these corrections range from @xmath12.2@xmath35 to 2.4@xmath35 .
the median @xmath53 depths of these spitzer / irac observations for a @xmath126-mag source is 25.5 mag in the @xmath58 m band and 25.3 mag in the @xmath59 m band .
many recent studies ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) have presented convincing evidence to support the presence of strong nebular line contamination in photometric filters , particularly for the spitzer / irac [ 3.6 ] and [ 4.5 ] bands .
the observed [ 3.6]@xmath6[4.5 ] irac color of galaxy candidates appear to be strongly impacted by the presence of these lines at different redshifts , in particular those of h@xmath13 and [ oiii ] .
figure [ fig : color ] provides an illustration of the expected dependence of the spitzer / irac @xmath61-[4.5]$ ] color , as a function of redshift , assuming an [ oiii]+h@xmath7 ew ( rest - frame ) of @xmath12250 , which is at the high end of what has been estimated for galaxies at @xmath62 ( @xcite ; @xcite ; @xcite ) .
the significant change in the @xmath61-[4.5]$ ] color of galaxies from @xmath34 - 7 to @xmath9 suggests this might be a promising way of segregating sources by redshift and in particular to identify galaxies at @xmath9
. such information would be especially useful for search fields like candels egs , which lack deep observations in @xmath3-band at @xmath11.1@xmath8 m to estimate the redshifts directly from the position of the lyman break .
smit et al .
( 2015 ) have shown that selecting sources with blue @xmath61-[4.5]$ ] colors can effectively single - out sources at @xmath63 - 6.9 over all candels fields , even in the absence of @xmath3-band coverage . here
we attempt to exploit this strong dependence of the @xmath61-[4.5]$ ] color on redshift to identify some of the brightest @xmath9 galaxies over the candels fields . in performing this selection , we start with the source catalogs derived by bouwens et al .
( 2015 ) and skelton et al .
( 2014 ) over a @xmath1900 arcmin@xmath23 region from the five candels fields . in general
, we will rely on the source catalogs from bouwens et al .
( 2015 ) where they exist ( covering a 750 arcmin@xmath23 area or @xmath183% of candels ) .
otherwise , we will rely on the skelton et al . ( 2014 ) catalogs and photometry .
we then apply color criterion to identify a base sample of lyman - break galaxies at @xmath26 - 9.0 . in particular , over the candels - uds , cosmos , and egs fields , we use a @xmath64 criterion . over the candels - gn and gs fields
, we require that sources satisfy one of the two color criteria defined by eq .
[ eq:2 ] or eq .
[ eq:3 ] : @xmath65 @xmath66 these color criteria are essentially identical to those from bouwens et al .
( 2015 ) but allow for @xmath67 colors as red as 0.5 mag to match up with the color criteria of oesch et al .
( 2014 ) and bouwens et al .
( 2015 ) in searching for @xmath68 galaxies ( i.e. , @xmath69 ) . in so doing , it was our goal to maximize the completeness of our selection for bright @xmath70 - 9 galaxies within the candels program .
these color criteria are motivated in figure 3 of bouwens et al .
( 2015 ) and result in a very similar redshift segregation as one achieves using photometric redshifts .
we require that sources have @xmath61-[4.5]$ ] colors redder than 0.5 mag ( see figure [ fig : color ] ) .
this color criterion was chosen ( 1 ) so as to require slightly redder colors than was the average color measured by labbe et al .
( 2013 ) for their faint @xmath32 sample from the hudf ( i.e. , @xmath10.4 ) and ( 2 ) such that sources would not easily satisfy the criterion simply due to noise ( requiring @xmath572@xmath16 deviations for the typical source ) . to be certain that the irac colors we measured are robust
, we exclude any sources where the subtracted flux from neighboring sources exceeds 65% of the original flux in a @xmath60-diameter aperture ( before subtraction ) .
to ensure our selection is free of @xmath71 galaxies , we required that sources show no statistically significant flux at optical wavelengths .
sources that show at least a @xmath72 detection in terms of the inverse - variance - weighted mean @xmath49 and @xmath51 flux with hst were excluded .
in addition , we also excluded sources detected at @xmath73 in the deep optical imaging observations available over each field from the ground .
we adopted a slightly less stringent threshold for detections in the ground - based observations , due to the impact of neighboring sources on the overall noise properties .
finally , we consider potential contamination by low - mass stars , particularly later @xmath74 and @xmath3 dwarfs ( t4 and later ) , where the @xmath61-[4.5]$ ] color can become quite a bit redder than 0.5 mag ( kirkpatrick et al .
2011 ; wilkins et al .
2014 ) . to exclude such sources from our samples , both the spatial information we had on each source from the sextractor stellarity parameter and total sed information were considered .
sources with measured stellarities @xmath570.9 were identified as probable stars ( where 0 and 1 corresponds to extended and point - like sources , respectively ) , as were sources with measured stellarity parameters @xmath570.5 if the flux information we had available for sources was significantly better fit ( @xmath75 ) with a low - mass stellar model from the specx prism library ( burgasser et al .
2004 ) than with the best - fit galaxy sed model , as derived by the easy and accurate zphot from yale ( eazy ; @xcite ) software . our sed fits with eazy considered both the standard sed templates from eazy and sed templates from the galaxy evolutionary synthesis models ( galev ; @xcite ) .
nebular emission lines , as described by @xcite , were added to the galev sed template models assuming a 0.2@xmath76 metallicity .
no sources were removed from our selection as probable low - mass stars .
the procedure we use here to exclude low - mass stars from our selection is identical to that utilized by @xcite .
cccccccc cosy-0237620370 & 10:00:23.76 & 02:20:37.00 & [email protected] & [email protected] & 7.14@xmath77 & @[email protected] & [ 1],[2],[3 ] + egs - zs8 - 1 & 14:20:34.89 & 53:00:15.35 & [email protected] & [email protected] & 7.92@xmath78 & [email protected] & [ 3 ] , [ 4 ] + egs - zs8 - 2 & 14:20:12.09 & 53:00:26.97 & [email protected] & [email protected] & 7.61@xmath79 & [email protected] & [ 3 ] + egsy-2008532660 & 14:20:08.50 & 52:53:26.60 & [email protected] & [email protected] & 8.57@xmath80 & before applying the selection criteria from 3.1 to the @xmath1900-arcmin@xmath23 candels + ers search fields , it is useful to first test these criteria on those data sets which feature deep @xmath20 and @xmath3-band observations .
the availability of observations at these wavelengths , together with observations at both redder and bluer wavelengths with hst , allows for very accurate estimates of the redshifts for individual sources .
there are five data sets that possess these observations : ( 1 ) candels goods - s , ( 2 ) candels goods - n , ( 3 ) ers , ( 4 ) candels uds , and ( 5 ) candels cosmos field . the first three feature these observations with hst and the latter two using ground - based telescopes .
we apply selection criteria from the previous section to a @xmath47-band limiting magnitude of 26.7 mag for the first three fields and 26.5 mag for the latter two .
our decision to use these depths is partially guided by the sensivity of the spitzer / irac data over these fields . applying the selection criteria from the previous section to the candels gn+gs and ers fields ( @xmath81 ) ,
we find 7 sources that satisfy our selection criteria .
for each of these sources , we estimate photometric redshifts with eazy . in fitting to the observed photometry
, we used the same standard eazy sed templates as we described in the previous section .
we also applied the above selection criteria to the candels - uds and candels - cosmos fields , where it is also possible to estimate photometric redshifts , making use of the available hst observations and ground - based optical and near - ir @xmath3 and @xmath82 band observations .
eight sources satisfy these criteria .
all 15 of the sources selected using the criteria from the previous section are presented in figure [ fig : validation_of_method ] and fall between @xmath83 and @xmath84 , which is the expected range if a high - ew [ oiii]+h@xmath7 line is responsible for red @xmath61-[4.5]$ ] colors in these galaxies .
this suggests that the criteria we propose in the previous section can be effective in identifying a fraction of @xmath9 galaxies that are present in fields with deep hst+spitzer observations .
the individual coordinates , colors , and estimated redshifts for individual sources from this validation sample can be found in table [ tab : valid_samp ] located in appendix b. in recommending the use of the irac photometry to subdivide @xmath34 - 9 samples by redshift , we should emphasize that the most robust results will be obtained making use of only those sources with the smallest confusion corrections .
while we took care in the selection of both our primary sample ( and the sample we used to validate the technique ) to avoid such souces , such sources were not excluded in making figure 1 of smit et al .
( 2015 : resulting in a few @xmath4 sources with anomalously blue spitzer / irac colors ) . despite this issue with smit et al .
( 2015 ) figure 1 , we emphasize that this is nevertheless not a major concern for sources in their @xmath85 - 6.9 sample . only 2 of the 15 sources in the latter sample were subject to a @xmath13@xmath35 correction for flux from neighboring sources and those 2 sources ( gsd-2504846559 and egs-1350184593 ) are flagged as less reliable .
here we focus on the identification of only the brightest @xmath86 @xmath9 galaxies using our spitzer / irac color criteria .
this is to keep the current selection small and to focus on sources whose surface density was particularly poorly defined from previous work .
prior to this work , the only study which identified such bright @xmath32 sources was bouwens et al .
focusing on the brightest sources is also valuable , since it allows us to obtain very precise constraints on sed shapes and its spitzer / irac colors of the sources , as well as providing opportunities for follow - up spectroscopy ( see 4.2 ) .
applying the selection criteria described in
[ subsec : method ] on the candels - gs , candels - gn , candels - uds , candels - cosmos and candels - egs fields , we identify a total of 4 especially bright ( @xmath86 ) candidate @xmath9 galaxies .
our 4 candidate @xmath317 galaxies are presented in table [ tab : table_details ] and in figure [ fig : postage_stamp ] .
we see from figure [ fig : postage_stamp ] that each candidate is clearly visible in the hst _
_ h__@xmath18 and _ _ j__@xmath39 filters , as well as the irac 3.6 @xmath8 m and 4.5 @xmath8 m bands . of course
, no significant detection is evident in the hst _
_ v__@xmath36 and _ _ i__@xmath37 bands for these sources .
this would suggest that these sources show a break in their spectrum somewhere between 0.9@xmath8 m and @xmath87 m and therefore have redshifts between @xmath34 and @xmath27 .
[ we discuss the impact of information from @xmath3-band observations available over 3 of the 4 candidates in 4.1 .
] 3 of these 4 bright sources are found in the candels - egs field .
sources from this field were not included in our earlier attempt to validate the present selection technique ( 3.2 ) , so only one of these new sources is in common with the 15 sources just discussed . to derive constraints on the redshift of each bright source , we again made use of eazy .
the photometry provided to eazy included fluxes from hst filters , irac 3.6 @xmath8 m and 4.5 @xmath8 m filters and ground based telescopes .
using eazy allows us to generate a best - fit sed of each galaxy candidate as well as its redshift likelihood distribution ( @xmath88 ) , which we present in figure [ fig : sed_pz ] , with the observed galaxy flux points overplotted . from the sed plots , we observe a near - flat rest - frame optical continuum , as well as emission lines dominating at the location of high flux points , highlighting the contribution of strong nebular emission lines to the instrument filters .
one of our @xmath9 candidates , i.e. , egs - zs8 - 2 , is sufficiently compact , as can be seen from figure [ fig : postage_stamp ] , that we considered the possibility that it may correspond to a star .
to test this possibility , we compare its sed to all the stellar seds in the specx prism library and find the best - fitting stellar sed . the @xmath89 goodness - of - fit for the stellar sed
is an order of magnitude greater than the galaxy sed .
in addition , the sextractor stellarity we measure for egs - zs8 - 2 in the @xmath46 and @xmath47 bands is 0.60 and 0.33 ( where 0 and 1 correspond to an extended and point source , respectively ) , which significantly favors egs - zs8 - 2 corresponding an extended source .
bouwens et al .
( 2015 ) ran an extensive number of end - to - end simulations to test the possibility that point - like sources could scatter to such low measured stellarities .
stellarities of @xmath10.60 are only found for @xmath90-mag point - like sources in @xmath912% of the simulations that bouwens et al .
( 2015 ) run .
therefore , both because of the spatial and spectral information , we can be confident that the egs - zs8 - 2 candidate is a @xmath9 galaxy and not a low - mass star .
however , as we show in 4.2 , perhaps the most convincing piece of evidence for this source corresponding to a @xmath4 galaxy is our discovery of a plausible 4.7@xmath16 ly@xmath13 line in the spectrum of this source at 1.031@xmath8 m ( figure [ fig : pascal_lya ] ) . a second candidate from our selection , cosy-0237620370 , is also very compact and could potentially also correspond to a low - mass star .
however , like egs - zs8 - 2 , the photometry of the source is better fit with a galaxy sed than a stellar sed ( with a @xmath92 ) and the source shows evidence for spatial extension , with a measured stellarity of 0.81 and 0.34 in the @xmath46 and @xmath47 bands , respectively .
stellarities even as high as 0.81 are only recovered in @xmath15% of the end - to - end simulations bouwens et al .
( 2015 ) run at @xmath93-band magnitudes of @xmath125.0 .
earlier , tilvi et al .
( 2013 ) obtained the same conclusion regarding this source based on medium - band observations over this candidate from the zfourge program , where consistent fluxes are found in the near - infrared medium bands strongly arguing against this source corresponding to a low - mass star .
bowler et al .
( 2014 ) also conclude this source is extended and not a low - mass star , based on its spatial profile ( see figure 6 from bowler et al .
2014 ) and based on its observed photometry where @xmath94 . flux information from hst , spitzer / irac , and ground - based observations all have value in constraining the redshifts of the candidate @xmath9 galaxies we have identified in the present probe .
while the hst flux information we have available for all three candidates in the @xmath95 bands only allow us to place them in the redshift interval @xmath96 - 9.0 ( _ blue line _ in figure [ fig : pz_subset ] ) , we can obtain improved constraints on the redshifts of the candidates incorporating the flux information from spitzer / irac and from deep ground - based observations .
each of these three candidates appears to have redshifts robustly between @xmath97 and @xmath98 ( _ red and black lines _ in figure [ fig : pz_subset ] ) .
in addition , as we discuss in 4.1 and show in figure [ fig : pz_subset ] , the availability of the @xmath3/@xmath45-band observations allows us to significantly improve our redshift constraints on all three candidates .
we used the bouwens et al .
( 2015 ) catalogs to search 83% of the total area of the candels fields and the skelton et al .
( 2014 ) photometric catalogs otherwise ( in those regions over the wfc3/ir candels fields which lack the deep hst / acs data ) . as a check on the search
results we obtained with the bouwens et al .
( 2015 ) catalogs , we applied the same selection criteria to the skelton et al .
( 2014 ) catalogs .
encouragingly , we identified 75% of our sample , with only one candidate missing due to its having a @xmath61-[4.5]$ ] color of 0.47 mag .
for all 4 candidates from our primary sample , we find that our derived [ 3.6]@xmath6[4.5 ] colors are almost identical to those quoted by skelton et al .
( 2014 ) , agreeing to @xmath990.1 mag ( and typically @xmath100[3.6]@xmath6[4.5 ] of 0.05 mag ) .
we also identified 1 additional bright @xmath101 ) @xmath9 candidate in the candels - egs field not identified in our primary search ( see appendix a ) .
it seems clear examining its photometry that this source is extremely likely to be at @xmath62 - 9 ( and indeed it appears in the bouwens et al . 2015 @xmath32 sample ) .
however , since its measured @xmath61-[4.5]$ ] color is [email protected] mag in our photometric catalog ( 0.3 mag bluer than in the skelton et al . 2014 catalog )
, we did not include it in our primary sample .
we remark that photometry for this source was more challenging due to its being located close to a bright neighbor and its being a two - component source . for very high redshift sources ( @xmath102 )
, it is expected that the sources with the brightest apparent magnitudes will benefit from gravitational lensing ( wyithe et al .
2011 ; barone - nugent et al . 2015 ; mason et al . 2015 ; fialkov & loeb 2015 ) , and indeed it is found that a small fraction of the brightest galaxies identified over the candels program are consistent with being boosted by gravitational lensing ( barone - nugent et al .
2015 ) . to investigate whether any of the bright @xmath9 galaxies identified in our search might be gravitationally lensed , we considered all sources within 5@xmath55 of our candidates in the skelton et al .
( 2014 ) catalogs and used the estimated redshifts , stellar masses , and sizes from these catalogs to derive einstein radii for the foreground sources assuming a single isothermal sphere model .
we then calculated the degree to which our bright @xmath9 galaxy candidates might be magnified by the foreground sources . in only one case
was the expected magnification level @xmath5710% and this was for our @xmath98 candidate egsy-2008532660 . in this case , we identified two foreground galaxies which could significantly magnify this candidate ( figure [ fig : lens ] ) .
the first was a @xmath103 @xmath104 mass , @xmath105 galaxy ( 14:20:08.81 , 52:53:27.2 ) with a separation of 2.8@xmath55 from our @xmath98 candidate .
the second was a @xmath106 @xmath104 mass , @xmath107 galaxy ( 14:20:08.37 , 52:53:29.1 ) with a separation of 2.7@xmath55 from our @xmath98 candidate . using the measured size of the two sources ,
we derive @xmath108 km / s and @xmath109 km / s for the velocity dispersion . we checked and these velocity dispersions are fairly similar to what fitting formula in mason et al .
( 2015 ) yield ( i.e. , using the relation in their table 1 and applying @xmath47-band or irac 3.6@xmath8
m apparent magnitudes depending on whether we are considering the @xmath105 or @xmath107 source ) .
based on the observed separation of this source from our @xmath98 galaxy , we estimate a lensing magnification of 20% and a factor of 1.8 from the former and latter foreground sources . in computing these magnification factors , we assume that the mass profile of galaxies is an isothermal sphere and taking the magnification factor to be @xmath110 where @xmath111 is separation from the neighboring sources and @xmath112 is the einstein radius .
looking at the morphology of egsy-2008532660 , we see no clear evidence to suggest that the galaxy is highly magnified and there is no obvious counterimage .
however , we clearly can not rule out smaller lensing amplification factors , particularly if the intrinsic size of the source is small . as the inferred stellar or halo masses for the neighboring galaxies is not precisely known ,
this translates into a modest uncertainty into the actual luminosity of this source ( as much as 0.3 dex ) .
given this fact , we consider it safest for us to exclude it from analyses of the @xmath22 lf .
here we attempt to determine the nature of the @xmath4 candidates we selected using the hst+spitzer / irac+ground - based observations using some @xmath3-band observations that became available over a few of our candidates and using the results of some follow - up spectroscopy that we performed ( first reported in oesch et al .
2015b ) .
deep observations at 1.05@xmath8 m are particularly useful in ascertaining the nature of these candidates and also their redshift , due to the @xmath3-band photometry providing constraints on the position of the lyman break as it redshifts from @xmath87 m to 0.9@xmath8 m .
deep observations at @xmath28 m are available for 3 of the 4 @xmath9 candidates that we selected as part of our @xmath86 sample .
@xmath3-band observations of the cosy-0237620370 candidate are available from the 3-year ultravista observations ( mccracken et al .
2012 ) , while hst @xmath45-band observations are available over 2 other candidates in our selection as a result of some recent observations from the z9-candels follow - up program ( bouwens 2014 ; bouwens et al .
2016).-10 galaxies over the candels - uds , cosmos , and egs fields . in some cases , bright candidate @xmath32 galaxies were located nearby bright @xmath113 - 10 candidates and could be readily observed in the same pointings . ]
we make these estimates in an identical way to what we did previously .
our redshift constraints , including the @xmath3-band , are presented in table [ tab : table_details ] .
furthermore , we present the hst _ _ y__@xmath38 filter images in figure [ fig : postage_stamp ] , where we observe a clear detection in the _ _ y__@xmath38 filters for egs - zs8 - 1 and egs - zs8 - 2 and no detection in the _ _ v__@xmath36 or _ _ i__@xmath37 filters , indicating a _ _ z__@xmath17 lyman dropout . for egs - zs8 - 1 ,
however , we observe little to no detection in the _ _ y__@xmath38 filter but a clear detection in the _
_ j__@xmath39 filter which indicates this galaxy is observed at _ _ z__@xmath18 .
figure [ fig : sed_pz ] and [ fig : pz_yband ] presents the redshift likelihood distributions on our @xmath9 candidates , incorporating the @xmath3-band observations from ultravista and hst .
it is evident from figure [ fig : pz_subset ] that the @xmath3-band data greatly improves our constraints on the redshift of the individual candidates in our selection .
together with the results in 3.2 and figure [ fig : validation_of_method ] , these results largely validate our selection technique .
in addition to using photometric data in the y - band to validate our method , we also tested this method by obtaining deep near - ir spectroscopy on 2 sources from the current selection .
oesch et al .
( 2015b ) already provided a first description of the observational set - up we utilized for half of our targets , so we keep the current discussion short .
a total of 4 hours of good y - band spectroscopy were obtained in the candels - egs field with the multi - object spectrometer for infra - red exploration ( mosfire : mclean et al .
2012 ) instrument on the keck i telescope .
two masks ( see fig . 2 of * ? ? ?
* ) were utilized and our spectra were taken with 180 s exposures at a spectral resolution of r=3500 and r=2850 ( for a 0.7 and 0.9 slit respectively ) over 3 nights ( april 18 , april 23 , april 25 , 2014 - although due to poor weather conditions , april 18 was effectively lost ) , with the aim of searching for ly@xmath13 emission in egs - zs8 - 1 and egs - zs8 - 2 .
each mask contains a slitlet placed on a star , which we use for monitoring the sky transparency and observing conditions of each exposure .
these observations were reduced using a modified version of the drp mosfire reduction code pipeline ( for details see oesch et al .
the spectra complement the photometric data sets for these two galaxies and allow us to confirm their redshifts .
the observations carried out with keck / mosfire revealed candidate ly@xmath13 emission lines in the spectra of both egs - zs8 - 1 and egs - zs8 - 2 .
the detection of a ly@xmath13 line for egs - zs8 - 1 appears to be robust ( a 6.1@xmath16 detection with a line flux of @xmath114 erg s@xmath115 ) and places that source at @[email protected] , as first reported by oesch et al .
( 2015b ) .
( e.g. , finkelstein et al .
this is in part due to the significantly poorer seeing conditions we were subject for the observations ( 1.00@xmath55 fwhm instead 0.65@xmath55 for finkelstein et al .
another potentially significant contributing factor is our relatively conservative account of the uncertainties in the line flux measurements , including uncertainties that arise from the sky subtraction .
the uncertainties we derive are consistent with typical values reported by the mosdef program ( kriek et al .
2015 ) . ]
the 1d and 2d spectra for our other targeted @xmath4 candidate egs - zs8 - 2 is presented in figure [ fig : pascal_lya ] ( see also figure 2 in oesch et al .
2015b for spectra of the confirmed @xmath117 candidate ) . using a simple gaussian to determine the central wavelength of the observed line at 1.031@xmath8 m ( and ignoring asymmetry and other effects due to skylines surrounding this candidate ly@xmath13 line )
, we determine the spectroscopic redshift for the source to be @xmath118 , with a detection significance of 4.7@xmath16 for the line and a line flux of @xmath119 erg s@xmath115 @xmath120 .
while this line is only detected at @xmath121 significance , its reality appears to be supported by subsequent near - infrared spectroscopy obtained on this source from independent observing efforts ( d. stark et al .
2016 , in prep ) .
in addition to the ly@xmath13-emission lines reported by oesch et al .
( 2015b ) and this work , zitrin et al .
( 2015 ) report the detection of a 7.5@xmath16 ly@xmath13 line for our egsy-2008532660 candidate in new keck / mosfire observations ( june 10 - 11 , 2015 ) .
this redshift measurement sets a new high - redshift distance record for galaxies with spectroscopic confirmation .
our photometric selection therefore contains 3 of the 4 most distant , spectroscopically - confirmed galaxies to date .
the @xmath122 , @xmath123 and @xmath124 redshifts for egs - zs8 - 1 , egs - zs8 - 2 and egsy-2008532660 , respectively , are in excellent agreement with the photometric redshifts derived for these galaxies using hst+irac+ground based observations and our color criteria .
the absolute magnitude and redshifts of egs - zs8 - 1 , egs - zs8 - 2 , and egsy-2008532660 are presented in the top panel of figure [ fig : pascal ] in relation to other @xmath125 galaxies with clear redshift determinations from ly@xmath13 .
the current specroscopy provides us with considerable reassurance that our proposed color technique is an effective method to search for bright , @xmath9 galaxies .
emission has been found in all 4 galaxies that make up our selection .
our entire sample has therefore been spectroscopically confirmed to lie in the redshift range @xmath126 - 9.1 , with the spectroscopic redshifts being in excellent agreement with our derived photometric redshifts . ]
three of our four candidates were already identified as part of previous work .
tilvi et al .
( 2013 ) identified cosy-0237620370 as a @xmath62 galaxy by applying lyman - break - like criteria to the deep medium - band zfourge data and estimate a redshift of 7.16@xmath127 .
this source was also identified by bowler et al .
( 2014 ) as a @xmath62 galaxy ( 211127 in the bowler et al . 2014 catalog ) using the deep near - ir observations from the ultra - vista program and derived a photometric redshift of 7.03@xmath128 for the source ( or 7.20 if the source exhibits prominent ly@xmath13 emission ) , similar to what we find here .
tilvi et al .
( 2013 ) derive a @xmath61-[4.5]$ ] color of @xmath129 mag , while bowler et al .
( 2014 ) find [email protected] mag , both of which are broadly consistent with what we find here .
bouwens et al .
( 2015 ) identified 3 of the 4 sources as part of their search for @xmath62 - 8 galaxies over the five candels fields and segregated the sources into different redshift bins using the photometric redshift estimates .
the full hst + subaru suprime - cam @xmath130 + cfht megacam @xmath131 + ultravista @xmath132 photometry was used to estimate these redshifts for the candidate in the cosmos field .
meanwhile , the hst + cfht megacam @xmath131 + wircam @xmath133 + spitzer / irac @xmath58m@xmath59 m photometry was used in the case of the two egs candidates .
bouwens et al .
( 2015 ) derived a photometric redshift of @xmath134 for cosy-0237620370 over the candels - cosmos field and derived photometric redshifts of 8.1 for the two sources over the candels - egs field ( egs - zs8 - 1 and egs - zs8 - 2 , respectively ) , so the latter two candidates were placed in the @xmath32 sample of bouwens et al .
( 2015 ) .
there was , however , some uncertainty as to both the robustness and also the precise redshifts of the candels - egs candidates from bouwens et al .
prior to the present study , the use of the @xmath61-[4.5]$ ] color has never been systematically demonstrated to work for the identification of galaxies with redshifts of @xmath4 despite there being @xmath15 prominent examples of @xmath9 galaxies with particularly red @xmath61-[4.5]$ ] colors ( bradley et al .
2008 ; ono et al . 2012 ; finkelstein et al . 2013 ; tilvi et al .
2013 ; laporte et al . 2014 , 2015 ) .
moreover , no @xmath45-band observations were available over either @xmath9 candidate from the candels - egs field in the bouwens et al .
( 2015 ) selection to validate potential @xmath9 galaxies ( though such observations have fortuitously become available as a result of observations made from the z9-candels follow - up program [ bouwens et al .
2016 ] ) .
the apparent magnitudes of the @xmath126 - 8.5 galaxies identified as part of the current selection are much brighter than the typical galaxy at @xmath32 , as is evident in both the upper and lower panels in figure [ fig : pascal ] . in fact
, 3 of the sources from our current irac - red @xmath61-[4.5]>0.5 $ ] selection appear to represent the brightest @xmath2 galaxies known in the entire candels program and constitute 3 of the 4 @xmath32 candidates shown in the lower panel of figure [ fig : pascal ] .
the only other especially bright @xmath135 @xmath32 candidate shown in that lower panel is presented in the appendix ( since it satisfies our @xmath61-[4.5]>0.5 $ ] selection criteria using an independent set of photometry , i.e. , skelton et al .
2014 ) .
interestingly enough , all 4 of the brightest candidates shown in the lower panel of figure [ fig : pascal ] are located in the candels egs field , providing a dramatic example of how substantial field - to - field variations in the surface densities of bright sources might be ( though we note that egsy-2008523660 is likely gravitationally lensed ) .
this seems to be just a chance occurrence , as none of these candidates is clearly in a similar redshift window .
the probability that the 4 brightest @xmath32 sources in the candels program would be found in the same candels field ( even if one is gravitationally lensed ) is @xmath11% .
selection with bouwens et al .
( 2015 ) which would be potentially easier to select as a @xmath32 galaxy over the candels egs field .
it is presented in appendix a. its redshift is not well constrained ( lying anywhere between @xmath136 and 8.5 ) , but it would be marginally easier to find over the candels egs field since the bouwens et al . (
2015 ) @xmath32 sample extends down to @xmath62 over that field while the bouwens et al .
( 2015 ) @xmath32 samples over the other fields only extend down to @xmath137 . ] previously , this point had been strongly made by bouwens et al .
( 2015 ) in discussing the number of bright sources over the different candels fields ( figure 14 , appendices e and f from bouwens et al .
2015 ) and also quite strikingly by bowler et al .
( 2015 ) in comparing the number of bright @xmath34 galaxies over the ultravista and uds fields .
in this section , we will examine the implications of our present search results for the volume density of luminous galaxies in the @xmath62 - 9 universe .
first , we estimate how complete we might expect our selection to be based on the @xmath61-[4.5]$ ] color distribution in fields where the redshift can be constrained using deep @xmath3-band data ( 6.1 ) .
second , we make use of our search results and our completeness estimates to set a constraint on the bright end of the @xmath62 - 9 lf ( 6.2 ) .
ccccccc egs - zs8 - 1 & 14:20:34.89 & 53:00:15.35 & [email protected] & [email protected] & [email protected] & [ 1 ] , [ 8 ] + & & & & ( @[email protected] ) + egs - zs8 - 2 & 14:20:12.09 & 53:00:26.97 & [email protected] & @xmath139 & [email protected] & [ 1 ] + & & & & ( @[email protected] ) + egsy-2008532660 & 14:20:08.50 & 52:53:26.60 & [email protected] & 8.57@xmath80 & [email protected] & + & & & & ( @xmath138=8.683@xmath15 ) + gndy-6379018085 & 12:36:37.90 & 62:18:08.50 & [email protected] & 7.508 & [email protected] & [ 6 ] + borgy-9469443552 & 04:39:46.94 & @xmath652:43:55.20 & [email protected] & 8.29@xmath140 & & [ 1],[2],[3 ] , [ 7 ] + gsdy-2499348180 & 03:32:49.93 & @xmath627:48:18.00 & [email protected] & 7.84@xmath141 & [email protected] & [ 1],[3],[4],[5 ] + borgy-6504943342 & 14:36:50.49 & 50:43:34.20 & [email protected] & 7.49@xmath142 & & [ 1 ] + cosy-0235624462 & 10:00:23.56 & 02:24:46.20 & [email protected] & 7.84@xmath143 & [email protected] & [ 1 ] + borgy-2463351294 & 22:02:46.33 & 18:51:29.40 & [email protected] & 7.93@xmath144 & & [ 1 ] + borgy-2447150300 & 10:32:44.71 & 50:50:30.00 & [email protected] & 7.93@xmath145 & & [ 1 ] + borgy-5550543040 & 07:55:55.05 & 30:43:04.00 & [email protected] & 7.66@xmath146 & & [ 1 ] + median & & & & & 0.82@xmath147 & + + + egsy-9597563148 & 14:19:59.75 & 52:56:31.40 & [email protected] & 8.19 & [email protected] & [ 1 ] in our attempts to identify bright @xmath4 galaxies , we only consider those sources with red @xmath61-[4.5]>0.5 $ ] spitzer / irac colors to ensure that the sources we select are robustly at @xmath4 ( see 3.2 ) .
however , by making this requirement , we potentially exclude those @xmath4 galaxies which have bluer [ 3.6]-[4.5 ] colors , either due to lower - ew [ oiii]+h@xmath7 lines or simply as a result of noise in the photometry . to determine how important this effect is , we look at the [ 3.6]-[4.5 ] color distribution of galaxies which we can robustly place at a redshift @xmath4 ( where both lines in [ oiii ] doublet fall in the [ 4.5 ] band ) .
the most relevant sources are those bright galaxies we can place at @xmath4 based on the available hst+ground - based photometry and which include deep flux measurements at @xmath148 m .
such measurements are available for the candels goods - s , goods - n , uds , and cosmos fields , and a small fraction of the candels egs field . for our fiducial results here , we only consider selected sources from those fields brightward of @xmath149 and with redshift estimates greater than @xmath2 .
this is to ensure that we only include bona - fide @xmath126 - 9.1 galaxies ( where the [ oiii]+h@xmath7 line in the 4.5@xmath8 m band ) in our selection .
photometric redshift errors often have an approximate size of @[email protected] to this magnitude limit , and so to avoid @xmath71 sources scattering into our selection , we kept our cuts fairly conservative .
the list of such sources at such bright magnitudes is still somewhat limited at present , with only the bright @xmath125.6-mag galaxy in the candels south from yan et al .
( 2012 ) and oesch et al .
( 2012 ) , a bright @xmath150-mag galaxy in the candels cosmos field from bouwens et al .
( 2015 ) , a bright @xmath125.5 mag source over the candels goods - north field by finkelstein et al .
( 2013 ) , two bright sources over the candels egs field where @xmath45-band photometry is available ( egs - zs8 - 1 , egs - zs8 - 2 ) , and a third bright source over the candels egs field where the @xmath67 color allows us to place it at @xmath151 ( egsy-2008532660 ) . of these sources ,
five out of the six sources have [ 3.6]@xmath6[4.5 ] colors in excess of 0.5 , and therefore for simplicity , we will assume that our irac red selection is 83% complete , but we emphasize that the completeness correction we derive from this selection is uncertain and could be much larger ( as indeed one would expect if the [ 3.6]@xmath6[4.5 ] color measurement derived by labb et al .
2013 , i.e. , @xmath10.4 mag , for the average stacked @xmath32 galaxy are indicative ) . to investigate this possibility
, we examined a slightly larger sample of objects over the four fields where we have photometric redshifts using @xmath3-band imaging .
considering sources to a @xmath47-band magnitude limit of 26.2 over the candels - uds and cosmos fields and 26.7 over the candels goods - north and goods - south while extending the photometric redshift selection to @xmath152 , 6 out of 9 sources satisfy the [ 3.6]@xmath6[4.5]@xmath570.5 criterion .
while this suggests the actual fraction of @xmath4 galaxies with such red irac colors may be less than 83% , this fainter sample is still consistent with our fiducial percentage .
it also reassuring that our suggested selection criteria would also apply to gn-108036 , the bright @xmath153 @xmath154 galaxy found by ono et al .
( 2012 ) , given its measured [ 3.6]@xmath6[4.5 ] color of [email protected] mag .
we include a list of those sources and other sources in table [ tab : brightz8 ] from the bouwens et al .
( 2015 ) catalog and also including the bright @xmath155 galaxy from finkelstein et al .
( 2013 ) . in figure
[ fig : histo_data ] , we present the [ 3.6]-[4.5 ] color distribution for the brightest sources we know to robustly lie at @xmath2 based on spectroscopy or from the available hst+ground - based photometry for those sources that lie in regions of candels with @xmath3-band observations or with @xmath67 colors red enough to confidently place the sources at @xmath151 . the median [ 3.6]@xmath6[4.5 ] color that we measure is 0.82@xmath156 mag .
such a color implies a minimum ew of @xmath11300@xmath12 for the [ oiii]+h@xmath7 lines , assuming a flat stellar continuum and no line contribution to the [ 3.6 ] band .
however , we emphasize that if there is also a substantial line contribution to the [ 3.6 ] band , e.g. , from h@xmath157 , h@xmath158 , and [ oii ] , then the implied ew of the [ oiii]+h@xmath7 lines would be much larger .
for example , adopting the line ratios from the anders & fritze - v .
alvensleben ( 2003 ) , 0.2@xmath159 model would imply an ew of @xmath52100@xmath160for the [ oiii]+h@xmath7 lines .
some correction is required to the median [ 3.6]@xmath6[4.5 ] color measurement to account for the fact that the @xmath4 sources from the candels egs field were explicitly selected because of their red irac colors .
if we assume that the intrinsic [ 3.6]@xmath6[4.5 ] color for sources over the candels egs field is @xmath10.6 mag ( which is the value we find from the 3 candidates over the other candels fields : see table [ tab : brightz8 ] ) and the noise + scatter is @xmath10.4 ( the value from the other fields ) , we compute a bias of 0.24 mag from a simple monte - carlo simulation . accounting for such biases reduces the median [ 3.6]-[4.5 ] color of the population by 0.24 mag , which implies a median ew of @xmath1800@xmath12and 1500 , ignoring and accounting for a possible nebular contribution to the [ 3.6 ] band respectively .
here we use the search results from the previous section to set a constraint on the bright end of the @xmath32 lf .
we begin this section by calculating the total selection volume in which we would expect to find bright @xmath9 galaxies with our selection criteria .
we will estimate the selection volumes in a similar way to the methodology used by bouwens et al .
( 2015 ) in deriving the lfs from the full candels program . in short , we create mock catalogs over each search field , with sources distributed over a range in both redshift @xmath34 - 10 and apparent magnitude ( @xmath161 to 26 ) .
we then take the two - dimensional @xmath50-band images of similar luminosity , randomly - selected @xmath162 galaxies from the hudf ( bouwens et al .
2007 , 2011 , 2015 ) and create mock images of the sources at higher - redshift using the two dimensional pixel - by - pixel @xmath162 galaxies as a guide ( see bouwens et al . 1998 , 2003 ) , adopting random orientations relative to their orientation in the hudf and scaling their physical sizes as @xmath163 , which is the approximate relationship that has been found comparing the mean size of galaxies at fixed luminosity , as a function of redshift ( oesch et al . 2010 ; grazian et al . 2012
; ono et al . 2013 ;
holwerda et al . 2015 ; kawamata et al .
2015 ; shibuya et al .
individual sources were assigned @xmath22 colors based on their @xmath22 luminosity , using the @xmath7-@xmath164 relationship derived by bouwens et al .
( 2014 ) and allowing for an intrinsic scatter @xmath165 of 0.35 at high luminosities ( @xmath166 $ ] ) as found by bouwens et al .
( 2009 , 2012 ) and systematically decreasing to 0.15 at lower luminosities as found by rogers et al .
( 2014 ) .
in addition to the hst images we created for individual sources , we also constructed simulated ground - based and spitzer / irac images for these sources which we added to the real ground - based + spitzer / irac data .
these simulated images were generated based on the mock @xmath47-band images we constructed for individual sources and convolving by the @xmath47-to - irac , @xmath47-to - ground kernels that mophongo ( labb et al .
2010 ) derived from the observations . in producing simulated irac images for the mock sources , we assume a rest - frame ew of 300@xmath12for h@xmath13+[nii ] emission and 500@xmath12for [ oiii]+h@xmath7 emission over the entire range @xmath167 - 9 , a flat rest - frame optical color , and a @xmath47-optical continuum color of 0.2 - 0.3 mag , to match the observational results of shim et al . (
2011 ) , stark et al .
( 2013 ) , gonzlez et al .
( 2012 , 2014 ) , labb et al .
( 2013 ) , smit et al .
( 2014 , 2015 ) , and oesch et al .
( 2013 ) .
we took the simulated images we created for individual sources and added them to the real hst , ground - based , and spitzer / irac observations .
these simulated images were , in turn , used to construct catalogs and our selection criteria applied to the derived catalogs in exactly the same way as we applied these criteria to the real observations ( including excluding sources which violated our confusion criteria ) .
summing the results over all five candels fields , we compute a total selection volume of 1.6@xmath3510@xmath168 mpc@xmath169 per 1-mag interval for galaxies with @xmath93 magnitudes brightward of 25.5 . if we assume that the present selection of @xmath32 galaxies is complete , this would imply a volume density of @xmath911.4@xmath3510@xmath170 mpc@xmath171 mag@xmath115 and @xmath172@xmath3510@xmath170 mpc@xmath171 mag@xmath115 for @xmath173 - 25.0 and @xmath135 - 25.5 galaxies , respectively .
we ignore the contribution of the @xmath98 candidate galaxy egsy-2008532660 , given the evidence that it may be slightly magnified ( 3.4 ) .
however , we can not assume that the present selection of bright @xmath32 galaxies is complete , since not every @xmath32 galaxy exhibits such a red @xmath61-[4.5]$ ] color .
in the previous section , we found that only 5 out of the 6 bright ( @xmath174 ) , secure @xmath175 sources within candels showed such red galaxies . correcting the volume densities given in the previous paragraph to account for this slight empirically - derived incompleteness ( 0.83@xmath176 ) , we estimate volume densities of @xmath911.7@xmath3510@xmath170 mpc@xmath171 mag@xmath115 and @xmath177@xmath3510@xmath170 mpc@xmath171 mag@xmath115 for @xmath173 - 25.0 and @xmath135 - 25.5 galaxies , respectively . the uncertainty in the completeness estimate is included in the error we quote for the volume density .
it is interesting to compare these constraints on the volume density of bright @xmath11 - 8.5 galaxies with other recent constraints which are available on the volume density of bright end of the @xmath22 lfs at @xmath162 , @xmath34 , @xmath32 , and @xmath178 galaxies from state - of - the - art studies ( e.g. , bouwens et al . 2015 ) .
the results are shown in figure [ fig : lf8 ] , and it is clear that our result lies somewhere midway between the @xmath62 and @xmath32 lfs , as one might expect given the redshift distribution of the sources that make up our @xmath126 - 8.5 sample .
in this paper , we take advantage of the deep spitzer / irac observations available over all five candels in conjunction with the hst+ground - based data to conduct a search over 900 arcmin@xmath23 to find bright @xmath32 galaxies . to identify galaxies at such high redshifts , we select those galaxies with especially red spitzer / irac [ 3.6]@xmath6[4.5 ] colors ( i.e. , @xmath570.5 ) , in the hopes of identifying those @xmath24 galaxies which show the presence of a strong [ oiii]+h@xmath7 line in 4.5@xmath8 m band . such a selection is useful for the candels program , given the lack of uniformly deep @xmath3-band observations over all five fields .
our selection yielded 4 @xmath9 candidates brighter than an @xmath93 magnitude of 25.5 .
each of these four selected candidates was requried to be undetected ( @xmath912.5@xmath16 ) at optical wavelengths ( @xmath911@xmath8 m ) , as defined by the inverse - variance - weighted mean flux measurement , be undetected in the @xmath49-band ( @xmath911.5@xmath16 ) , and show a @xmath179 color redward of 1.5 . fortuitously , 3 of our 4 selected @xmath9 candidates had deep @xmath3-band observations available from either deep ground - based observations or from the new z9-candels follow - up program ( bouwens 2014 ; bouwens et al .
2016 ) with hst .
the available @xmath3-band observations provide clear confirmation of the @xmath180 redshifts we estimate for three of four candidates found in our search .
the redshift estimates we obtain for three of our selected candidates lie significantly above @xmath62 , with egs - zs8 - 2 having a redshift estimate of [email protected] , egs - zs8 - 1 having a redshift estimate of [email protected] , and egsy-2008532660 having a redshift estimate of 8.6@xmath181 .
we also obtained spectroscopic observations on two of our candidate @xmath4 galaxies in the near - ir and find probable ly@xmath13 lines in their spectra consistent with redshifts of [email protected] and [email protected] .
the detection of ly@xmath13 emission for these candidates is significant at 4.7@xmath16 and 6.1@xmath16 significance , respectively .
the second of these sources was featured in oesch et al .
( 2015b ) .
remarkably enough , a third candidate from our list was spectroscopically confirmed to lie at @xmath182 by zitrin et al .
( 2015 ) .
these sources represent the brightest @xmath183 candidates we identified over the entire candels program and are 0.5-mag brighter than @xmath183 candidates identified anywhere else on the sky .
coindentally enough , they all lie in the same candels field , again suggesting large field - to - field variations for brightest @xmath9 galaxies .
see also discussion in bouwens et al .
( 2015 ) and bowler et al .
( 2015 ) . using these candidates
, we estimate the volume density of bright ( @xmath86 ) @xmath9 galaxies in the early universe based on our selected sample and estimate that 17@xmath184% of @xmath4 galaxies do not show such red colors .
the volume density estimate we derive lies midway between the volume density of luminous @xmath34 galaxies bouwens et al .
( 2015 ) derive and the volume density of luminous @xmath32 galaxies .
the median [ 3.6]@xmath6[4.5 ] color distribution for our selection and other bright @xmath2 galaxies from the literature is 0.82@xmath156 mag ( observed ) and 0.58@xmath156 ( correcting for the approximate selection bias : see 6.1 ) .
this strongly points to the existence of extremely high ew nebular emission lines in typical star - forming galaxies at @xmath4 . assuming no contribution from nebular line emission to the [ 3.6 ] band implies a [ oiii]+h@xmath7 ew of @xmath1800 .
however , allowing for contamination of the [ 3.6 ] band in accordance with the expectations of anders & fritze - v .
alvensleben ( 2003 ) would imply a median ew of @xmath11500 .
these results are in reasonable agreement but perhaps slightly higher than smit et al .
( 2015 ) estimate for the irac - blue sources they selected at @xmath85 - 6.9 .
smit et al .
( 2015 ) estimate a typical @xmath185+h_{\beta}$ ] ew of 1085@xmath12 for their selected sources .
these estimates are similar albeit slightly higher than those estimated by labb et al .
( 2013 ) , laporte et al .
( 2014 , 2015 ) , and huang et al .
( 2016 ) . in the near future
, we would expect the brightest @xmath32 galaxies to be identified within the @xmath11 deg@xmath23 wide - area ultravista field ( mccracken et al .
2012 ) by combining the progressively deeper @xmath132 observations with constraints from the optical subaru+cfht observations and spitzer / irac observations from splash ( capak et al .
2013 ) and smuvs ( caputi et al .
another significant source of bright @xmath32 candidates will be the new borg@xmath186}$ ] program ( trenti 2014 ) , which uses a huge allotment of 500 orbits to cover a 500 arcmin@xmath23 area to @xmath187 mag depth ( @xmath53 ) .
we thank robert barone - nugent , daniel schaerer and dan stark for valuable conversations .
this work has benefited significantly from the public reductions of the seds program and hence the efforts of matt ashby , giovanni fazio , steve willner , and jiasheng huang .
we are grateful to dan stark , sirio belli , and richard ellis for communicating with us with some unpublished results they also obtained on egs - zs8 - 2 ( april 2015 ) where they also find a @xmath573@xmath16 line ( putatively ly@xmath13 ) at 1.031@xmath8 m ( to appear in d. stark et al .
2016 , in prep ) .
we acknowledge the support of nasa grant nag5 - 7697 , nasa grant hst - go-11563 , and a nwo vrij competitie grant 600.065.140.11n211 .
anders , p. , & fritze - v .
alvensleben , u. 2003 , , 401 , 1063 ashby , m. l. n. , willner , s. p. , fazio , g. g. , et al .
2013 , , 769 , 80 atek , h. , richard , j. , jauzac , m. , et al .
2015 , , 814 , 69 barone - nugent , r. l. , wyithe , j. s. b. , trenti , m. , et al .
2015 , , 450 , 1224 bertin , e. , & arnouts , s. 1996 , , 117 , 393 bouwens , r. , broadhurst , t. and silk , j. 1998 , , 506 , 557 bouwens , r. j. , illingworth , g. d. , rosati , p. , et al .
2003 , , 595 , 589 bouwens , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , , 670 , 928 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2011 , , 737 , 90 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2012 , , 754 , 83 bouwens , r. 2014 , hst proposal , 13792 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2015 , , 803 , 34 bouwens , r. j. , oesch , p. a. , labbe , i. , et al .
2016 , submitted , arxiv:1506.01035 bowler , r. a. a. , dunlop , j. s. , mclure , r. j. , et al .
2014 , , 440 , 2810 bowler , r. a. a. , dunlop , j. s. , mclure , r. j. , et al .
2015 , , 452 , 1817 burgasser , a. j. , mcelwain , m. w. , kirkpatrick , j. d. , et al .
2004 , , 127 , 2856 bradley , l. d. , bouwens , r. j. , ford , h. c. , et al .
2008 , , 678 , 647 bradley , l. d. , trenti , m. , oesch , p. a. , et al .
2012 , , 760 , 108 bradley , l. d. , zitrin , a. , coe , d. , et al .
2014 , , 792 , 76 brammer , g. b. , van dokkum , p. g. , & coppi , p. 2008
, , 686 , 1503 capak , p. , aussel , h. , ajiki , m. , et al .
2007 , , 172 , 99 capak , p. , aussel , h. , bundy , k. , et al .
2013 , spitzer proposal , 10042 caputi , k. , ashby , m. , fazio , g. , et al .
2014 , spitzer proposal , 11016 de barros , s. , schaerer , d. , & stark , d. p. 2014
, , 563 , a81 ellis , r. s. , mclure , r. j. , dunlop , j. s. , et al .
2013 , , 763 , ll7 fialkov , a. , & loeb , a. 2015 , , 806 , 256 finkelstein , s. l. , papovich , c. , dickinson , m. , et al .
2013 , , 502 , 524 finkelstein , s. l. , ryan , r. e. , jr . ,
papovich , c. , et al .
2014 , arxiv e - prints furusawa , h. , kosugi , g. , akiyama , m. , et al .
2008 , , 176 , 1 fontana , a. , dunlop , j. s. , paris , d. , et al .
2014 , , 570 , aa11 galametz , a. , grazian , a. , fontana , a. , et al . 2013 , ,
206 , 10 gonzlez , v. , bouwens , r. j. , labb , i. , et al .
2012 , , 755 , 148 gonzlez , v. , bouwens , r. , illingworth , g. , et al .
2014 , , 781 , 34 grazian , a. , castellano , m. , fontana , a. , et al .
2012 , , 547 , a51 grogin , n. a. , kocevski , d. d. , faber , s. m. , et al . 2011 , , 197 , 35 guo , y. , ferguson , h. c. , giavalisco , m. , et al .
2013 , , 207 , 24 holwerda , b. w. , bouwens , r. , oesch , p. , et al .
2015 , , 808 , 6 huang , k .- h .
, bradav c , m. , lemaux , b. c. , et al .
2016 , , 817 , 11 ishigaki , m. , kawamata , r. , ouchi , m. , et al .
2015 , , 799 , 12 jiang , l. , egami , e. , mechtley , m. , et al .
2013 , , 772 , 99 kawamata , r. , ishigaki , m. , shimasaku , k. , oguri , m. , & ouchi , m. 2015 , , 804 , 103 kirkpatrick , j. d. , cushing , m. c. , gelino , c. r. , et al .
2011 , , 197 , 19 koekemoer , a. m. , faber , s. m. , ferguson , h. c. , et al .
2011 , , 197 , 36 kotulla , r. , fritze , u. , weilbacher , p. , & anders , p. 2009
, , 396 , 462 kriek , m. , shapley , a. e. , reddy , n. a. , et al .
2015 , , 218 , 15 kron , r. g. 1980 , , 43 , 305 labb , i. , et al . 2010a , , 708 , l26 labb , i. , et al .
2010b , , 716 , l103 labb , i. , oesch , p. a. , bouwens , r. j. , et al .
2013 , , 777 , ll19 labb , i. , oesch , p. a. , illingworth , g. d. , et al .
2015 , , 221 , 23 laporte , n. , streblyanska , a. , clement , b. , et al .
2014 , , 562 , l8 laporte , n. , streblyanska , a. , kim , s. , et al .
2015 , , 575 , a92 lorenzoni , s. , bunker , a. j. , wilkins , s. m. , et al .
2013 , , 429 , 150 mason , c. a. , treu , t. , schmidt , k. b. , et al .
2015 , , 805 , 79 mclean , i. s. , steidel , c. c. , epps , h. w. , et al .
2012 , , 8446 , 84460j mccracken , h. j. , milvang - jensen , b. , dunlop , j. , et al .
2012 , , 544 , a156 mcleod , d. j. , mclure , r. j. , dunlop , j. s. , et al .
2015 , , 450 , 3032 mclure , r. j. , dunlop , j. s. , bowler , r. a. a. , et al .
2013 , , 432 , 2696 oesch , p.a . , et al . 2010a ,
, 709 , l16 oesch , p. a. , bouwens , r. j. , illingworth , g. d. , et al .
2012 , , 759 , 135 oesch , p. a. , labb , i. , bouwens , r. j. , et al .
2013 , , 772 , 136 oesch , p. a. , bouwens , r. j. , illingworth , g. d. , et al .
2014 , , 786 , 108 oesch , p. a. , bouwens , r. j. , illingworth , g. d. , et al . 2015a , , 808 , 104 oesch , p. a. , van dokkum , p. g. , illingworth , g. d. , et al .
2015b , , 804 , l30 oke , j. b. , & gunn , j. e. 1983 , , 266 , 713 ono , y. , ouchi , m. , mobasher , b. , et al .
2012 , , 744 , 83 ono , y. , ouchi , m. , curtis - lake , e. , et al .
2013 , , 777 , 155 pentericci , l. , fontana , a. , vanzella , e. , et al .
2011 , , 743 , 132 rogers , a. b. , mclure , r. j. , dunlop , j. s. , et al .
2014 , , 440 , 3714 schaerer , d. , & de barros , s. 2009 , , 502 , 423 schenker , m. a. , stark , d. p. , ellis , r. s. , et al .
2012 , , 744 , 179 schenker , m. a. , robertson , b. e. , ellis , r. s. , et al .
2013 , , 768 , 196 schmidt , k. b. , treu , t. , trenti , m. , et al .
2014 , , 786 , 57 shibuya , t. , kashikawa , n. , ota , k. , et al .
2012 , , 752 , 114 shibuya , t. , ouchi , m. , & harikane , y. 2015 , , 219 , 15 shim , h. , chary , r .-
r . , dickinson , m. , et al .
2011 , , 738 , 69 skelton , r. e. , whitaker , k. e. , momcheva , i. g. , et al .
2014 , , 214 , 24 smit , r. , bouwens , r. j. , labb , i. , et al .
2014 , , 784 , 58 smit , r. , bouwens , r. j. , franx , m. , et al .
2015 , , 801 , 122 stark , d. p. , schenker , m. a. , ellis , r. , et al .
2013 , , 763 , 129 szalay , a. s. , connolly , a. j. , & szokoly , g. p. 1999
, , 117 , 68 tilvi , v. , papovich , c. , tran , k .- v .
h. , et al .
2013 , , 768 , 56 trenti , m. , bradley , l. d. , stiavelli , m. , et al .
2011 , , 727 , l39 treu , t. , schmidt , k. b. , trenti , m. , bradley , l. d. , & stiavelli , m. 2013 , , 775 , ll29 trenti , m. 2014 , hst proposal , 13767 vanzella , e. , pentericci , l. , fontana , a. , et al .
2011 , , 730 , l35 wilkins , s. m. , coulton , w. , caruana , j. , et al .
2013 , , 435 , 2885 wilkins , s. m. , stanway , e. r. , & bremer , m. n. 2014 , , 439 , 1038 windhorst , r. a. , cohen , s. h. , hathi , n. p. , et al .
2011 , , 193 , 27 wyithe , j. s. b. , yan , h. , windhorst , r. a. , & mao , s. 2011 , , 469 , 181 yan , h. , yan , l. , zamojski , m. a. , et al .
2011 , , 728 , ll22 yan , h. , finkelstein , s. l. , huang , k .- h . , et al . 2012 , , 761 , 177 zheng , w. , postman , m. , zitrin , a. , et al .
2012 , , 489 , 406 ( z12 ) zheng , w. , shu , x. , moustakas , j. , et al .
2014 , , 795 , 93 zitrin , a. , zheng , w. , broadhurst , t. , et al .
2014 , , 793 , l12
in addition to our application of our criteria to the catalogs bouwens et al .
( 2015 ) compiled over a 750 arcmin@xmath23 search area within candels , we also made use of the catalogs from the 3d - hst team ( skelton et al .
2014 ) over the same region .
our rationale to do so was to maximize the completeness of our selection for bright @xmath9 galaxies .
one additional @xmath9 galaxy candidate is found which did not make it into our fiducial selection using the bouwens et al .
( 2015 ) catalogs ( because it had a measured @xmath61-[4.5]$ ] color of @xmath10.2 mag).@xmath6[4.5 ] colors we measure for the 4 other sources in our selection agree to @xmath910.1 mag with the skelton et al .
( 2014 ) values ( @xmath10.05 mag differences are typical ) . ]
we tabulate its coordinates , @xmath93-band magnitude , @xmath61-[4.5]$ ] color , and estimated redshift in table [ tab : table_details2 ] .
postage images of the @xmath32 candidate is presented in figure [ fig : postage_extra ] .
model fits to the photometry skelton et al .
( 2014 ) provide on the source , as well as the inferred redshift likelihood distribution , are also presented in figure [ fig : sed_extra ] .
ccccccc egsy-9597563148 & 14:19:59.76 & 52:56:31.40 & [email protected] & [email protected] & 8.19@xmath188 & [ 1 ] + @xmath189component - a & 14:19:59.78 & 52:36:31.30 & [email protected] + @xmath189component - b & 14:19:59.73 & 52:36:31.70 & [email protected]
in 3.2 , we considered a selection of sources from the four candels fields with deep @xmath3-band observations to test the idea that we could use an irac color criterion , i.e. , @xmath61-[4.5]>0.5 $ ] , combined with an optical dropout criterion to identify galaxies at @xmath4 even in the absence of @xmath3-band data .
in table [ tab : valid_samp ] , we provide a compilation of the 15 sources that we identified which satisfied the primary selection criteria fom the paper but which are brighter than 26.7 mag in the @xmath47-band ( and brighter than 26.5 mag over the candels uds and cosmos fields ) .
cccccc gndy-6487514332 & 12:36:48.752 & 62:14:33.29 & 26.4 & @xmath190 & 7.66 + gndy-7048017191 & 12:37:04.805 & 62:17:19.14 & 26.2 & @xmath191 & 7.84 + gnwy-7379420231 & 12:37:37.941 & 62:20:23.14 & 26.5 & @xmath192 & 8.29 + gnwz-7455218088 & 12:37:45.529 & 62:18:08.87 & 26.5 & @xmath193 & 7.16 + gsdz-2468850074 & 03:32:46.889 & @xmath627:50:07.45 & 26.0 & @xmath194 & 7.24 + gswy-2249353259 & 03:32:24.934 & @xmath627:53:25.94 & 26.1 & @xmath195 & 8.11 + gsdy-2209651370 & 03:32:20.964 & @xmath627:51:37.02 & 26.3 & @xmath196 & 7.84 + cosy-0439027359 & 10:00:43.90 & 2:27:35.9 & 26.6 & @xmath193 & 7.33 + cosz-0237620370 & 10:00:23.76 & 2:20:37.0 & 25.1 & @xmath197 & 7.14 + cosy-0235624462 & 10:00:23.56 & 2:24:46.2 & 25.7 & @xmath198 & 7.84 + udsy-4133353345 & 02:17:41.333 & @xmath65:15:33.45 & 25.8 & @xmath199 & 7.41 + udsy-4308785165 & 02:17:43.087 & @xmath65:08:51.65 & 26.3 & @xmath200 & 7.84 + udsz-4199355469 & 02:17:41.993 & @xmath65:15:54.69 & 26.5 & @xmath201 & 7.08 + udsy-1765825082 & 02:17:17.658 & @xmath65:12:50.82 & 26.3 & @xmath194 & 7.93 + udsy-5428621201 & 02:16:54.286 & @xmath65:12:12.01 & 26.1 & @xmath202 & 7.49 + | we identify 4 unusually bright ( h@xmath0<25.5 ) galaxies from hst and spitzer candels data with probable redshifts _ _ z__@xmath17 - 9 .
these identifications include the brightest - known galaxies to date at @xmath2 . as @xmath3-band observations are not available over the full candels program to perform a standard lyman - break selection of @xmath4 galaxies , we employ an alternate strategy using deep spitzer / irac data .
we identify [email protected] - 9.1 galaxies by selecting _ _ z__@xmath56 galaxies from the hst candels data that show quite red irac [ 3.6]@xmath6[4.5 ] colors , indicating strong [ oiii]+h@xmath7 lines in the 4.5@xmath8 m band .
this selection strategy was validated using a modest sample for which we have deep y - band coverage , and subsequently used to select the brightest @xmath9 sources . applying the irac criteria to all hst - selected optical - dropout galaxies over the full @xmath1900 arcmin@xmath10 of the candels survey revealed four unusually bright @xmath11 , 7.6 , 7.9 and 8.6 candidates .
the median [ 3.6]@xmath6[4.5 ] color of our selected @xmath11 - 9.1 sample is consistent with rest - frame [ oiii]+h@xmath7 ews of @xmath11500@xmath12 in the [ 4.5 ] band .
keck / mosfire spectroscopy has been independently reported for two of our selected sources , showing ly@xmath13 at redshifts of [email protected] and 8.683@xmath15 , respectively .
we present similar keck / mosfire spectroscopy for a third selected galaxy with a probable 4.7@xmath16 ly@xmath13 line at @[email protected] .
all three have h@xmath18-band magnitudes of @xmath125 mag and are @xmath10.5 mag more luminous ( @xmath19 ) than any previously discovered _ _
z__@xmath18 galaxy , with important implications for the uv lf .
our 3 brightest , highest redshift @xmath20>7 galaxies all lie within the candels egs field , providing a dramatic illustration of the potential impact of field - to - field variance . |
factorization theorems are at the heart of the theory of high - energy inclusive processes involving a large momentum transfer .
examples include the inclusive production in hadron - hadron collisions of new , heavy particles , such as the higgs boson or the supersymmetric partners of the currently known particles .
these theorems apply to _ inclusive _ cross sections for the production of a set @xmath0 of heavy particles or of a system of jets with a total mass @xmath1 , @xmath2 the theorem assumes that @xmath0 is defined in such a way that all final states that differ by the emission of soft hadrons and by the collinear rearrangement of momenta between outgoing light hadrons are summed over .
it is important here that @xmath1 be large .
one usually assumes that @xmath1 is of the order of the @xmath3 invariant mass @xmath4 . for @xmath5 ,
one should sum logarithms of @xmath6 to improve the usefulness of the perturbative expansion used in the calculations . according to the factorization theorem , the cross section for ( [ abfx ] )
may be written as a convolution of parton distributions and perturbative hard scattering functions , @xmath7 here , the nonperturbative functions @xmath8 give the distribution of parton @xmath9 in hadron @xmath10 as a function of the longitudinal momentum fraction @xmath11 , evolved to factorization scale @xmath12 .
the hard - scattering function @xmath13 is computed in perturbation theory up to some order , @xmath14 , leaving corrections of order @xmath15 .
as indicated , corrections to the factorization formula are suppressed by a positive power , @xmath16 , of the large mass . despite their central role ,
the foundations of the factorization theorems have received rather little theoretical attention in recent years beyond the arguments given in ref .
@xcite and reviewed in ref .
an exception is a recent paper by gotsman , levin , and maor @xcite , which discusses the possibility of corrections to the right - hand side of eq .
( [ factth ] ) that are not power - suppressed at all .
these conjectured corrections are associated with the exchange of ladders of the reggeon type .
it is our aim in this paper to show that such exchanges do _ not _ require a revision of eq .
( [ factth ] ) , and that , indeed , in an inclusive cross section of the type described above , contributions from the regions of momentum space identified in @xcite cancel
. we shall rely heavily on the discussion of ref .
@xcite , and will not find it necessary to modify the reasoning presented there . indeed
, the discussion of that paper already includes the treatment of the effect identified in ref .
@xcite , to all orders in perturbation theory . nevertheless , since ref .
@xcite does not include low - order examples , the treatment of a rapidity - ordered ladder will , we hope , illustrate and perhaps clarify the proof .
we consider the cross section @xmath17 for the production of a higgs boson with rapidity @xmath18 in the collision of two hadrons with c.m.energy @xmath4 .
our hadrons each consist of a simple quark - antiquark pair with a pointlike coupling to the hadron field .
the hard scale in the problem is the higgs boson mass , which we denote by @xmath1 .
we may have @xmath19 or @xmath20 .
our analysis works in either case .
we analyze cut feynman diagrams for higgs boson production that arise from the uncut graph shown in fig .
[ basic ] .
this choice of graph is motivated by the suggestion of gotsman _
@xcite , who examine graphs similar to this .
this graph contains enough of the physics to be instructive , while it is of low enough order to be treated in detail .
we hope that it will be evident that the argument given below for the graph of fig . [
basic ] could be applied to more complicated graphs . in fig .
[ basic ] , the star represents the hard scattering matrix element for @xmath21 . the lowest order diagram for this , containing a top quark loop ,
is shown in fig .
we shall treat only the case where higgs production appears on the outer " sides of the ladders , as in fig .
[ basic ] . moving the starred vertex to the other side of
either or both ladders @xcite leads to diagrams that can be treated by exactly the same methods as fig.[basic ] .
( basic.eps width 8 cm ) ( hard.eps width 3 cm ) we take @xmath22 and @xmath23 , where @xmath24 .
then @xmath25 . let the higgs boson momentum be @xmath26 , so @xmath27 .
it is helpful to have a specific reference frame in mind .
let us choose the reference frame in which @xmath28 .
there are various integration regions that make leading contributions to the cross section .
we consider the region in which all transverse momenta are of order @xmath29 ( @xmath30 ) and in which all lines outside the hard scattering have virtuality at most of order @xmath31 .
we take quark masses to be of order @xmath29 .
the gluons entering the hard interaction have momenta @xmath32 , @xmath33 where @xmath34 and @xmath35 are of order @xmath1 . the transverse components @xmath36 and @xmath37 are of order @xmath29
. then @xmath38 and @xmath39 are much smaller than @xmath29 , of order @xmath40 in order that @xmath41 . in the integration region of interest ,
all of the lines in the upper part of the graph carry large @xmath42 momenta , with positive @xmath42 momentum carried from left to right through the graph .
the transverse momentum components are all of order @xmath29 and the @xmath43 components are all small .
now consider the bottom part of the graph , not including the two gluons exchanged between top and bottom .
all of the of the lines here carry large @xmath43 momenta , with positive @xmath43 momentum carried from left to right through the graph .
the transverse momentum components are all of order @xmath29 and the @xmath42 components are all small
. we shall refer to this as the ladder region " of momentum space .
there are two one - rung ladders in the diagram .
for the left hand ladder , for instance , we choose the labeling shown in fig .
[ nascent ] .
as mentioned above , we work in the integration region with @xmath44 . we take @xmath45 and @xmath46 with @xmath47 . if @xmath48 then @xmath49-integration over this range can give a large logarithm @xmath50 , where the logarithm comes from the strongly ordered region @xmath51 .
however , we do not need strong ordering .
what is important for us is that @xmath52 , @xmath53 and @xmath54 are large and positive , at least of order @xmath1 .
( nascent.eps width 6 cm ) by assumption , we are in a region of momentum where the exchanged gluon with momentum @xmath55 has @xmath56 and all the lines outside the hard scattering have virtuality at most of order @xmath31 .
then the important integration region for @xmath57 is @xmath58 so that propagators in the @xmath59 system are not far off shell .
similarly , the important integration region for @xmath60 is @xmath61 .
then @xmath62 , so we can simplify our problem by replacing @xmath63 in the gluon propagator . with this replacement for @xmath64 ,
the soft gluon line has @xmath65 , so that it can never cross the final state cut .
( this is just to reduce the number of cuts considered .
our standard analysis @xcite does not use this assumption , which allows us to consider also @xmath66 . )
exactly similar remarks apply to the soft gluon , of momentum @xmath67 , in the other ladder .
now , following the argument in ref .
@xcite , we break up our graph into @xmath68 and @xmath69 , as shown in fig .
[ split ] .
the jet subgraph @xmath68 contains all of the lines near the mass shell with large plus momenta , @xmath70 .
the remaining subdiagram contains all of the other lines , including another jet subgraph with lines with large minus momenta , the two soft lines carrying momenta @xmath55 and @xmath71 , and the hard subgraph .
( split.eps width 8 cm )
we now show the cancellation of the ladder region in the sum over cuts of fig . [ basic ] .
we proceed in a number of steps that follow the arguments of ref .
these start with some approximations that give power - suppressed errors .
_ neglect @xmath72 and @xmath73 in @xmath68 so that subgraph @xmath69 includes @xmath74 .
similarly , neglect @xmath38 and @xmath75 in @xmath69 so that subgraph @xmath68 includes @xmath76 .
_ neglect @xmath57 and @xmath77 in @xmath68 .
this gives @xmath78 .
also take @xmath79 .
_ step 3 . _ show that @xmath80 is independent of which side of the final state cut the upper ends of the two soft gluons are on .
steps 4 and 5 will be explained later and illustrated by explicit calculation . before proceeding to them ,
we need to discuss _ step 3 _ in detail , since it is not obviously correct .
we need to show that the four cuts of @xmath69 shown in fig .
[ rcut ] give equal contributions to @xmath69 .
( rcut1.eps width 7 cm ) ( rcut2.eps width 7 cm ) ( rcut3.eps width 7 cm ) ( rcut4.eps width 7 cm ) in ref .
@xcite , we show this by writing the diagrams in @xmath81 ordered perturbation theory . here
we have an especially simple situation , so we can simplify the argument .
consider , for example , the first graph in fig .
[ rcut ] .
( rcutrouting.eps width 8 cm ) we can route @xmath57 and @xmath77 differently in different graphs because we are integrating over @xmath72 and we can shift the origin of the @xmath72 integration . in this particular graph , we choose to route @xmath57 directly to the left hand hard interaction , as illustrated in fig . [ rcutrouting ] . with this momentum routing
, we perform the @xmath57 integration first . after accounting for the replacement ( [ softprops ] )
, we see that there are two poles in the complex @xmath57 plane , one from the quark line carrying momentum @xmath82 and one from the gluon line carrying momentum @xmath83 . in the ladder region of integration ,
the quark lines in the figure all carry positive momentum fractions @xmath84 from left to right .
then the two poles in the @xmath57 plane are on opposite sides of the real @xmath57 axis .
we close the @xmath57 integration contour in the lower half plane to put the quark line on shell .
the result is the same as that obtained by replacing @xmath85 just as if the quark were crossing a final state cut .
we make a similar choice for the @xmath77 integration .
the result is shown in the first graph in fig .
[ rcutcut ] , where we indicate the factors @xmath86 with `` cut '' symbols .
( rcutcut1.eps width 7 cm ) ( rcutcut2.eps width 7 cm ) ( rcutcut3.eps width 7 cm ) ( rcutcut4.eps width 7 cm ) now we do this for each of the four graphs .
we get the results shown in fig .
[ rcutcut ] . evidently , all four of these cut graphs are equal , except for the factors of @xmath87 associated with the vertices and the uncut propagators . but to go from one graph to another we transfer one or more uncut propagators and the same number of vertices from one side of the final - state cut to the other
. hence the different cuts are in fact equal .
_ now we look at @xmath68 . in ref .
@xcite we argue that , after summing over cuts , we can replace @xmath88 and @xmath89 in @xmath68 .
this involves showing ( by writing the diagrams in @xmath81 ordered perturbation theory ) that \a ) after summing over cuts , poles in @xmath60 and @xmath90 near the origin are in the lower half planes ; \b ) we can deform the @xmath91 integration contours far into the upper half planes ; \c ) on the deformed contours , if @xmath92 is the momentum of a propagator through which @xmath93 flows , then @xmath94 becomes large : @xmath95 ; thus we can set @xmath96 ; \d ) finally , we can return the contours to the real axes .
the argument in ref .
@xcite proceeds to combine the results from different feynman graphs , using gauge invariance , to finally write the result in a factored form . in the present case
we have an especially simple situation , so we can simplify the argument .
in fact , the contribution from the ladder region to the graph we are considering vanishes at leading order in @xmath97 by itself .
thus we do not need _
step 5_. we will therefore make a straightforward analysis of the graphs to show that . after summing over cuts , @xmath98 in the ladder region .
since subgraph @xmath69 is the same no matter where we draw the cut , this implies that the sum over cuts of the original graph vanishes at the leading power of @xmath97 .
we now show explicitly that the sum over cuts of the jet subgraph @xmath68 gives zero .
there are nine cuts of @xmath68 to consider .
we look first at the three cuts shown in fig .
[ jetcut ] . call these cut graphs @xmath99 , @xmath59 , and @xmath10 .
we have oriented the lines to indicate the direction of flow of the @xmath42 component of momentum ; recall that the @xmath42 components of @xmath100 and @xmath101 have been set to zero in @xmath68 .
( jetcuta.eps width 7 cm ) ( jetcutb.eps width 7 cm ) ( jetcutc.eps width 7 cm ) we consider first cut graph @xmath99 .
this cut graph has a virtual loop , shown in more detail in fig .
[ jetcutafate ] .
let loop momentum @xmath102 circulate around this loop as shown in the first part of fig .
[ jetcutafate ] .
since , in the ladder integration region , positive @xmath42 momentum flows from left to right through the propagators in the figure , there are three poles in the upper half @xmath102 plane and one in the lower half @xmath102 plane .
we close the contour in the lower half @xmath102 plane .
this amounts to putting the quark line on shell , as indicated by the cut symbol in second part of fig .
[ jetcutafate ] .
next , we route @xmath60 to the hard vertex through the two gluon propagators , as depicted in the second part of fig .
[ jetcutafate ] .
there are two poles in the @xmath60 plane .
we perform the @xmath60 integral by closing the contour in the upper half plane , which amounts to putting the first gluon through which @xmath60 flows on shell .
the result is indicated in the first graph , @xmath99 , of fig .
[ jetcutcut ] .
( jetcutaa.eps width 7 cm ) ( jetcutaaa.eps width 7 cm ) in cut graph @xmath59 , we route @xmath60 to the hard vertex through the two gluon propagators , indicated by heavy lines in fig .
[ jetcut ] . then closing the @xmath60 contour in the upper half plane
gives the result indicated in the second graph , @xmath59 , of fig .
[ jetcutcut ] in cut graph @xmath10 , we route @xmath60 to the hard vertex to the right , through a gluon propagator , two quark propagators , and two more gluon propagators , indicated with heavy lines .
there are five poles in the @xmath60 plane , two in the upper half plane and three in the lower half plane .
we perform the @xmath60 integral by closing the contour in the upper half plane .
then there are two terms .
the two terms are indicated in the third and fourth graphs , @xmath103 and @xmath104 , of fig .
[ jetcutcut ] .
now we note that @xmath103 cancels @xmath99 , while @xmath104 cancels @xmath59 : in each of the canceling pairs of graphs , the propagators and vertices are the same up to sign and there are an odd number of sign changes resulting from moving uncut propagators and vertices from one side of the final state cut to the other .
( jetcutcuta.eps width 7 cm ) ( jetcutcutb.eps width 7 cm ) ( jetcutcutc1.eps width 7 cm ) ( jetcutcutc2.eps width 7 cm ) there are more cut graphs . for each way that we cut the right hand ladder , there are three ways to cut the left hand ladder .
thus we could label the graphs @xmath105 with @xmath106 denoting the cut for the left hand ladder and @xmath107 denoting the cut for the right hand ladder . in this notation ,
what we have seen is that @xmath108 . the same argument shows that @xmath109 a similar argument also shows that @xmath110 thus there is a double cancellation .
we conclude that the sum over all cuts of fig .
[ basic ] is @xmath111 and we do not have a violation of factorization .
gotsman _ et al . _ consider more elaborate diagrams than the diagram of fig .
[ basic ] that we have analyzed here .
for instance , they consider exchanges of gluon ladders with arbitrary numbers of rungs . in addition , they concentrate on the region of integration space in which the rapidities of the rungs are strongly ordered
. it should be evident that the calculation given above could be applied to more complicated diagrams
. there would be little to change in the pattern of cancellation found here , which illustrates the general argument of ref .
@xcite , although the notation would have to be more complicated .
we draw attention to the fact that although factorization applies in the strongly ordered region that gives the leading small @xmath112 logarithms , factorization also applies in other regions : strong ordering of rapidities is not essential to the argument .
the argument presented here applies when the exchanged gluons have transverse momentum of order @xmath29 , and when none of the lines outside of the hard scattering has virtuality much larger than @xmath31 .
this is also the region discussed in ref .
other regions than the one we considered here are correctly treated by other parts of the factorization proof in @xcite .
let us close with a few additional comments relating our work to that of ref.@xcite .
use modified agk cutting rules to perform their calculation , finding a remainder after summing over final states .
such an analysis , however , assumes that the incoming particles to which the ladders attach are on - shell .
in addition , we believe that this analysis is exact only at the level of leading logarithms . however , their noncancelling result ( eq . ( 8) in ref.@xcite ) , which is proportional to the square of the real part of their single - ladder exchange amplitude , is suppressed by two logarithms compared to leading logarithms .
we may note two specific differences between our analysis and that of gotsman _
, both of which are important beyond the leading logarithmic approximation .
first , they ignore certain final - state cuts , for example cut b in fig .
[ jetcut ] .
this cut crosses the side of a ladder .
such a cut is not allowed in the leading logarithm approximation , but it is present when one goes beyond leading logarithms . observe that the cut in question would not be kinematically allowed if the gluon ladder coupled to on - shell quark lines instead of internal off - shell lines .
second , gotsman _
_ appear to have ignored the loop integral that we label @xmath113 , which we use to prove that the subgraph @xmath69 is independent of the final state cut of @xmath114 ( _ step 3 _ ) .
both the extra cuts and the integral over @xmath113 are necessary to evaluate the ladder region fully .
the proof of factorization that we gave in ref .
@xcite applies to the full leading - power contribution and not just to leading logarithms .
it is this more general method that we have explained in the context of particular feynman graphs .
although we believe that factorization works , we should emphasize our concern that improvements are needed in the rigor of existing proofs of factorization @xcite .
conspicuously lacking are iterative algorithms for demonstrating ( [ factth ] ) along the lines of those available for proofs of renormalization .
our presentation here is not a step in this direction , rather it is an illustration of already - existing arguments .
we believe , however , that it is possible to construct such an algorithm ( _ cf .
improvements in the proof would help both to increase confidence in factorization and to clarify its variant applications , for instance to transverse momentum distributions @xcite in the range of @xmath115 .
we therefore welcome the motivation provided by @xcite to reopen the discussion of these important issues .
we thank the aspen center for physics for its hospitality during the formative stages of this analysis .
this work was supported by the u.s .
department of energy grants de - fg02 - 90er-40577 and de - fg03 - 96er40969 and the u. s. national science foundation , grant phy9722101 .
e. gotsman , e. levin , and u. maor , phys .
b 406 ( 1997 ) 89 , e - print archive : hep - ph/9705205 ; e. levin , in _ dis 97 , proceedings of the 5th international workshop on deep inelastic scattering and qcd _ , edited by j. repond and d. krakauer , aip conf .
407 , ( aip , new york , 1997 ) , p. 970 , e - print archive : hep - ph/9705375 .
collins , d.e .
soper and g. sterman , nucl .
b 250 ( 1985 ) 199 ; g. altarelli , r.k .
ellis , m. greco and g. martinelli , nucl .
b 246 ( 1984 ) 12 ; c.t.h .
davies and w.j .
stirling , nucl .
b 244 ( 1984 ) 337 ; c.t.h .
davies , b.r .
webber and w.j .
stirling , nucl .
b 256 ( 1985 ) 413 ; p.b .
arnold and r.p .
kauffman , nucl .
b 349 ( 1991 ) 381 ; g.a .
ladinsky and c.p .
yuan , phys .
d 50 ( 1994 ) r4239 , e - print archive : hep - ph/9311341 . | we show that existing proofs of factorization imply the cancellation of certain multiladder contributions that gotsman , levin , and maor had suggested would invalidate the basic factorization theorem in qcd .
no modifications of the original argument are necessary , although the details of the example offer useful insight into the mechanisms of factorization .
= 6.5 in = -0.5 in epsf.tex ( # 1 width # 2)=#2 |
in the supplemental material , we give details for exact diagonalization of hamiltonian ( 1 ) considered in the main text .
the hamiltonian is given by @xmath88 with @xmath27 .
under the periodic boundary condition , we can transform the hamiltonian from the real space to the momentum space through a fourier transformation . for @xmath27 , there are 3 sites in each unit cell and the single - particle energy spectrum are composed of 3 bands . for convenience ,
we use @xmath89 , @xmath90 , and @xmath91 to represent the annihilation operator at different sites in a unit cell , respectively . to represent it in momentum space
, we make the fourier transformation @xmath92 where @xmath93 , @xmath94 , @xmath95 and m is the index of the unit cell .
so the hamiltonian @xmath96 can be represented in momentum space , and the noninteracting part @xmath97 is given by @xmath98 the interacting part @xmath99 can be represented as @xmath100 with @xmath101 and @xmath102 taking @xmath70 , @xmath103 and @xmath104 . for the case of @xmath105 , @xmath106}{(3\eta)^3 } } -\frac{v}{2n_{cell}}\frac{\cos{\frac{(k_3-k_4)n_{cell}}{2}}}{(\frac{3n_{cell}}{2})^3}\ } \gamma_{\beta k_1}^{\dag}\gamma_{\beta k_2}\gamma_{\beta k_3}^{\dag}\gamma_{\beta k_4}\delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi ) } , \nonumber \\ h_{int}^{12}&=&\frac{v}{n_{cell}}\sum_{k_1,k_2,k_3,k_4}\sum_{\eta=0}^{n_{cell}/2 - 1}[\frac{e^{i(k_3-k_4)\eta}}{(3\eta+1)^3 } + \frac{e^{-i(k_3-k_4)(\eta+1)}}{(3\eta+2)^3}]\gamma_{1k_1}^{\dag}\gamma_{1k_2}\gamma_{2k_3}^{\dag}\gamma_{2k_4}\delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi ) } , \nonumber \\ h_{int}^{13}&=&\frac{v}{n_{cell}}\sum_{k_1,k_2,k_3,k_4}\sum_{\eta=0}^{n_{cell}/2 - 1}[\frac{e^{i(k_3-k_4)\eta}}{(3\eta+2)^3 } + \frac{e^{-i(k_3-k_4)(\eta+1)}}{(3\eta+1)^3}]\gamma_{1k_1}^{\dag}\gamma_{1k_2}\gamma_{3k_3}^{\dag}\gamma_{3k_4}\delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi)}. \nonumber\end{aligned}\ ] ] for the case of @xmath107 , @xmath108}{(3\eta)^3 } } \gamma_{\beta k_1}^{\dag}\gamma_{\beta k_2}\gamma_{\beta k_3}^{\dag}\gamma_{\beta k_4}\delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi ) } , \nonumber \\
h_{int}^{12}&=&\frac{v}{n_{cell}}\{\sum_{k_1,k_2,k_3,k_4}\sum_{\eta=1}^{(n_{cell}-1)/2}[\frac{e^{i(k_3-k_4)\eta}}{(3\eta+1)^3 } + \frac{e^{-i(k_3-k_4)\eta}}{(3\eta-1)^3}]+1\}\gamma_{1k_1}^{\dag}\gamma_{1k_2}\gamma_{2k_3}^{\dag}\gamma_{2k_4}\delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi ) } , \nonumber \\
h_{int}^{13}&=&\frac{v}{n_{cell}}\{\sum_{k_1,k_2,k_3,k_4}\sum_{\eta=1}^{(n_{cell}-1)/2}[\frac{e^{i(k_3-k_4)(\eta-1)}}{(3\eta-1)^3 } + \frac{e^{-i(k_3-k_4)(\eta+1)}}{(3\eta+1)^3}]+e^{-i(k_3-k_4)}\}\gamma_{1k_1}^{\dag}\gamma_{1k_2}\gamma_{3k_3}^{\dag}\gamma_{3k_4 } \delta_{(k_1-k_2+k_3-k_4)\mathrm{mod}(2\pi)}. \nonumber\end{aligned}\ ] ] here @xmath109 is for long - range interactions between sites @xmath110 and @xmath111 in the hamiltonian , and @xmath110 , @xmath111 are for the indexes of the sites in a unit cell .
we note that @xmath112 and @xmath113 have the same form for both cases . under the periodic boundary condition ,
total quasi - momentum of the system @xmath114 is conservative .
it can decompose the total hilbert space into subspaces according to the irreducible representations .
the hamiltonian can not couple two states belonging to two different total momentum @xmath115 and thus it is block diagonalized .
we can diagonalize the hamiltonian in the each subspace of total momentum @xmath115 . under the twisted boundary condition , each momentum @xmath53 in brillouin zone
makes a shift @xmath116 .
while the total momentum @xmath115 is still a good quantum number , the hamiltonian can be diagonalized in each subspace of total momentum @xmath115 . by changing @xmath52 ,
the spectrum flux can be obtained . for the fractionally filling flat band system considered in @xcite
, the hamiltonian can be diagonalized in the filled band by approximately neglecting the other bands , which is similar to the projecting to the lowest landau level in traditional fqh effect @xcite .
the dimension of the hilbert space of the hamiltonian is thus greatly reduced . in our case , we take a standard hopping form with a cosine dispersion .
it is hard to ensure the availability of the projection method , due to the extended band and the large @xmath13 .
taking @xmath117 as an example , in the current parameters of @xmath27 and @xmath21 , we need choose @xmath118 .
the projecting method only considers the filled bands ( here is the lowest band ) , and the dimension of the hamiltonian is reduced to about @xmath119 for each total momentum sector . while in our case , we need consider all the bands , the dimension ( about @xmath120 for each total momentum sector ) is too huge to calculate by using the full diagonalization method . and different @xmath13 under the periodic boundary condition .
( d)-(f ) spectrum flow as the function of @xmath121 for the system with coulomb interaction , @xmath122 and different @xmath13 under the twist boundary condition . here ,
@xmath34 and from top to bottom , @xmath123 . , width=340,height=340 ] in the main text , we study the case with the long - range dipole - dipole interaction decaying as @xmath12 , which corresponds to the dipolar fermi gas .
we would like to indicate that the main conclusions in the main text do not depend on the specifical form of dipole - dipole interaction . for systems with the coulomb interaction decaying as @xmath124
, we can obtain similar conclusions . to give an example , we study the system with coulomb interaction described by the hamiltonian : @xmath125 with @xmath27 .
the hamiltonian can be diagonalized in the momentum space by following procedures described in the above section . in fig.[sfig1](a)-(c ) , we show the low - energy spectrum for the above system with @xmath14 , @xmath26 , @xmath27 , @xmath28 , @xmath34 and different @xmath13 under the periodic boundary condition . similar to the dipolar system , when @xmath13 is small , for example , @xmath45
, the low - energy parts assemble together and it is hard to distinguish the lowest excited states apart from the higher excited states by an obvious gap . when @xmath126 , the lowest three states are separated from the higher excited states with an obvious gap . as @xmath13 increases further , the gap becomes more obvious and the lowest three states become nearly degenerate at @xmath56 .
similar to the dipolar system considered in the main text , we also calculate the spectrum flow of the system with coulomb interaction under twist boundary conditions . in fig.[sfig1](d)-(f ) , we show the spectrum flow as the function of @xmath52 for the system described by eq.([coulomb ] ) with @xmath14 , @xmath26 , @xmath27 , @xmath28 , @xmath34 and different @xmath13 under twist boundary conditions . as shown in fig.[sfig1](d ) , when @xmath45 , the lower energy levels overlap together with the change of @xmath52 . when @xmath126 , the lowest three energy spectra are separated from the higher excited states with a small gap , while they flow into each other with the change of @xmath52 . for @xmath56 ,
the lowest three states are nearly degenerate , and the nearly degenerate ground states are well separated from the higher excited states by an obvious gap .
in contrast to the dipolar fermi system , a smaller @xmath13 is needed for the emergence of degenerate ground states due to the coulomb interaction is much stronger than the dipole - dipole interaction . following the procedure in the main text
, we can also verify that the total chern number of the three degenerate ground states is @xmath70 as the dipolar system .
such an example indicates that the fractional topological phase can also appear in one - dimensional superlattice systems with the coulomb long - range interactions . in ref.@xcite
, fractional topological states are found in a two - dimensional lattice system with topological nontrivial flat band , in which the short - range interaction is used . while in our system the energy band
is split into three bands with finite band widths , the long - range interaction plays an important role in the formation of fractional topological states . to see it clearly , we consider the fermi system trapped in the one - dimensional bichromatic optical lattices with a short - range interaction @xmath132 where the last term of eq.[eqn6 ] describes the short - range interaction with @xmath128 controlling the range of the interaction and the prime denoting @xmath133 due to the pauli principle .
the hamiltonian can be also diagonalized under the twist boundary condition . in order to compare with the dipolar system considered in the main text
, we calculate the spectrum flux of the hamiltonian ( [ eqn6 ] ) with @xmath134 , @xmath103 and @xmath104 . in fig.[sfig2]a - fig.[sfig2]c
, we show the spectrum flux for the system with @xmath31 , @xmath14 , @xmath27 , @xmath28 , @xmath34 and @xmath129 , respectively .
in contrast to fig.3f in the main text , the lowest three energy levels are not separated with the higher excited states and no a degenerate ground - state manifold is formed for @xmath31 . moreover ,
the lowest three levels and the exited ones are not well separated even for a much larger @xmath130 with @xmath131 . as shown in fig.[sfig2]d ,
the spectrum flux is almost the same as the one displayed in fig.[sfig2]c .
our results show that no 3-fold degenerate ground states can be found and no fractional topological states emerge for the superlattice system with short - range interactions even in the strongly interacting limit . for system with @xmath14 , @xmath27 , @xmath28 , and @xmath135 under the periodic boundary condition .
( b ) @xmath136 and @xmath137 as the function of @xmath9 for the system with @xmath14 , @xmath27 , @xmath28 , @xmath31 , @xmath34 and @xmath42 under the periodic boundary condition .
( c)-(f ) spectrum flux as the function of @xmath52 for the system with @xmath14 , @xmath27 , @xmath28 , @xmath31 , @xmath34 , @xmath42 and different @xmath9 under the twist boundary condition .
here , @xmath138 , @xmath139 , @xmath140 , @xmath104 for ( c)-(f ) , respectively.,width=453,height=340 ] it is worth to point that the periodically modulated potential provides an additional dimension to ensure the topologically nontrivial property in our system .
such a one - dimensional system can be mapped to the lattice version of the two - dimensional integer hall effect ( the hofstadter problem ) @xcite by substituting @xmath141 , @xmath142 , @xmath143 and @xmath144 where @xmath145 is the magnetic flux per unit cell @xcite .
it is well known that the later model has nontrivial topological properties . in the main text
, we choose the amplitude of the periodic potential as @xmath26 . to understand how the amplitude of @xmath9 affects our calculation , we present the single - particle spectra as a function of @xmath9 for system with @xmath14 , @xmath27 , @xmath28 and @xmath135 under the periodic boundary condition in fig.[sfig3](a ) .
it is shown that the spectra are split into three bands and the gap between the first and second band becomes narrow when one decreases @xmath9 .
to get degenerate ground states which are well separated from the higher excited states , one need choose a suitable @xmath9 to generate an obvious gap between the first and second band . in fig.[sfig3](b )
, we show the width of lowest three states defined as @xmath146 and the energy gap between the fourth state and third state defined as @xmath147 versus the strength of modulation potential @xmath9 for the system with @xmath14 , @xmath27 , @xmath28 , @xmath31 and @xmath34 , where @xmath72 , @xmath148 , @xmath149 and @xmath150 denote energies of the lowest four eigenstates . from fig.[sfig3](b )
, it can be seen that @xmath136 decreases whereas @xmath137 increases with the increase of @xmath9 . in the limit of @xmath151
, we can find that @xmath136 and @xmath137 have the same scale , which is not favorable for the formation of degenerate ground states .
fig.[sfig3](c)-(f ) show the spectrum flux for the system with @xmath14 , @xmath27 , @xmath28 , @xmath31 , @xmath34 and different @xmath9 . for @xmath138 ,
the lowest three states are not well separated from the higher states . with the increase of @xmath9 ,
the gap becomes more obvious .
for @xmath26 chosen in the main text for the calculation , the lowest three states are nearly degenerate with an obvious gap separated them from the higher states . | we study the properties of dipolar fermions trapped in one - dimensional bichromatic optical lattices and show the existence of fractional topological states in the presence of strong dipole - dipole interactions .
we find some interesting connections between fractional topological states in one - dimensional superlattices and the fractional quantum hall states : ( i ) the one - dimensional fractional topological states for systems at filling factor @xmath0 have @xmath1-fold degeneracy , ( ii ) the quasihole excitations fulfill the same counting rule as that of fractional quantum hall states , and ( iii ) the total chern number of @xmath1-fold degenerate states is a nonzero integer .
the existence of crystalline order in our system is also consistent with the thin - torus limit of the fractional quantum hall state on a torus .
the possible experimental realization in cold atomic systems offers a new platform for the study of fractional topological phases in one - dimensional superlattice systems . _
introduction.- _ fractional quantum hall ( fqh ) effects have attracted intensive studies in the past decades as an important subject in condensed matter physics .
the traditional fqh states were realized in two - dimensional ( 2d ) electron gases within a strong external magnetic field .
in addition to 2d electron gases , great effort has been made to study quantum hall effects in some other physical systems , for example , lattice systems without a magnetic field and cold atomic systems .
effective landau levels can be realized in cold atomic systems in the presence of a rapidly rotating trap @xcite or a laser - induced gauge field @xcite . because of the existence of long - range interaction , the dipolar fermi gas is a good candidate to realize fqh states .
the fqh effects in a 2d dipolar fermi gas with either isotropic @xcite or anisotropic dipole - dipole interaction ( ddi ) @xcite have been studied recently .
as most of the previous studies on topological nontrivial states focus on 2d systems @xcite , the one - dimensional ( 1d ) systems attracted less attention until very recently @xcite .
although 1d systems without any symmetry are generally topologically trivial , the 1d superlattice model was recently found to be topologically nontrivial @xcite as the periodic parameter in these superlattice models can be considered as an additional dimension and thus the systems may have a nontrivial chern number in an effective 2d parameter space .
it has been shown that the 1d superlattice system with subbands being filled is not a trivial band insulator but a topological insulator with a nonzero chern number @xcite , which can be viewed as a correspondence of the integer quantum hall state of a 2d square lattice @xcite in the reduced 1d system .
it is well known that the fqh effect emerges from the integer quantum hall state in the presence of strong long - range interactions .
it is natural to ask whether a fractional topological state is available for the 1d superlattice system when interactions are included . in this letter
, we explore the nontrivial topological properties of dipolar fermions in 1d bichromatic optical lattices , which can be realized in cold atom experiments by loading the dipolar fermions into the lattice superposed by two 1d optical lattices with different wavelengths @xcite .
the noninteracting part of our hamiltonian is the recently studied 1d superlattice model with topologically nontrivial bands @xcite .
the presence of dipolar interactions breaks down the band description within a noninteracting picture . to characterize topological features of the interacting system ,
we study the low - energy spectrum and the topological chern number of the dipolar system based on exact calculation of finite - size systems . the existence of nontrivial topological states for the strongly interacting system at fractional filling is demonstrated by the topological degeneracy and nontrivial chern number of the low - energy states .
particularly , recent progress in manipulating ultracold polar molecules @xcite offers the possibility of exploring exotic quantum states of fermi gases with strong dipolar interactions in the topologically nontrivial optical lattices .
_ model hamiltonian.- _ we consider a 1d fermi gas with ddis in a bichromatic optical superlattice : @xmath2 with @xmath3 where @xmath4 ( @xmath5 ) is the creation ( annihilation ) operator of fermions , @xmath6 the density operator , and @xmath7 the hopping strength . here
@xmath8 is the periodic potential with @xmath9 being the modulation amplitude , @xmath10 determining the modulation period and @xmath11 being an arbitrary phase . the last term of eq .
( [ eqn1 ] ) is for ddis which are long - range interactions decayed with @xmath12 with @xmath13 the strength of ddi . for convenience
, we shall set @xmath14 as the unit of energy and choose @xmath15 . in the absence of interactions
, it has been demonstrated that the system with its subbands fully filled by fermions is an insulator with a nontrivial topological chern number in a 2d parameter space spanned by momentum and the phase of @xmath11 @xcite .
if the subband is only partially filled , the system is a topologically trivial conductor . in this letter , we shall study the case with the lowest band being partially filled by fermions subjected to the long - range interaction .
given that the number of fermions is @xmath16 and the lattice size is @xmath17 , the filling factor is defined as @xmath18 with @xmath19 being the number of primitive cells . while @xmath20 corresponds to the lowest band being fully filled , in this letter we shall consider the system with a fractional filling factor , for example , @xmath21 and @xmath22 .
_ low - energy spectrum and ground state degeneracy.- _ in the presence of the long - range ddi term , we diagonalize the hamiltonian ( [ eqn1 ] ) in each momentum subspace with @xmath23 under the periodic boundary condition ( pbc ) @xcite , where @xmath24 takes @xmath25 . for finite - size systems with
a given fractional filling , we study the change of low - energy spectrum with the increase in the interaction strength @xmath13 . in fig.[fig1 ] , we display the low - energy spectrum in momentum sectors for systems with @xmath14 , @xmath21 , @xmath26 , @xmath27 , @xmath28 , and different @xmath13 .
various cases with particle numbers @xmath29 are shown in the same figure .
when @xmath13 is small , it is hard to distinguish the lowest states apart from the higher excited states by an obvious gap .
when @xmath13 exceeds @xmath30 , the lowest three states tend to form a ground - state ( gs ) manifold with an obvious gap separating them from higher ones .
as @xmath13 increases further , the gap becomes more obvious and the lowest three states become nearly degenerate at @xmath31 .
the threefold degeneracy does not depend on particle numbers , but is only relevant to the filling factor @xmath21 . in the large @xmath13 case
, the lowest three states in the gs manifold always appear at some determinate positions in the momentum space . for cases of @xmath32 ( even ) ,
the total momenta locate at @xmath33 , whereas for @xmath34 ( odd ) , at @xmath35 .
we observe that one can make a connection between the momenta in our system and the orbital momenta of the fqh system in the thin - torus limit by setting @xmath36 with @xmath37 being the number of the flux quanta .
according to the exclusion rule known from the thin - torus limit of the fqh system @xcite , the total momenta of the @xmath1-fold degenerate gss emerge at @xmath38\mathrm{mod}n_{\mathrm{cell}}\}/n_{\mathrm{cell}}$ ] , where @xmath39 for a system with @xmath0 .
we also check the low - energy spectra for systems with @xmath14 , @xmath22 , @xmath26 , @xmath27 , @xmath28 , and different @xmath13 .
similar to cases of @xmath21 , both systems with @xmath40 and @xmath34 show the fivefold degenerate gs manifold in the large @xmath13 limit and the positions of momenta fulfill the above expression determined by the exclusion rule . , @xmath14 , @xmath26 , @xmath27 , @xmath28 and different @xmath13 under the pbc .
( a)-(f ) @xmath13 takes 0 , 1 , 10 , 50 , 300 , and 500 , respectively.,width=377,height=340 ] _ quasihole excitation spectrum.- _ for fqh states , the existence of quasihole excitations , which fulfill fractional statistics @xcite , is an important characteristic feature of the system .
according to the general counting rule @xcite , the number of quasiholes for the fqh system of @xmath21 reads as @xmath41 .
next we study the quasihole excitations by removing a particle from our system and check whether a similarity to the fqh system exists . on the left part of fig .
[ fig2 ] , we show the quasihole excitation spectra for the system with @xmath40 and @xmath42 produced by removing a particle from the system of @xmath21 with @xmath34 and @xmath42 , whereas the right part of fig.[fig2 ] gives spectra for @xmath34 and @xmath43 by removing a particle from the system of @xmath21 with @xmath44 and @xmath43 .
for both parts , from top to bottom @xmath45 , @xmath30 , and @xmath46 , respectively . in the regime of small
@xmath13 , the number of quasihole excitation is much larger than that given according to the above accounting rule of fqh systems . as @xmath13 increases
, the low - energy parts are excited into the upper part . for @xmath31 , as shown in fig.[fig2](c ) , below the gap , the number of quasihole excitations is @xmath47 with two states on each momentum sector . in fig.[fig2](f ) , the total number of states under the dash line is @xmath48 and for momentum sectors @xmath49\mathrm{mod}3=0 $ ] , the number of states below the gap is four while in the other parts is three , due to the finite - size effect . for both cases with @xmath31 ,
the total number of states below the gap is consistent with the number obtained by the counting rule for the @xmath21 fqh state . ,
@xmath42 and @xmath50 from top to bottom .
the right part is for @xmath34 , @xmath43 and @xmath50 from up to down .
, width=340,height=302 ] _ topological feature of ground - state manifold.- _ to characterize the topological feature of the many - body states , we introduce the twist boundary condition @xmath51 , where @xmath52 is the introduced phase factor . under the twist boundary condition , the momentum @xmath53 in brillouin zone
gets a shift @xmath54 with @xmath55 .
correspondingly , the energies vary continuously with the change of @xmath52 . in fig.3
, we show the low energy spectra as a function of @xmath52 ( the spectrum flux ) at a fixed @xmath28 for systems with @xmath40 [ ( a)-(c ) ] and @xmath34 [ ( d)-(f ) ] . in the small @xmath13 regime of @xmath45
, the lower energy levels overlap together with the change of @xmath52 . for @xmath56 , the lowest three energy spectra flow into each other but are already separated from the higher states . for @xmath31 , the lowest three states are nearly degenerate , and the gs manifold is well separated from the higher states by a gap .
similarly , if the phase @xmath11 varies from @xmath57 to @xmath58 , the spectrum for a given @xmath52 changes continuously with the gs manifold well separated from the other states , which indicates the robustness of gs manifold in the large @xmath13 regime .
an example for @xmath59 corresponding to the pbc is given in fig.4(a ) .
for systems with @xmath14 , @xmath21 , @xmath26 , @xmath27 , @xmath28 and different @xmath13 .
( a)- ( c ) @xmath40 , @xmath45 , @xmath30 , and @xmath46 , respectively .
( e)-(f ) @xmath34 , @xmath45 , @xmath30 , and @xmath46 , respectively .
, width=340,height=340 ] in the 2d parameter space of @xmath60 , the chern number of the many - body state is defined as an integral invariant @xmath61 , where @xmath62 is the berry curvature @xcite .
considering the system with @xmath31 shown in fig.1(f ) , we calculate the chern numbers of the lowest three nearly degenerate states in the gs manifold . for system of @xmath40 ,
the chern numbers of the three states are @xmath63 , @xmath64 , and @xmath65 , respectively . for each state , the chern number is not an integer , but their sum is an integer @xmath66 .
similarly , for @xmath34 , we have @xmath67 , @xmath68 , and @xmath69 with their summation being @xmath70 .
the existence of a nonzero total chern number characterizes the system at fractional filling having nontrivial topological properties .
effectively , the total chern number is shared by the @xmath71 degenerate states , which is similar to the fqh system with its @xmath71-fold gss sharing an integer total chern number @xcite . for the system with @xmath42 , @xmath34 , @xmath14 , @xmath26 , @xmath27 , @xmath31 . here
@xmath72 represents the gs energy of the system .
( a ) is for the pbc ; ( b ) is for the obc ; ( c)-(f ) is the density distribution for the in - gap mode with @xmath73 , @xmath74 , @xmath75 , and @xmath76 , respectively.,width=302,height=302 ] the emergence of edge states under the open boundary condition ( obc ) is generally characteristic of topologically nontrivial phases . in fig.4(b ) , we show the low - energy spectra as a function of phase @xmath11 under the obc , which is obtained by setting the hopping amplitude between the first and @xmath17-th site as zero .
in contrast to the spectra under the pbc , inside the gap regime , there emerge edge modes which connect the gs and excitation branch of the bulk spectrum as one varies phase @xmath11 . as shown in the fig.4(c)-fig.4(f ) , the state is adiabatically changed with @xmath11 varying from @xmath77 to @xmath78 .
the density distribution @xmath79 shows that the in - gap state with @xmath73 is pinned down on the right edge , whereas the state with @xmath80 is pinned down on the left edge . of the lowest three energy states versus momentum @xmath53 for the system with different @xmath13 . here ,
@xmath81.,width=340,height=302 ] the density distributions shown in fig.4(c)-fig.4(f ) already display the signature of a crystalized phase . to reveal the crystalline character of systems with the pbc
, we introduce the static structure factor @xmath82 defined as @xmath83,\ ] ] where @xmath84 is the sum of the particle number operators in the @xmath85th primitive cell and @xmath86 .
figure [ fig5 ] shows the static structure factor of the three lowest eigenstates versus momentum for the system with @xmath44 , @xmath43 , and different @xmath13 . in the strong repulsive limit ,
peaks emerge at @xmath87 for all the lowest three eigenstates . with the increases of @xmath13 , the height of the peaks increase and the central parts of the @xmath82 decrease dramatically , which suggests that these states are crystalized with a periodic structure in the large @xmath13 limit .
the existence of topologically nontrivial crystallized phase in our system is consistent with previous works on the evolution of fqh states on a torus @xcite , which have shown that the fqh state on a torus is adiabatically connected to a crystallized phase as the 2d system is deformed to the 1d thin - torus limit .
the emergence of fractional topological states is consequence of interplay of nontrivial topology of superlattices and long - range interactions , which is illustrated by the shift of edge mode from one to the other edge driven by the phase @xmath11 .
while the long - range interaction is responsible for the formation of degenerate gs manifold , the nontrivial topology of superlattices guarantees the existence of edge states .
consequently , when the pbc is changed to the obc , the threefold degenerate gss are lifted and one of them develops into the edge mode as shown in fig.4(b ) .
this is quite different from the noninteracting case , for which the gs is nondegenerate and nontrivial edge states only appear at the integer filling . to see clearly the effect of long - range interaction
, we also check cases with the coulomb interaction and short - range interactions @xcite .
while our conclusions also hold true for the case with coulomb interaction , no degenerate gs manifold and fractional topological states are found for the case with short - range interactions even in the strongly interacting limit . in summary
, we demonstrate the existence of fractional topological states for dipolar fermions in topologically nontrivial 1d superlattices , which are characterized by the gs degeneracy , nontrivial total chern number of gss , and quasihole excitations fulfilling the same counting rule as the fqh states .
the existence of crystallized order in the 1d fractional topological phases is also identified by calculating the structure factor .
our study provides a way of creating nontrivial fractional topological states by trapping the dipolar fermions in 1d bichromatic optical lattices which are realizable in current cold atomic experiments .
we thank x. wan and z. liu for helpful discussions .
this work has been supported by national program for basic research of most , the nsf of china under grants no.11174360 and no.11121063 , and 973 grant .
n. r. cooper , adv .
phys . * 57 * , 539 ( 2008 ) .
a. l. fetter , rev .
mod .
phys . * 81 * , 647 ( 2009 ) .
y .- j .
lin , r. l. compton , k. jimnez - garcia , j.v .
porto , and i. b. spielman , nature ( london ) * 462 * , 628 ( 2009 ) .
k. osterloh , n. barbern , and m. lewenstein , phys .
rev .
lett . * 99 * , 160403 ( 2007 ) .
r .- z .
qiu , f. d. m. haldane , x. wan , k. yang , and s. yi , phys .
rev .
b * 85 * , 115308 ( 2012 ) ; r .- z .
qiu , s .-
p .
kou , z .- x .
hu , xin wan , and s. yi , phys .
rev .
a * 83 * , 063633 ( 2011 ) .
k. osterloh , m. baig , l. santos , p. zoller , and m. lewenstein , phys .
rev .
lett . * 95 * , 010403 ( 2005 ) . l. b. shao , s .- l .
zhu , l. sheng , d. y. xing , and z. d. wang , phys
. rev .
lett .
* 101 * , 246810 ( 2008 ) .
m. z. hasan and c. l. kane , rev .
mod .
phys . * 82 * , 3045 ( 2010 ) ; x .-
l .
qi and s .- c .
zhang , rev .
mod .
phys . * 83 * , 1057 ( 2011 ) .
l .- j .
lang , x. m. cai , and s. chen , phys .
rev .
lett . * 108 * , 220401 ( 2012 ) .
y. e. kraus , y. lahini , z. ringel , m. verbin , and o. zilberberg , phys .
rev .
lett . * 109 * , 106402 ( 2012 ) .
f. mei , s. l. zhu , z. m. zhang , c. h. oh , and n. goldman phys .
rev .
a * 85 * , 013638 ( 2012 ) .
d. j. thouless , m. kohmoto , m. p. nightingale , and m. den nijs , phys .
rev .
lett .
* 49 * , 405 ( 1982 ) .
d. r. hofstadter , phys .
rev .
b * 14 * , 2239 ( 1976 ) .
l. fallani , j. e. lye , v. guarrera , c. fort , and m. inguscio , phys .
rev .
lett . * 98 * , 130404 ( 2007 ) .
g. roati _
et al_. , nature ( london ) * 453 * , 895 ( 2008 ) ; b. deissler _ et al_. , nat . phys .
* 6 * , 354 ( 2010 ) .
k .- k .
ni , s. ospelkaus , m. h. g. de miranda , a. peer , b. neyenhuis , j. j. zirbel , s. kotochigova , p. s. julienne , d. s. jin , and j. ye , science * 322 * , 231 ( 2008 ) ; s. ospelkaus , a. peer , k .- k .
ni , j. j. zirbel , b. neyenhuis , s. kotochigova , p. s. julienne , j. ye , and d. s. jin , nat . phys . * 4 * , 622 ( 2008 ) .
see supplemental material for details .
n. regnault and b. a. bernevig , phys .
rev .
x * 1 * , 021014 ( 2011 ) .
e. j. bergholtz and a. karlhede , phys .
rev .
b * 77 * , 155308 ( 2008 ) .
r. b. laughlin , phys .
rev .
lett .
* 50 * , 1395 ( 1983 ) .
b. i. halperin , phys .
rev .
lett .
* 52 * , 1583 , ( 1984 ) ; d. arovas , j. r. schrieffer , and f. wilczek , phys .
rev .
lett . * 53 * , 722 ( 1984 ) .
q. niu , d. j. thouless , and y. s. wu , phys .
rev .
b * 31 * , 3372 ( 1985 )
. d. n. sheng , z. c. gu , k. sun and l. sheng , nat . commun .
* 2 * , 389 ( 2011 ) .
a. seidel , h. fu , d .- h .
lee , j. m. leinaas , and j. moore , phys .
rev .
lett . * 95 * , 266405 ( 2005 ) .
e. j. bergholtz and a. karlhede , phys .
rev .
lett . * 94 * , 026802 ( 2005 ) .
e. j. bergholtz and a. karlhede , j. stat .
mech .
( 2006 ) l04001 ; a. seidel and d .- h .
lee , phys .
rev .
lett .
* 97 * , 056804 ( 2006 ) . |
glassiness and jamming in grains and colloids entails mechanical solid - like behavior accompanied by divergent relaxation times , apparently without spatial order @xcite .
a rather simple way to theoretically study such non - equilibrium transitions is by lattice models in which disorder and slow ( or nonexistent ) dynamics arise naturally from the underlying static or dynamic local rules of the models @xcite .
thus , a promising approach for describing the glass and jamming transitions has evolved in recent years around _ kinetically - constrained models _
@xcite . the first such model due to fredrickson and andersen ( fa ) @xcite considers non - interacting spins on an ordered lattice such that each spin can flip only if at least some number @xmath0 of its neighbors are up .
kob and andersen ( ka ) later suggested the corresponding lattice - gas model @xcite .
although these two models were thought to exhibit a glass transition , it was recently shown that they jam only due to finite - size effects @xcite .
namely , for any temperature ( fa ) or density ( ka ) there is a finite length - scale , such that in systems larger than this scale , all particles eventually move .
subsequently , two - dimensional ( 2d ) jamming - percolation models were introduced and proven to jam at finite temperature ( spin models ) or density ( lattice gases ) even in the thermodynamic limit @xcite . in such models ,
the kinetic rules defining which particles are constrained from flipping ( or moving ) depend not only on the number of up ( or occupied ) neighbors but also on their relative positions . consistently with the glass and jamming scenarios , these models exhibit a phase - transition of mixed nature @xcite ; the fraction of frozen particles jumps discontinuously , yet there are critical scaling and diverging length- and time - scales .
glass- or jamming - transitions result from the fact that particles are _ blocked _ by their neighbors , which in turn are blocked by their neighbors , and so on , hence forming large clusters of particles that can not move . in jamming - percolation models ,
the blocked particles form similar clusters , which above the critical density become infinite and span the entire system , thus a finite fraction of the system is completely _
frozen_. in spite of the qualitative differences between percolation in two and three dimension , most work so far was on 2d models , and it is not clear to what extent the results obtained there would be relevant for actual 3d systems . in this letter
we propose a 3d kinetically - constrained model , for which we prove there is a mixed - order jamming transition at finite density .
our numerical investigation supports the theoretical predictions we give here that the static properties of this model are qualitatively similar to those of the 2d spiral model @xcite , whereas its dynamics are qualitatively different . in both models
, jamming occurs because 1d strings of blocked particles span the system . in 2d
, these strings of frozen particles confine the mobile particles in compact domains , see fig .
[ fig : face]a , hence these can not diffuse on long time scales .
our 3d model is also jammed by 1d system - spanning frozen strings .
however , the mobile particles in this case can use the third dimension to bypass the frozen strings , see fig . [
fig : face]b , and thus they exhibit long - time diffusive behavior even above the critical density for jamming . some attempts have been previously made to extend several force - balance - percolation models to 3d @xcite , but to our knowledge , we are the first to provide exact results , and to study the dynamics of such models in three dimension .
the 2d spiral model @xcite was defined on the square lattice by dividing each site s 4 nearest- and 4 next - nearest - neighbors into four pairs of adjacent sites as depicted in fig .
[ fig : rules]a . here , the central site ( ) is unblocked if its ( ( w _ or _
e ) _ and _ ( s _ or _
n ) ) pairs do not contain any particles .
inspired by the 2d spiral model , we define our 3d model on the cubic lattice , and divide each site s 6 nearest- and 12 next - nearest - neighbors into 6 sets , each set consisting of one nearest neighbor and two next - nearest neighbors as shown in fig [ fig : rules]b .
we define the central site ( ) as unblocked if its ( ( w _ or _
e ) _ and _ ( s _ or _
n ) _ and _ ( b _ or _
t ) ) sets are completely empty . at each time step of the dynamics of both models
a particle is randomly chosen and moved to one of its randomly selected nearest - neighbor sites only if the target site is vacant and the particle is unblocked both before and after the move . here
we focus on these lattice - gas dynamics , in which there is sense to motion and hence to diffusion , yet our static results are equally valid for the ising variant of this model , in which individual unblocked sites can flip their state at some temperature - dependent rates .
our 3d kinetic constraint is clearly more restrictive than that of the 2d spiral model , since in each of the three orthogonal planes passing through a given site in the cubic lattice we have the 2d spiral rules .
thus for @xmath1 there is a finite fraction of permanently - frozen particles , with @xmath2 the critical density of 2d directed percolation ( dp ) , at which the 2d spiral model jams @xcite . before proving that our 3d model undergoes a discontinuous transition at the critical density @xmath3 @xcite of 3d dp , we present in fig .
[ fig : npf],a - b numerical results for the fraction of permanently frozen particles vs. density .
these indicate a jamming transition at @xmath4 in our 3d model and at @xmath5 in the 2d spiral model . of the lattices .
we will prove that in our model the number of frozen particles is macroscopic for @xmath6 , and that for @xmath7 in the thermodynamic limit there are no frozen particles .
together this implies that the fraction of frozen particles jumps discontinuously at @xmath8 from zero to a finite value .
our proof extends to three dimensions the corresponding theoretical work regarding the 2d spiral model @xcite . yet , as will become evident below , our extension is not as straightforward as expected .
we start by considering the directed cubic lattice formed by drawing arrows from each lattice site to its three t neighbors , see fig [ fig : tetrahedron]a .
this maps the kinetic constraint to 3d dp , thus for @xmath6 there is an infinite sequence of frozen particles which lie along a b - t path , and a finite fraction of frozen particles even in the thermodynamic limit . to prove that for @xmath9 there are no frozen particles ,
as opposed to the proof for the 2d spiral model , here we will demonstrate that with finite probability a small unblocked region can expand to unblock only six - eights of the system , and not all of it .
since in an infinite system there are an infinite number of such finite initially unblocked regions , the entire system can thus be unblocked . to simplify the presentation , we consider _ culling _ dynamics , in which unblocked particles are _ removed _ rather than _ moved _ to neighboring sites .
this immediately proves the transition for ising spin - flip dynamics , and we expect that the same results for the aforementioned lattice - gas dynamics should follow . around any lattice site @xmath10 , for any direction @xmath11 from it to one of its third order neighbors , with @xmath12 , and for any integer distance @xmath13 , we define the _ trirectangular tetrahedron _
@xmath14 as all sites @xmath15 for which @xmath16 , @xmath17 , @xmath18 , and @xmath19 .
the _ legs _ of this tetrahedron are its edges which lie along the @xmath20 , @xmath21 and @xmath22 directions , and its _ diagonals _ are its other three edges .
now assume that for a given @xmath23 all sites in @xmath24 are empty , and that the three diagonals of @xmath25 are empty , and ask whether all sites in the larger tetrahedron @xmath25 may also be emptied . fig .
[ fig : tetrahedron],b - c shows that for @xmath26 this is possible , yet for @xmath27 it is not .
one can verify that such expansion is possible for all @xmath28 vectors except for @xmath29 and @xmath30 .
we thus define the almost - octahedron @xmath31 as the union of the six tetrahedra @xmath32 for all @xmath28 vectors except for @xmath33 . applying the above result for each tetrahedron
, we deduce that @xmath34 may be emptied if @xmath35 is empty and all sites of the 18 diagonals of @xmath34 can be emptied .
( yellow ) and the diagonals of @xmath36 ( purple ) are empty , then for @xmath26 ( b ) the remaining sites ( green ) of @xmath36 may be emptied by the e - s - t sets in the order indicated by the numbers . for @xmath27 ( c )
the remaining sites have only one or two of their e , n and t sets empty as indicated in the figure , thus they are not necessarily emptiable . ] now , for a site in location @xmath37 along a diagonal of @xmath34 to be emptiable , there should not be a directed path from this site to any of the sides of the cube confining the tetrahedron to which it belongs .
the lengths of these three paths are @xmath38 , @xmath39 and @xmath13 .
thus , the probabilities of having them are @xmath40 , @xmath41 and @xmath42 , respectively ( see appendix a ) , with @xmath43 the dp correlation length , which diverges at @xmath8 . due to the positive correlations between probabilities of different sites on the diagonal to be emptiable , the probability that @xmath35 may be expanded to @xmath34 is bounded by ( see appendix b ) : @xmath44 with @xmath45 .
the probability to expand @xmath34 to infinity is thus @xmath46>0 , \label{eq : two}\end{aligned}\ ] ] where the last inequality follows from @xmath43 being finite .
thus for @xmath9 , with finite probability a small region may be expanded to an infinite almost - octahedron , and empty six - eights of the system .
now , any site @xmath15 in the lattice may be emptied if there is an almost - octahedron centered at any other site @xmath10 except for those with @xmath47 . as there is an infinite number of such @xmath48 sites
, any @xmath49 can be emptied , and for @xmath50 there are no blocked particles in the infinite - system - size limit . to conclude the proof
, we will now show that the fraction of frozen particles is discontinuous , namely at @xmath8 each site has a finite probability of being frozen . extending the proof for the 2d spiral model , we start with a tilted rhombus oriented along the s - n direction of some size @xmath51 , with @xmath52 . starting from the two far edges of this rhombus we construct the series shown in fig . [
fig : rhomboids ] of pairs of tilted rhombi of size @xmath53 that alternate in their long direction between the s - n and w - e directions , such that the far ends of each rhombus touch the next two rhombi , and with @xmath54 and @xmath55 .
the value @xmath52 is chosen so that different rhombi will not intersect .
the original site is frozen if all these rhombi have dp paths along their long directions and each path is connected to the previous path . for any density and for any length @xmath56 , there is some finite probability @xmath57 that the first rhombus contains a dp path of length @xmath56 in the s - n direction , and that this path continues until the edges of the two rhombi adjacent to the first one , so that its total length in the s - n direction is @xmath58 .
now , consider each of the adjacent rhombi and divide it into @xmath59 channels , each of size @xmath60 , with @xmath61 the dp exponent relating length and width of dp paths .
namely , typical dp clusters of parallel length @xmath62 have typical transverse length @xmath63 @xcite . by this construction , for @xmath6 , each such channel contains a dp path with some finite probability @xmath64 , which is independent of @xmath65 .
the path spanning the first rhombus passes through @xmath66 such channels , and so the probability that at least one of these channels will have a dp path spanning it is bounded from below by @xmath67 .
this process can be continued indefinitely since the probability that the original site is frozen is bounded from below by @xmath68 , which is finite .
we have shown that the fraction of frozen particles is discontinuous at @xmath8 . as discussed above
, the phase transition we observe in our 3d model also exhibits some features of continuous transitions . to probe time- and length - scales we run the following culling dynamics both for our 3d model and for the 2d spiral model .
we start with a randomly occupied lattice at a given density .
we then check the kinetic constraint on all sites and identify the mobile particles . in a single culling step all mobile particles are removed , and subsequently other particles blocked by them may become mobile .
these in turn are removed in the following step , and this process is continued iteratively until either the lattice is empty or no more particles can be removed .
we obtain a _ time scale _ from the average number of culling iterations @xmath69 needed to remove all movable particles .
this is related to the time it would take a particle , which is not permanently frozen to eventually relax under the system s actual dynamics .
clearly , @xmath70 is not directly proportional to the structural relaxation time or to the persistence time in the system , but we expect to find strong correlations between these two measures . near @xmath71
not many particles can initially move , and one has to cull many layers of particles until the frozen core is reached .
thus , @xmath69 peaks at @xmath71 , as shown in fig .
[ fig : npf],c - d .
system - size analysis shows that the peak culling time grows algebraically with system size , @xmath72 , see fig .
[ fig : exponents]a
. the exponents we find are @xmath73 in 2d that compares well with a previous study @xcite , and @xmath74 in 3d .
we probe a diverging _ length - scale _ at the transition by finite - size scaling .
[ fig : npf],a - b shows how the transition becomes sharper as the system size is increased . in the thermodynamic limit
, we expect the typical cluster - size to diverge as @xmath75 $ ] , where @xmath76 .
the values of the dp exponents are @xmath77 and @xmath78 @xcite , thus we expect @xmath79 .
a system of linear dimension @xmath80 jams at density @xmath81 for which this cluster size equals the system size : @xmath82\propto l$ ] , where @xmath83 is the theoretical value which we expect to obtain in the infinite system - size limit . inferring @xmath84 from @xmath81 is impractical because even for the largest systems we can simulate ( @xmath85 ) @xmath86 is too far from the 3d dp value of @xmath87 .
instead of looking at the system - size dependence of @xmath71 , we claim @xcite that the transition width @xmath88 is proportional to @xmath89 , and therefore expect to have @xmath90 .
this functional form agrees with our numerical results , see fig .
[ fig : exponents]b , but with scaling exponents @xmath91 which are about a factor of two smaller from the theoretical values given above .
this discrepancy has been previously observed for the 2d spiral model and attributed to the limited range of systems sizes probed in such simulations @xcite . with @xmath92 .
( b ) transition width fits @xmath93 with @xmath94 . ]
the static properties of our 3d model are qualitatively similar to those of the previously studied 2d spiral model .
interestingly , the dynamics of these two models are qualitatively different . in both models ,
the backbone of frozen particles is comprised of perpendicular sets of dp chains . in the spiral model ,
these chains form a 2d mesh , and mobile particles are confined within the cells of this mesh ( see fig .
[ fig : face]a ) .
thus , above the critical density @xmath95 , for which these percolating clusters of permanently - frozen particles appear , the mobile particles can diffuse only over finite distances , and the mean - square - displacement saturates with time . in our 3d model , this mesh is also comprised of 1d chains of permanently - frozen particles , hence the mobile particles can use the third dimension to travel between the cells of this mesh ( see fig . [
fig : face]b ) and the long - time behavior may be diffusive . fig .
[ fig : npf],e - f shows that in 3d diffusivity is indeed finite also above @xmath8 , whereas in 2d it vanishes at @xmath95 .
interestingly , the diffusion coefficient does not seem to exhibit any singularity at @xmath8 .
this is because below @xmath8 although all particles eventually move , many of them are already blocked by a hierarchy of many other particles ( as may be seen from the huge values of the culling time there ) , and the particle - averaged mean - square - displacement is dominated by the particles that do manage to move more quickly ( and do not require many culling iterations in order to be removed in the analysis of which particles are frozen ) . for the diffusion of these particles ,
no qualitative change occurs at @xmath8 , thus @xmath96 is smooth there . in our 3d model , above @xmath8 diffusing particles travel through windows in the mesh of string - like dp chains .
as the density is increased beyond @xmath8 , more particles are permanently frozen , thus these chains become thicker , until at some second critical density @xmath97 the windows in the mesh close , such that particles can no longer diffuse between neighboring cages , and the diffusivity should vanish .
@xmath97 is the critical density for directed percolation of a surface in 3d , which should be bounded from above by the critical density @xmath98 of surface percolation in 3d @xcite .
in summary , we propose the first 3d jamming - percolation model , and provide a theoretical proof that it exhibits a mixed - order jamming transition at a non - trivial critical density .
the system - size dependence of the transition width indicates that the length scale associated with the size of frozen clusters diverges faster than a power law .
the 3d divergent length - scale exponent is slightly smaller than in 2d .
many theoretical studies , including those based on lattice models , as well as recent experiments done on glassy or jammed matter use 2d or quasi-2d systems @xcite .
the behavior in lower dimensions may differ from that of the corresponding 3d systems .
a relevant question thus remains whether substantial differences occur when going from two to three dimensions and what would be the origin of such differences .
the dynamic results we present here are one example showing that diffusive behavior in 3d differs from that of 2d systems .
this occurs due to the way percolating structures span 3d systems , providing an additional dimension for the particles to diffuse in .
this dimensionality dependence , related to diffusion in porous media and in chemical gels , is quite general
. it would be interesting to test to what extent this is related to the well - known decoupling of self diffusion and viscosity in glasses .
it is certainly useful that we have a concrete glassy example where the diffusion coefficient is still non - zero but the structure does not relax anymore .
thus , our simple kinetically - constrained lattice model provides scope for future studies of various phenomena related to glassiness , jamming and dynamics in three dimensions .
we thank haye hinrichsen , dov levine , and andrea liu for helpful discussions , and amir natan for computing resources .
this research was supported by the israel science foundation grants no .
@xmath99 , @xmath100 .
we want to prove that in 3d dp , for @xmath9 the probability that a site is part of a directed path of @xmath101 sites is smaller than or equal to @xmath102 , where @xmath43 is finite and depends on the density .
let @xmath103^{3}$ ] be a cube of side @xmath104 , and @xmath105 its boundary .
we denote the number of sites on the boundary @xmath106 which have a directed path to them starting from the origin @xmath107 by @xmath108 .
the total number of sites in an infinite lattice that have a directed path to them from the origin is thus @xmath109 , where @xmath110 is the expectation value of @xmath108 .
since @xmath9 , the size @xmath111 of the dp cluster starting from the origin is finite .
hence , there must be some @xmath0 for which @xmath112 .
consider the probability @xmath113 that the origin has a directed path to the boundary of the cube @xmath114 with some @xmath115 , i.e. that the origin has a directed path of length @xmath116 starting from it .
this is equal to the probability that there is a directed path from the origin to some site @xmath49 on the boundary of a smaller cube @xmath117 and from that site to the boundary of the larger cube @xmath118 the probability that the site @xmath49 is connected to the larger boundary is smaller or equal to the probability that it is connected to the boundary of a cube with one corner at @xmath49 and the opposite corner at @xmath119 , because in order to reach the boundary @xmath120 it must first pass through the boundary of @xmath121 .
hence , @xmath122 the probabilities that the origin is connected to @xmath49 and that @xmath49 is connected to the boundary are independent , and so @xmath123 since we can choose @xmath0 such that @xmath112 . for any @xmath124 , we write @xmath125 where @xmath126 and @xmath38 are integers with @xmath127 .
thus @xmath128 , which decays exponentially with @xmath101 , as required .
we want to show that @xmath129 we first note that by symmetry @xmath130 .
such that @xmath131 .\end{aligned}\ ] ] the argument of the logarithm is smaller than @xmath132 for all @xmath133 , and so adding more terms to the sum would only decrease it .
hence we can write @xmath134 $ ] . in order to show that the infinite sum converges to a finite value , we change it to an integral over @xmath135 , @xmath136 , and thus @xmath137 , with @xmath138 . for finite @xmath43 and large @xmath13 , we may approximate @xmath139^{\ell e^{-\ell/\xi } } \nonumber\\ & \approx & \exp\left(-\ell e^{-\ell/\xi}\right ) .\end{aligned}\ ] ] j.p .
garrahan , p. sollich and c. toninelli , dynamical heterogeneities in glasses , colloids , and granular media .
l. berthier , g. biroli , j .-
bouchaud , l. cipelletti and w. van saarloos .
oxford university press ( 2011 ) | we introduce a three - dimensional model for jamming and glasses , and prove that the fraction of frozen particles is discontinuous at the directed - percolation critical density . in agreement with the accepted scenario for jamming- and glass - transitions ,
this is a mixed - order transition ; the discontinuity is accompanied by diverging length- and time - scales .
because one - dimensional directed - percolation paths comprise the backbone of frozen particles , the unfrozen rattlers may use the third dimension to travel between their cages .
thus the dynamics are diffusive on long - times even above the critical density for jamming . |
the boson @xmath0 discovered @xcite in 2012 by the cms and atlas collaborations was reported to have a mass about 125 gev and some decay modes predicted for the standard model ( sm ) higgs boson .
since that time , the observed particle , called the higgs boson , has been intensively studied ( see , for example , papers @xcite ) .
a main goal of experiments on the higgs boson physics has been to prove or disprove the hypothesis that @xmath0 is the sm higgs boson .
apart from the decay channels , the sm predicts that @xmath0 has @xmath13 .
the followed thorough analysis has fine - tuned the mass of @xmath0 , which is @xmath14 gev according to ref .
@xcite , and has yielded some information on its spin and its @xmath2 parity . in particular , the observation of the @xmath15 and @xmath16 modes ( see , for example , ref .
@xcite ) means that the higgs boson spin is 0 , 1 , or 2 while the fact that @xmath0 decays @xcite to @xmath17 and the landau - yang theorem exclude the spin - one variant .
further , the analyses presented in @xcite rule out many spin - two hypotheses at a 99% confidence level ( cl ) or higher . therefore , we conclude that the spin of the higgs boson is 0 with a probability of about 99% . to clarify the @xmath2 properties of @xmath0 , in ref .
we study the decay of a spin - zero particle @xmath1 into two off - mass - shell @xmath3 bosons @xmath4 and @xmath5 .
since @xmath1 is defined as an elementary neutral particle with zero spin , our study applies to the higgs boson .
moreover , it can apply to the dilaton if this boson actually exists .
the amplitude of the decay @xmath18 depends ( see eq .
( 4 ) in ) on 3 complex - valued functions of the invariant masses of @xmath4 and @xmath5 .
these functions determine the @xmath2 properties of the boson @xmath1 and are called the @xmath19 couplings . using the cms and atlas experimental data on the decay @xmath8 ( where @xmath9 stands for @xmath10 , @xmath11 , or @xmath12 ) , these collaborations in refs .
@xcite and we in ref .
have obtained some constraints on the @xmath20 couplings .
these constraints demonstrate that @xmath0 is not a @xmath2-odd state and it may be the sm higgs boson , another @xmath2-even state , or a boson with indefinite @xmath2 parity .
besides , as shown in , a non - zero imaginary part of the @xmath20 couplings is not excluded , which can be related to small loop corrections and possibly to a non - hermiticity of the @xmath20 interaction . thus , the @xmath2 parity of the higgs boson is not yet fully ascertained .
moreover , in some supersymmetric extensions of the sm there are @xcite neutral bosons with negative or indefinite @xmath2 parity .
that is why it is now important to establish the @xmath2 properties of the higgs boson .
aiming at that , we consider the decay of the particle @xmath1 into @xmath4 and @xmath5 which then decay to fermion - antifermion pairs @xmath21 and @xmath22 respectively . while in ref .
we study in detail the decays with the non - identical fermions , @xmath23 , in the present paper the case @xmath24 is under investigation .
the masses of the fermions @xmath25 and @xmath26 is neglected in both papers .
we are motivated to consider the decay into identical fermions by the following . in refs .
@xcite the cms and atlas collaborations analyze 95 decays @xmath8 .
53 of them are the decays to identical leptons , namely to @xmath10 or @xmath11 . in spite of the fact that the decays to the identical leptons make up
about 55% of the measured decays @xmath8 , the distributions of the former decays have not been properly analytically studied .
the sm total widths of the decays into identical fermions are studied in refs .
@xcite and are calculated in @xcite .
some distributions of the decay @xmath27 are plotted in @xcite for the sm higgs boson and some spin - zero states beyond the sm . in the present paper
we perform a more general study and consider the decay @xmath28 with allowance for all the possible @xmath2 properties of the particle @xmath1 . in sec .
[ section : the fully differential width ] we derive an analytical formula for the fully differential width of the decay to identical fermions .
section [ section : invariant mass and angular distributions ] shows a comparison of some distributions of the decay to identical leptons with those for the decay into non - identical ones . for this comparison
we obtain an exact analytical formula for a certain differential width of the decay to non - identical fermions ( see appendix [ appendix : a formula for d gamma over d a of the decay into non - identical fermions ] ) .
we analyze the usefulness of all the compared distributions for obtaining constraints on the @xmath20 couplings . in sec .
[ section : comparison with experimental data ] we derive some sm histogram distributions of the decay @xmath8 by monte carlo ( mc ) integration and compare them with the corresponding simulations presented in ref .
@xcite and with the experimental distributions from @xcite .
we consider a neutral particle @xmath1 with zero spin and arbitrary @xmath2 parity .
it can decay into two fermion - antifermion pairs , @xmath21 and @xmath22 , through the two off - mass - shell @xmath3 bosons ( @xmath4 and @xmath5 ) : @xmath29 if @xmath30 $ ] ( @xmath31 is the mass of the particle @xmath1 , @xmath32 is the mass of the @xmath33 quark , @xmath34 is the mass of the @xmath35 quark ) , which holds for @xmath36 , then @xmath37 . if @xmath38 , which is possible @xcite if @xmath1 is the dilaton , then @xmath39 can be the top quark as well . in ref .
we considered decays @xmath40 at the tree level .
the present paper shows our analysis of decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) in the case of the identical fermions , @xmath41 : @xmath42 the matrix element of decay ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) is @xmath43 where the matrix elements @xmath44 and @xmath45 correspond to the diagrams ( a ) and ( b ) in fig .
[ the feynman diagrams for the decay into identical fermions ] respectively .
namely , @xmath46 where * @xmath47 and @xmath48 ( @xmath49 and @xmath50 ) are the 4-momenta of the particles @xmath25 and @xmath51 ( @xmath26 and @xmath52 ) in the rest frame of @xmath1 ; * @xmath53 and @xmath54 are the 4-momenta of @xmath4 and @xmath5 respectively in the rest frame of @xmath1 in diagram fig .
[ the feynman diagrams for the decay into identical fermions ] ( a ) ; * @xmath55 ; * @xmath56 and @xmath57 are respectively the pole mass and the total width of the @xmath3 boson ; * @xmath58 is the amplitude of the decay @xmath18 where @xmath59 and @xmath60 are respectively the momentum and the helicity of the boson @xmath61 in the rest frame of @xmath1 ; * @xmath62 is the amplitude of the decay @xmath63 where @xmath64 and @xmath65 ( @xmath66 and @xmath67 ) are respectively the momentum and the polarization of @xmath68 ( @xmath69 ) in the rest frame of @xmath3 , @xmath70 is the helicity of decaying @xmath3 ; * @xmath71 and @xmath72 are the 4-momenta of @xmath4 and @xmath5 respectively in the rest frame of @xmath1 in diagram fig .
[ the feynman diagrams for the decay into identical fermions ] ( b ) ; * @xmath73 . from the conservation of the energy - momentum 4-vectors we find all the possible values of @xmath74 and @xmath75 : @xmath76 where @xmath77 is the mass of the fermion @xmath39 .
the amplitude @xmath58 is @xmath78 where @xmath79 , @xmath80 is the fermi constant , @xmath81 , @xmath82 , and @xmath83 are some complex - valued dimensionless functions of @xmath74 and @xmath75 , @xmath84 , @xmath85 is the polarization 4-vector of the @xmath3 boson with a momentum @xmath86 and a helicity @xmath70 , @xmath87 is the 4-momentum of the boson @xmath1 in its own rest frame , @xmath88 is the levi - civita symbol ( @xmath89 ) .
the values of the couplings @xmath90 , @xmath91 , and @xmath92 reflect the @xmath2 properties of the particle @xmath1 .
specifically , at the tree level the correspondence shown in table [ the cp parity of x in the case of various sets of values of a_z , b_z , and c_z ] takes place .
.the @xmath2 parity of the particle @xmath1 for various values of @xmath90 , @xmath91 , and @xmath92 .
[ cols="^,^,^,^",options="header " , ]
in this paper , we have considered the decay of a neutral particle @xmath1 with zero spin and arbitrary @xmath2 parity into two off - mass - shell @xmath3 bosons ( @xmath4 and @xmath5 ) each of which decays to identical fermion - antifermion pairs ( @xmath6 ) : @xmath7 .
analytical formulas for the fully differential width of the decay in question and for the fully differential width of the decay @xmath8 are derived ( see eqs .
( [ the fully differential width ] ) and ( [ an explicit formula for the fully differential distribution of the decay h to z1 z2 to 4 ell ] ) ) .
moreover , we present an exact formula for the differential width @xmath93 of a decay @xmath94 with @xmath23 ( see eq .
( [ an explicit formula for d gamma over d a for the decay into non - identical fermions ] ) ) . integrating eq .
( [ the fully differential width ] ) with a mc method , we have obtained some non - histogram distributions for any decay @xmath95 ( @xmath96 ) with @xmath97 .
these distributions are compared to those for the decay @xmath95 with @xmath98 ( see figs .
[ plots of the distributions of a1 and a2 ] and [ plots of the distributions of a , theta , and phi ] ) .
the comparison has revealed significant distinctions between the distributions for the case @xmath97 and the corresponding ones for @xmath98 .
the dependences shown in fig .
[ plots of the distributions of a , theta , and phi ] are calculated at 4 possible sets ( [ four sets of the possible values of a , b , c ] ) of values of the @xmath20 couplings @xmath90 , @xmath91 , and @xmath92 . at all
the 4 sets these distributions almost coincide
. therefore their measurement can yield notable constraints on @xmath90 , @xmath91 , and @xmath92 only if the distributions are measured at very high precision . in order to determine the usefulness of eq .
( [ the fully differential width ] ) , we have computed some sm histogram distributions of the decay @xmath8 by means of integration of eq .
( [ an explicit formula for the fully differential distribution of the decay h to z1 z2 to 4 ell with the atlas or cms limitations ] ) .
the distributions are calculated for atlas kinematical limitations ( [ the atlas limitations on m12 , m34 , eta_e , and eta_mu ] ) and for cms ones ( [ the cms limitations on m12 , m34 , eta_e , and eta_mu ] ) .
we have compared our distributions with the atlas experimental ones and the atlas mc simulated ones ( see ref .
the way our distributions are derived is almost purely analytical its only numerical part is integration of eq .
( [ an explicit formula for the fully differential distribution of the decay h to z1 z2 to 4 ell with the atlas or cms limitations ] ) .
besides , we have chosen the total number of decays more accurately than atlas during its simulations
. however , our calculation does not allow for the fact that the invariant mass of @xmath9 may differ from @xmath99 while this fact is taken into account in the atlas simulations .
the pros and cons of our technique and the atlas simulations make our distributions and the atlas simulated ones somewhat different but equally close to the atlas experimental data .
we have also presented our distributions of @xmath100 , @xmath101 , @xmath102 , @xmath103 , and @xmath104 for the kinematic conditions specific for cms . in summary , various distributions of the decays @xmath7 or @xmath8
have been obtained with a rather simple integration of eqs .
( [ the fully differential width ] ) and ( [ an explicit formula for the fully differential distribution of the decay h to z1 z2 to 4 ell ] ) respectively .
this way of calculation gives an alternative to more traditional mc simulation .
this research is partially supported by the national academy of sciences of ukraine ( project tso-1 - 4/2017 ) and the ministry of education and science of ukraine ( project
0115u000473 and project `` electromagnetic processes and processes involving short - lived particles in crystals at energies attainable at cern '' ) .
the fully differential width of decay ( [ x- > z_1^ * z_2^ * - > f antif f antif ] ) is @xmath105 , & \end{aligned}\ ] ] where @xmath106 & \end{aligned}\ ] ] is the fully differential width of decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) ( see eq .
( 5 ) in ref . ) , @xmath107 is the weak isospin projection of the fermion @xmath68 , @xmath108 , @xmath109 is the electric charge of @xmath68 , @xmath110 is the electric charge of the positron , @xmath111 is the weak mixing angle , @xmath112 , @xmath113 @xmath114 @xmath115 @xmath116 , @xmath117 is the momentum of the fermion @xmath25 in the center - of - momentum frame of the particles @xmath25 and @xmath52 , @xmath118 , @xmath119 and @xmath120 are any unit and mutually orthogonal vectors
such that @xmath121 , @xmath122 @xmath123 is the azimuthal angle of the @xmath25 momentum in the @xmath4 rest frame formed by the vectors ( @xmath119 , @xmath120 , @xmath124 ) , @xmath125 is the azimuthal angle of the @xmath26 momentum in the @xmath5 rest frame formed by the vectors ( @xmath119 , @xmath126 , @xmath127 ) , @xmath128 @xmath129 , @xmath130 is the momentum of the fermion @xmath26 in the center - of - momentum frame of the particles @xmath26 and @xmath51 , @xmath131 @xmath132 @xmath133 @xmath134 @xmath135 note that the dependence of expression ( [ the fully differential width ] ) on @xmath123 and @xmath125 reduces to a dependence on @xmath136 and in eq .
( [ the fully differential width ] ) the latter sum has to be substituted by @xmath104 .
it follows from eq .
( [ an explicit formula for d gamma over d a for any decay x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) that for any decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ,
f_1 neq f_2 ] ) @xmath137 where the differential width @xmath138 is determined by eq .
( 8) from ref . . if the functions @xmath139 , @xmath140 , @xmath141 , and @xmath142 are independent of @xmath74 and @xmath75 , integration of @xmath138 in eq
( [ a formula for d gamma over d a for the decay into non - identical fermions ] ) yields @xmath143 , & \end{aligned}\ ] ] where @xmath144 , @xmath145 , @xmath146 , in place of @xmath147 one may take either @xmath148 or @xmath149 ( this choice does not influence the dependence of @xmath93 on @xmath150 ) , @xmath151 we define the argument @xmath152 of a complex number @xmath153 as follows : @xmath154 where @xmath155 , @xmath156 , ~~\theta ( x ) \equiv 1 ~~ \forall x \in ( 0 , + \infty).\end{aligned}\ ] ] according to definition ( [ a definition of arg(z ) ] ) , @xmath157 .
in this appendix we propose some general definitions of the differential widths @xmath158 and @xmath159 for any decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) , and show that the differential widths defined this way coincide with those defined in the standard fashion for decays ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) and ( [ x- > z_1^ * z_2^ * - > f antif f antif ] ) separately . therefore , the distributions presented in fig . [ plots of the distributions of a , theta , and phi]a are general distributions defined for any decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) which are calculated for the decay into non - identical leptons and the distributions in fig .
[ plots of the distributions of a , theta , and phi]b are the same general distributions calculated for the decay into identical leptons .
thus , comparison of fig .
[ plots of the distributions of a , theta , and phi]a and fig .
[ plots of the distributions of a , theta , and phi]b is sensible thanks to the existence of the general definitions of @xmath158 and @xmath159 .
we define the function @xmath158 as @xmath160 where @xmath161 is the probability that in decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) there is a @xmath3 boson whose squared invariant mass lies in an interval @xmath162 $ ] . to derive an explicit formula for the distribution @xmath163
, we should recall that for decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ,
f_1 neq f_2 ] ) @xmath164 where @xmath165 , @xmath166 is the number of the decays ( [ x- >
z_1^ * z_2^ * - > f_1
antif_1 f_2 antif_2 , f_1 neq f_2 ] ) in which the squared invariant mass of @xmath4 ( @xmath5 ) is in an interval @xmath167 $ ] ( @xmath168 $ ] ) , the polar angle of @xmath25 ( @xmath26 ) lies in @xmath169 $ ] ( @xmath170 $ ] ) , and the azimuthal angle between the planes of the decays @xmath171 and @xmath172 is in an interval @xmath173 $ ] , among @xmath174 decays ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] )
. eq . ( [ a relation between the fully differential width and the fully differential number of decays for the decay to non - identical fermions ] ) is consistent with the fact that for any decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) @xmath175 because @xmath176 using eqs .
( [ a definition of 1 over gamma d gamma over d a ] ) and ( [ a relation between the fully differential width and the fully differential number of decays for the decay to non - identical fermions ] ) , we obtain that for decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) @xmath177 since if we neglect @xmath178 and @xmath179 , then @xmath180 ( see eq .
( 8) in ref . ) . for any decay ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) @xmath181 where @xmath182 is the number of the decays ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) in which there is a @xmath3 boson @xmath4 with a squared invariant mass lying in an interval @xmath167 $ ] and a @xmath3 boson @xmath5 whose squared invariant mass is in @xmath168 $ ] , the polar angle of @xmath25 ( @xmath26 ) lies in an interval @xmath169 $ ] ( @xmath170 $ ] ) , and the azimuthal angle between the planes of the decays @xmath171 and @xmath172 is in @xmath173 $ ] , among @xmath174 decays ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) .
note that while for decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) @xmath4 ( @xmath5 ) is defined as the @xmath3 boson decaying into @xmath21 ( @xmath22 ) , for decay ( [ x- > z_1^ * z_2^ * - > f antif f antif ] ) the choice of @xmath4 and @xmath5 is arbitrary , which leads to the difference between the definitions of @xmath166 and @xmath182 .
( [ a relation between the fully differential width and the fully differential number of decays for the decay to identical fermions ] ) accords with eq .
( [ a formula for gamma of any decay x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) due to the fact that @xmath183 the `` 2 '' in the right - hand side of eq .
( [ an integral of d^5 n_f_1 = f_2 over d^5 p ] ) emerges because of the double counting during the integration of @xmath184 on @xmath74 and @xmath75 .
it follows from eqs .
( [ a definition of 1 over gamma d gamma over d a ] ) and ( [ a relation between the fully differential width and the fully differential number of decays for the decay to identical fermions ] ) that for decay ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) @xmath185 combining eqs .
( [ an explicit formula for 1 over gamma d gamma over d a for the decay into non - identical fermions ] ) and ( [ an explicit formula for 1 over gamma d gamma over d a for the decay into identical fermions ] ) , we infer that in the approximation @xmath186 for any decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) @xmath187 analogously , we define the differential width @xmath159 as @xmath188 where @xmath189 is the probability that in decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) there is a fermion whose polar angle lies in an interval @xmath190 $ ] .
( [ a definition of 1 over gamma d gamma over d theta ] ) and ( [ a relation between the fully differential width and the fully differential number of decays for the decay to non - identical fermions ] ) yield that for any decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) @xmath191 according to eq .
( [ the fully differential width for the decays with f_1 different from f_2 ] ) , the differential width @xmath192 of decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) is invariant under the substitution @xmath193 and @xmath194 if @xmath195 ( see eq .
( [ a definition of a_f ] ) for the definition of the quantity @xmath196 ) .
that is why for decay ( [ x- >
z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 , f_1 neq f_2 ] ) in the case @xmath195 @xmath197 and therefore @xmath198 we find from eqs .
( [ a definition of 1 over gamma d gamma over d theta ] ) and ( [ a relation between the fully differential width and the fully differential number of decays for the decay to identical fermions ] ) that for decay ( [ x- >
z_1^ * z_2^ * - > f antif f antif ] ) @xmath199 combination of eqs .
( [ an explicit formula for 1 over gamma d gamma over d theta for the decay into non - identical fermions ] ) and ( [ an explicit formula for 1 over gamma d gamma over d theta for the decay into identical fermions ] ) yields that for any decay ( [ x- > z_1^ * z_2^ * - > f_1 antif_1 f_2 antif_2 ] ) wherein @xmath195 @xmath200
it follows from eqs .
( [ a relation between the fully differential width and the fully differential number of decays for the decay to identical fermions ] ) , ( [ a relation between the fully differential width and the fully differential number of decays for the decay to non - identical fermions ] ) , ( [ the atlas limitations on m12 , m34 , eta_e , and eta_mu ] ) , and ( [ the cms limitations on m12 , m34 , eta_e , and eta_mu ] ) that @xmath201 where @xmath202 is the number of the decays @xmath203 selected by atlas or cms , among @xmath204 decays @xmath203 , @xmath205 @xmath206 , @xmath207 is the fully differential width of the decay @xmath203 ( see eqs .
( [ the fully differential width ] ) and ( [ the fully differential width for the decays with f_1 different from f_2 ] ) ) , @xmath208 @xmath209 @xmath210 moreover , the fully differential distribution of the decay @xmath8 is @xmath211 where * @xmath212 is the number of the decays @xmath8 in which @xmath213 $ ] , @xmath214 $ ] , the polar angle of @xmath25 ( @xmath26 ) lies in an interval @xmath169 $ ] ( @xmath170 $ ] ) , and the azimuthal angle between the planes of the decays @xmath171 and @xmath172 is in @xmath173 $ ] , among @xmath215 decays @xmath8 ; * @xmath216 is the number of the decays @xmath217 in which @xmath213 $ ] , @xmath214 $ ] , the polar angle of @xmath25 ( @xmath26 ) lies in an interval @xmath169 $ ] ( @xmath170 $ ] ) , and the azimuthal angle between the planes of the decays @xmath171 and @xmath172 is in @xmath173 $ ] , among @xmath215 decays @xmath8 .
it follows from eqs .
( [ a relation between the fully differential width and the fully differential number of decays for the decay to non - identical fermions ] ) and ( [ the fully differential width for the decays with f_1 different from f_2 ] ) that @xmath218 using eqs .
( [ some formulas for the fully differential distribution of the decay h to z1 z2 to 4 ell ] ) and ( [ d^5 n_2 e 2 mu^prime = 2 d^5 n_2 e 2 mu ] ) , we derive that @xmath219 integration of eq .
( [ an explicit formula for the fully differential distribution of the decay h to z1 z2 to 4 ell ] ) yields @xmath220 where @xmath221 is the number of the decays @xmath8 selected by atlas or cms , among @xmath215 decays @xmath8 . g. aad et al .
( atlas collaboration ) , phys .
b * 716 * , 1 ( 2012 ) ; + s. chatrchyan et al .
( cms collaboration ) , phys .
b * 716 * , 30 ( 2012 ) .
g. aad et al .
( atlas and cms collaborations ) , jhep * 1608 * , 045 ( 2016 ) ; + l. t. hue , a. b. arbuzov , n. t. k. ngan , and h. n. long , arxiv:1611.06801 [ hep - ph ] ; + m. merchand and m. sher , arxiv:1611.06887 [ hep - ph ] ; + m. kumar , x. ruan , r. islam , a. s. cornell , m. klein , u. klein and b. mellado , arxiv:1509.04016 [ hep - ph ] ; + m. kawasaki , k. mukaida , and t. t. yanagida , phys . rev .
d * 94 * , no .
6 , 063509 ( 2016 ) ; + c. pallis , jcap * 1610 * , no .
10 , 037 ( 2016 ) ; + j. baglio , phys .
b * 764 * , 54 ( 2017 ) ; + o. evnin and r. nivesvivat , arxiv:1604.00521 [ nlin.si ] ; + a. l. maroto and f. prada , phys . rev .
d * 90 * , no . 12 , 123541 ( 2014 ) , erratum : [ phys . rev .
d * 93 * , no .
6 , 069904 ( 2016 ) ] ; + r. grber , arxiv:1611.07391 [ hep - ph ] ; + r. romero and j. besprosvany , arxiv:1611.07446 [ hep - ph ] ; + z. chen , y. yang , m. ruan , d. wang , g. li , s. jin , and y. ban , arxiv:1601.05352 [ hep - ex ] ; + a. basirnia , s. macaluso , and d. shih , arxiv:1605.08442 [ hep - ph ] ; + r. barbieri , l. j. hall , and k. harigaya , arxiv:1609.05589 [ hep - ph ] ; + a. angelescu and g. arcadi , arxiv:1611.06186 [ hep - ph ] ; + v. prilepina and y. tsai , arxiv:1611.05879 [ hep - ph ] ; + g. c. dorsch , s. j. huber , t. konstandin , and j. m. no , arxiv:1611.05874 [ hep - ph ] ; + a. haarr , a. kvellestad , and t. c. petersen , arxiv:1611.05757 [ hep - ph ] ; + m. spannowsky and c. tamarit , arxiv:1611.05466 [ hep - ph ] ; + l. m. carpenter , t. han , k. hendricks , z. qian , and n. zhou , arxiv:1611.05463 [ hep - ph ] ; + c. englert , k. nordstrm , k. sakurai , and m. spannowsky , arxiv:1611.05445 [ hep - ph ] .
v. khachatryan et al .
( cms collaboration ) , phys .
d * 92 * , no . 1 , 012004 ( 2015 ) . t. v. zagoskin and a. y. korchin , jetp * 149 * , 770 ( 2016 ) .
a. pilaftsis , c.e.m .
wagner , nucl .
b * 553 * , 3 ( 1999 ) .
v. barger , p. langacker , m. mccaskey et al .
d * 79 * , 015018 ( 2009 ) . g.c
. branco , p.m. ferreira , l. lavoura et al .
, phys . rep .
* 516 * , 1 ( 2012 ) .
a. bredenstein , a. denner , s. dittmaier and m. m. weber , phys .
d * 74 * , 013004 ( 2006 ) .
s. dittmaier et al .
( lhc higgs cross section working group collaboration ) , doi:10.5170/cern-2011 - 002 , arxiv:1101.0593 [ hep - ph ] .
m. gasperini , phys .
b * 327 * , 214 ( 1994 ) .
s. bolognesi , y. gao , a. v. gritsan et al .
d * 86 * , 095031 ( 2012 ) .
i. anderson , s. bolognesi , f. caola et al .
d * 89 * , no . 3 , 035007 ( 2014 ) .
a. soni and r. m. xu , phys .
d * 48 * , 5259 ( 1993 ) .
v. a. novikov , l. b. okun , a. n. rozanov , and m. i. vysotsky , rept .
phys . * 62 * , 1275 ( 1999 ) .
olive et al .
( particle data group ) , chin .
c * 38 * , 090001 ( 2014 ) . | in order to investigate various decay channels of the higgs boson @xmath0 or the hypothetical dilaton , we consider a neutral particle @xmath1 with zero spin and arbitrary @xmath2 parity .
this particle can decay into two off - mass - shell @xmath3 bosons ( @xmath4 and @xmath5 ) decaying to identical fermion - antifermion pairs ( @xmath6 ) : @xmath7 .
we derive analytical formulas for the fully differential width of this decay and for the fully differential width of @xmath8 ( @xmath9 stands for @xmath10 , @xmath11 , or @xmath12 ) .
integration of these formulas yields some standard model histogram distributions of the decay @xmath8 which are compared with corresponding monte carlo simulated distributions obtained by atlas and with atlas experimental data . |
within the nucleon excited states ( @xmath5 ) the @xmath0 resonance , also known as roper , plays a special role .
contrarily to the @xmath6 and other nucleon excitations , the roper was not identified as a bump in a reaction cross - section but was instead found in the analysis of phase - shifts @xcite .
nowadays there is evidence that the roper should be identified as the first radial excitation of the nucleon quark core , although meson excitations are also important for the internal structure
. calculations based on valence quark degrees of freedom are consistent with the @xmath1 transition form factors for large @xmath2 ( @xmath7 gev@xmath4 ) @xcite .
however , estimates based exclusively on quark degrees of freedom fail to describe the small @xmath2 data ( @xmath8 gev@xmath4 ) @xcite .
the gap between valence quark models and the data at low @xmath2 has been interpreted as the manifestation of the meson cloud effects @xcite . when the meson cloud contributions are included in quark models the estimates approach the data @xcite .
besides , the calculations based on dynamical coupled - channel reaction models , where the baryon excitations are described as baryon - meson states with extended baryon cores @xcite , corroborates the importance of the meson cloud effects . in those models
the mass associated with the roper bare core is about 1.7 gev . only when the meson cloud dressing is considered the roper mass is reduced to the experimental value @xcite .
the interpretation of the roper as a radial excitation of the nucleon combined with a dynamical meson cloud dressing solves the long standing problem of the roper mass in the context of a quark model @xcite .
the roper decay widths into @xmath9 , @xmath10 and @xmath11 are large comparative to other @xmath5 decays .
those decay widths are also difficult to explain in the context of a quark model .
the meson cloud dressing helps to explain the @xmath12 width , where the contributions associated with baryon - meson - meson states play an important role @xcite .
overall the recent developments in the study of the roper electromagnetic structure point to the picture of a radial excitation of the nucleon surrounded by a cloud of mesons @xcite . there is therefore a strong motivation to study the roper internal structure and to disentangle the effects of the valence quark component from the meson cloud component . in particular , one can use the knowledge of the baryon core structure to infer the contribution due to the meson cloud .
this procedure was used in previous works based on different frameworks for the baryon core @xcite . in the case of the nucleon
the valence quark degrees of freedom produce the dominant effect in the elastic form factors .
the meson cloud contribution to the nucleon wave function is estimated to be of the order of a few percent @xcite .
as for the roper , the meson cloud seems to play a more prominent role .
we conclude at the end that the valence quark degrees of freedom provide a very good description of the data , but the meson cloud contribution are important below 1.5 gev@xmath4 , particularly for the dirac form factor . in the present work we propose a new framework to analyze the role of the valence quarks and the meson cloud in the roper .
we use llight - front holography to estimate the leading order ( lowest fock state ) contribution to the @xmath1 transition form factors .
since the leading order calculation includes only pure valence quark effects in the baryon wave functions ( @xmath13 contributions ) , there is no contribution from the meson cloud . in those conditions
the gap between the calculations and the data must be essentially the consequence of the meson cloud effects . the light - front ( lf ) formalism is particularly appropriate to study hadron systems , ruled by qcd , and to describe the hadronic structure in terms of the constituents @xcite .
the lf wave functions ( lfwf ) are relativistic and frame independent @xcite .
the connection between lf quantization of qcd and anti - de - sitter conformal field theory ( ads / cft ) leads to light - front holography @xcite .
lf holography have been used to study the structure of hadron properties , such as the hadron mass spectrum , parton distribution functions , meson and baryon form factors etc .
in particular , the formalism was recently applied to the study of the nucleon @xcite and roper @xcite electromagnetic structure .
an important advantage of the lf formalism applied to hadronic physics is the systematic expansion of the wave functions into fock states with different number of constituents @xcite . in the case of the baryons
the leading order contributions is restricted to the three - valence quark configuration .
although the restriction to the lowest order fock state ( three valence - quark system ) may look as a rough simulation of the real world , it may provide an excellent first approximation when the confinement is included in the lfwf , defined at the light - front time @xcite . in those conditions
non - perturbative physics is effectively taken into account by lf holography @xcite . under the assumption that lf holography can describe accurately the large-@xmath2 region dominated by valence quark degrees of freedom , we calibrate the free parameters of the model by the available data above a given threshold @xmath14 ( @xmath152.5 gev@xmath4 ) .
this procedure differs from the previous studies where the free parameters , associated with the nucleon anomalous magnetic moments were fixed at @xmath16 , where the meson cloud contamination is expected to be stronger .
once the free parameters are fixed by the nucleon data , one uses the model to calculate the @xmath1 transition form factors . since no parameter is adjusted by the roper data , our calculations are true predictions for the @xmath1 transition form factors .
this article is organized as follows : in sec .
[ secholography ] we discuss lf holography . in the following section ( sec . [ secnucleon ] ) ,
we review the results for the nucleon form factors and fix the parameters by the nucleon data .
our results for the valence quark contribution to the @xmath1 transition form factors and their discussion are presented in sec .
[ secroper ] .
the outlook and conclusions are presented in sec .
[ secconclusions ] .
it is by now well known that string theory or gravity in anti - de - sitter ( ads ) space can provide a description of some lower - dimensional ( conformal ) gauge theories ( cft ) at strong coupling @xcite .
this ads / cft correspondence , or holography , can , with certain important restrictions , be applied to qcd - like theories , e.g. @xcite .
it remains an open question , however , whether the _ top - down _ string theory methods are practical enough to model qcd itself . instead it proved efficient to apply _ bottom - up _ approaches to qcd @xcite .
such approaches are motivated phenomenologically by the string models , but lack a first - principle justification . among those ,
the approach of lf holography is based on a comparison of the results of the lf quantization of qcd @xcite with the predictions of holographic ( ads / cft ) models @xcite .
light - front quantization provides a powerful tool to study systems ruled by microscopic qcd dynamics .
it starts with introducing a special ( light - front ) parametrization of the hamiltonian , which consequently acts on a hilbert space of the lfwf spanned by partonic fock states associated with quarks and gluons @xcite .
using lf parametrization one can separate the relative and the overall kinematics of partons obtaining wave functions @xmath17 where @xmath18 is the fraction of momentum of the parton @xmath19 , @xmath20 is the transverse momentum and @xmath21 is the number of components of the fock state .
functions @xmath17 define the probability of the given @xmath21-parton fock state .
in the massless quark limit , considering the 2d fourier transform from @xmath22 to the transverse impact variable @xmath23 , one passes to a wave function defined in the transverse impact space , @xmath24 @xcite . in the semiclassical approximation , when the quark loops are neglected and the quarks are massless
, one can replace the dependence on the variables @xmath25 and @xmath23 in @xmath26 by dependence on a single parameter @xmath27 , except for an overall factor @xmath28 @xcite .
the variable @xmath29 measures the separation between the active quark and the remaining spectator partons .
the key observation of lf holography is the equivalence of characteristic lfwfs @xmath30 and wave functions of matter fields in five - dimensional anti - de - sitter ( ads@xmath31 ) space , assuming that one identifies @xmath29 with the radial coordinate of ads@xmath31 @xcite .
the residual @xmath25 dependent factor @xmath32 is not relevant in the holographic correspondence , since its effect can be absorbed in the normalization factors in the 5d theory calculation of matrix elements .
the full lfwf is an expansion over @xmath21-parton fock states @xmath33 with coefficients @xmath30 . in the standard terminology those are the states of a given _ twist _ @xmath34 .
the purpose of this work is to understand the leading twist contribution ( @xmath35 ) , which corresponds to the valence quark approximation to certain qcd processes . to study the electromagnetic properties of the nucleon excitations we note that a three - quark system ( @xmath35 ) , as the nucleon and the nucleon radial excitations can be regarded as two - body systems with an active quark and a spectator cluster with masses ( eigenvalues ) and wave functions determined by the lf wave equation @xcite .
in the bottom - up holographic approach to qcd the nucleon and the nucleon excitations can be introduced as fermion fields in five - dimensional ads space @xcite . to define this curved space one requires the metric tensor @xmath36 , which can be conveniently introduced via line element @xmath37 in the poincar coordinates :
[ eqmetric ] ds^2 = ( _ dx^dx^- dz^2 ) , where @xmath38 is the 4d minkowski metric tensor and @xmath39 is a parameter called ads radius .
the latter sets the scale for the space s curvature .
ads@xmath31 radial coordinate @xmath40 is to be identified with the lf parameter @xmath29 .
from the metric one can define the dirac operator @xmath41 , which we write as = e^m_a^a ( _ m + _ m^ab[^a,^b ] ) . here
@xmath42 , @xmath43 is the standard set of five dirac gamma matrices and we choose to represent in the chiral basis .
frame @xmath44 and spin - connection @xmath45 tensor fields can be computed from the metric tensor . for completeness
we summarize the explicit formulas here e_a^m ^a & = & ( ^ , - i _ 5 ) , [ eqframe ] + _ m^ab & = & ( ^az_m^b-^bz_m^a ) it is also understood that @xmath46 substantiating the above claim , the kinematic properties of nucleons , _
e.g. _ the spectrum , can be described by a theory of a massive fermion in ads@xmath31 : [ eqaction ] s_n & = & d^4 x d z ( -- ) , where @xmath47 is the 5d fermion mass to be fixed below .
the function @xmath48 is an effective scalar potential , whose role is to introduce the complex dynamics of qcd , essentially the confinement . when @xmath49 , the equations of motion of the 5d theory can be cast in the form corresponding to the lf schrdinger equation in the conformal limit @xcite .
a phenomenologically compelling choice of the potential is @xcite [ potential ] = ^2 z^2 .
the dimensionful parameter @xmath50 breaks conformal symmetry and introduces a mass scale , which determines the meson and baryon spectrum @xcite .
it can be thus related to @xmath51 .
a conventional way to introduce this potential is through the dilaton field background @xcite .
this procedure makes the connection with top - down ads / cft constructions .
the corresponding bottom - up case is dubbed the soft - wall model . in four dimensions
the four - component dirac spinor describes two independent chirality modes .
although five - dimensional dirac spinor @xmath52 equally has four components , it will correspond to a nucleon with a specific single choice of chirality .
this is a consequence of boundary conditions necessary in the holographic approach @xcite . in order to preserve cp invariance ,
the second chirality mode is introduced similarly as eq .
( [ eqaction ] ) , but with an opposite sign in front of @xmath47 and @xmath53 , _
e.g. _ @xcite .
for simplicity of presentation we only review one chirality mode here . in order to solve the dirac equations derived from action ( [ eqaction ] ) the fermion field
( spin 1/2 and positive parity ) @xmath52 is decomposed into its left- and right- chirality components : ( x , z ) = _ l(x , z ) + _ r(x , z ) , where @xmath54 .
the solution can then be found via separation of variables , assuming a plane wave dependence on the 4d spacetime coordinates _
l / r(x , z ) = _ l / r(x)z^2f_l / r(z ) , where left and right weyl spinors @xmath55 satisfy the 4d dirac equation with mass @xmath56 .
we also pull out the @xmath57 factor for convenience . in the end , one is left with a pair of coupled equations for scalar _ profile _ functions @xmath58 : f_l / r ( z ) = m
f_r / l(z ) .
[ eqholog1 ] the equations must be solved with an appropriate choice of boundary conditions .
first , holography requires regularity of the wave function @xmath59 as @xmath60 . for @xmath61
there are two linearly independent solutions @xmath62 . in holography one
usually chooses to call one of those solutions the source and another one the vev of the operator dual to the bulk ( 5d ) field @xmath52 in this case baryon interpolating operator . for the problem in question it only makes sense to choose the vev solution as @xmath63 and the source @xmath64 .
the opposite choice , source @xmath65 and vev @xmath66 leads to a non - physical spectrum . note that for the opposite chirality case one changes @xmath67 , and the two choices of the @xmath61 boundary conditions are interchanged , with the physical one being @xmath65 and @xmath68 .
after fixing the boundary conditions it is standard to rewrite system ( [ eqholog1 ] ) as a second order schrdinger - like differential equation for either @xmath69 , or @xmath70 . since we are solving the boundary problem , any solution can be expanded over an eigenfunction basis , labeled by an integer @xmath21 .
equation @xmath71 has a set of eigenfunctions @xmath72 expressed in terms of the generalized laguerre polynomials @xmath73 : [ eqfr ] f_r , n(z ) z^m e^- ^2z^2/2l_n^m-1/2(^2z^2 ) .
the second component @xmath70 can now be calculated using eqs .
( [ eqholog1 ] ) : [ eqfl ] f_l , n(z ) z^m+1 e^-^2z^2/2l_n^m+1/2(^2z^2 ) .
these eigenfunctions correspond to eigenvalues given by m_n^2= 4 ^2 ( n + m + ) .
[ leigenvalues ] we recall that in the previous expressions @xmath74 .
the eigenvalues provide the result for the 4d spectrum of nucleon excitations with radial excitation quantum number @xmath21 and angular momentum quantum number linear in the parameter @xmath75 , which we will fix below . the above choice of the potential ( [ potential ] ) leads to the spectrum consistent with regge trajectories @xcite .
summing up , the full solution for the 4d positive chirality mode reads [ poschirmod ] _
n^+(x , z)= z^2 ( c f_l , n(z)_n(x ) + f_r , n(z ) _ n ( x ) + ) , where @xmath76 are two component spinors related with the weyl spinors : @xmath77 and @xmath78 .
the negative chirality modes is obtained via an appropriate change of signs and exchange of the solutions for @xmath79 @xcite : [ negchirmod ] _
n^-(x , z)= z^2 ( c f_r , n(z)_n(x ) + -f_l , n(z)_n(x ) + ) . using the component associated with the index
@xmath21 we can write the fermion fields with positive and negative parities as ^(x , z ) = _ n _ n^(x , z ) . in hadron physics ,
information on the structure of nucleon resonances is contained in the matrix elements of interaction currents between the initial and final hadrons , including the nucleon and the nucleon excitations ( baryons ) . in holography ,
4d operators of conserved currents @xmath80 are described by massless gauge fields @xmath81 in ads@xmath82 space .
interacting bulk baryon fields @xmath52 couple to the gauge fields , which is a dual holographic description of the coupling of baryons to conserved currents .
matrix elements of the currents , can be computed from the interaction terms , which are , schematically , overlaps of the bulk fields with the source gauge field @xcite : [ sint ] s_int = d^4 x d z |(x , z ) ( x , z ) ( x , z).here @xmath83 encodes a coupling of field @xmath81 to fermions .
we consider in particular the decomposition ( x , z ) = _ 0(x , z ) + _ 1(x , z ) + _ 2(x , z ) , [ eqvtotal ] as discussed next . for simplicity ,
we only present the interaction for positive chirality . for the opposite chirality case one
has to adjust the signs in the above expression to preserve the cp invariance @xcite .
the term @xmath84 is representing the minimal dirac coupling given by @xcite _ 0(x , z ) = q ^m v_m ( x , z ) , where @xmath81 could be either the abelian dual of the electromagnetic current , or the non - abelian isovector current .
following the parametrization of @xcite @xmath85 , the nucleon charge ( @xmath86 ) , for the elastic electromagnetic transitions , and @xmath87 for the transition between the nucleon and higher mass @xmath88 resonances ( @xmath89 is the pauli isospin operator ) .
minimal _ coupling only yields the dirac form factor .
to produce the pauli form factor more input from the holographic side is necessary .
abidin and carlson @xcite proposed to consider a non - minimal extension of the coupling @xmath90 , adding a term [ v1 ] _ 1 ( x , z ) = _ n [ ^m,^n]v_mn(x , z ) , [ eqv1 ] where @xmath91 $ ] and one can consider different couplings @xmath92 and @xmath93 for the isoscalar and isovector parts respectively [ @xmath94 . as one may expect from its lorentz structure , this term yields the pauli form factors , as well as a correction to the dirac form factors .
moreover , since the non - minimal term contains derivatives of the field @xmath81 and extra powers of @xmath40 , the pauli form factor turns out to be subleading with respect to the dirac form factor at large momentum transfer , as expected , while the correction to the dirac form factor is of the same order . the effect of the term @xmath95 on the nucleon form factors is studied in refs .
we provide additional details in what follows . more non - minimal couplings can be considered . in refs .
@xcite the following non - minimal ( isovector ) term was suggested [ v2 ] _ 2 ( x , z ) = g_v _ 3 ^m^5v_m ( x , z ) . where @xmath96 is a coupling constant .
this coupling is consistent with gauge invariance and discrete symmetries , that can be added to the 5d action to improve the fitting of experimental data @xcite .
one first observation about this term is that it naively breaks cp in 4d .
however , this is not the case , since 5d fermion @xmath52 describes only one 4d chirality . adding a similar interaction term for the opposite chirality , with the opposite sign of coupling , ensures 4d cp invariance .
a more serious issue is that eq .
( [ v2 ] ) breaks 5d covariance .
adding such term requires an implicit background 5d vector field .
such background fields are more common in the higher - dimensional top - down holographic models with flux compactifications . here
we assume that such a flux compactification exists and we assess its effect on the observable form factors .
we further comment on the general structure of the term ( [ sint ] ) in ref .
@xcite .
in order to calculate transition matrix elements of the currents using eq .
( [ sint ] ) one needs to specify the 5d gauge field @xmath97 .
similarly to fermions , a bosonic gauge field dual to the isovector or isoscalar current is introduced by a 5d lagrangian .
specifically , one needs to solve equations following from the action s_j = -d^4 x
d z trv_mnv^mn , subject to appropriate boundary conditions .
these boundary conditions differ from the boundary conditions for the fields associated with the nucleon and the nucleon excitations characterized by the quantum numbers @xmath21 and @xmath75 , since they have to be of the _ source _ type at @xmath98 ( the leading solution as opposed to the subleading vev - type ) .
the regularity condition requires @xmath81 to vanish for @xmath99 . in
what follows we will be interested in the abelian electromagnetic current , so the non - abelian details of this discussion can be omitted .
altogether , the solution can be cast in the form @xcite [ fourier ] v_(x , z ) = e^-i q x _ ( q)v(-q^2,z ) , where @xmath100 is the photon polarization vector , and [ vpropagator ] v(q^2,z)= ^2 z^2 _ 0 ^ 1 x^ e^- , with @xmath101 . in the limit @xmath102 , one has @xmath103 . substituting eqs .
( [ fourier ] ) and ( [ vpropagator ] ) together with the modes ( [ poschirmod ] ) and ( [ negchirmod ] ) into the interaction lagrangian ( [ sint ] ) yields matrix elements of the current up to some normalization terms , from which form factors are read off directly .
we present the results in sec .
[ secnucleon ] for the nucleon and in sec .
[ secroper ] for the roper . in this section
we specify the details of the holographic model and relating it to the observable spectrum of the nucleon radial excitations and the @xmath104 mesons . first , the nucleon ( @xmath105 ) and the roper ( @xmath106 ) are the states with radial quantum numbers @xmath107 and angular momentum @xmath108 .
any of these states must be regarded as a superposition of the given twist ( number of partons ) fock states , so we assume that the 5d mass @xmath47 is encoding both the angular momentum @xmath109 and the twist @xmath110 , @xmath111 .
a way to fix the relation for @xmath108 is to analyze the large @xmath2 scaling of @xmath72 and consequently the form factors . as discussed later
, the choice @xmath112 yields the correct falloff estimated by perturbative qcd ( pqcd ) @xcite .
consequently , the spectrum ( [ leigenvalues ] ) takes the following form @xcite , m_nn= 2 .
[ eqmassn ] thus , holography provides an estimate for the spectrum of the nucleon radial excitations in terms of a single scale parameter @xmath50 .
similarly , one obtains the spectrum of mesons and baryons with different angular momenta and parity @xcite .
the @xmath104 meson ( mass @xmath113 ) is a traditional reference for the hadron states . in lf
holography we can write @xmath114 .
the nucleon mass , for example , approximately satisfies @xmath115 .
if we assume that the nucleon mass is primarily composed of the leading twist @xmath35 contribution , @xmath116 , then the phenomenological value of @xmath50 is fixed at [ kappa ] = ~_qcd .
the leading twist estimate of the masses of nucleon radial radial excitations is now given by @xmath117 , @xmath118 .
this yields @xmath119 for the roper mass in leading twist approximation ( @xmath35 ) .
it is worth mentioning that the lf estimate of the roper mass , @xmath120 , is not so accurate as for the nucleon and the @xmath104 meson .
this is consistent with a general expectation that leading twist approximation is not so accurate for baryons as it is for mesons . in particular for the roper , there are indications that the baryon - meson - meson corrections of the twist order @xmath121 are important @xcite . in the next subsection
we collect further comments on the spectrum and other issues of lf holography .
the spectrum of the hadrons , as predicted by lf holographic approach has further issues . a similar approach to mesons yields similar regge - behaved spectra as the one given by eq .
( [ eqmassn ] ) . for the @xmath104 meson family one
gets @xmath122 for @xmath123 @xcite .
brodsky and teramond argue , however , that those twist-2 mass poles should be shifted to their physical values , suggesting that @xmath124 @xcite .
these formula gives a very good agreement with the vector meson masses for @xmath125 gev @xcite , as in eq .
( [ kappa ] ) , provided that @xmath126 mev .
similar values were used in refs .
@xcite .
the scaling dimension @xmath127 is defined by the behavior of the wave function in the limit @xmath102 : @xmath128 @xcite .
the holographic approach relates the 5d mass parameter @xmath75 with the scaling dimension @xmath127 of the hadron interpolating operator .
for example , for the vector field this relation is = 2 + . for the massless vector field @xmath81 above , this infers @xmath129 , the correct scaling dimension of a current operator @xcite . for fermions
however the relation is = 2 + |m| . in the leading twist
this yields @xmath130 for the nucleon / roper interpolating operator , in contrast to the scaling dimension @xmath131 of the operator @xmath132 , composed of three quark fields .
this discrepancy is even more pronounced if one applies the identification @xmath133 used in ref .
indeed , comparing eqs .
( [ eqfr ] ) and ( [ leigenvalues ] ) with eqs .
( 5.34 ) and ( 5.38 ) of ref .
@xcite one ends up with the relation @xmath134 , which implies @xmath135 . in qcd
the ( anomalous ) scaling dimensions of the interpolating operators are not well - defined , because of the logarithmic running of the gauge coupling . in a putative conformal theory , which would approximate the slow running of the coupling
, the anomalous scaling dimensions would exist and correspond to a dual gravity theory of the type studied here .
in particular , the massless vector field of a gravity dual would correspond to @xmath129 conserved symmetry current . in the meantime we have to accept that the correct choice of the operator dimension leads to unpleasant effects like incorrect scaling of the form factors for very large @xmath2 .
one can argue , in principle , that predictions of holographic models are generally valid in the strongly coupled regime of a theory , thus the pqcd prediction should not be compared with the outcome of the holographic analysis .
we believe , however , that this argument does not quite apply to the scaling of the form factors , since the pqcd prediction is rather based on the assumption of a particular partonic structure of the hadron , which is not exactly the full perturbative quark picture @xcite .
we hope that some more advanced holographic model can resolve the issues mentioned above . in particular , top - down models , as more complex ones , naturally share more intricate details with qcd . a prototypical example would be ( the glueball sector of ) the klebanov - strassler theory @xcite .
albeit supersymmetric , this theory encodes a logarithmically running gauge coupling , thus , extracting dimensions of the operators in this theory requires removing the logarithmic scale dependence ( _ e.g. _ @xcite ) .
moreover , in this theory , the states with the same quantum numbers tend to mix with each other
. therefore the mass eigenstates are superpositions of states with different twist .
mixing has effect of changing the resulting spectrum as well as shifting the values of the dimensions @xcite .
the transition current @xmath136 between two nucleon states ( elastic transition ) can be expressed , omitting the asymptotic spin states ( spinors ) and the electric charge @xmath137 , as @xcite j^= f_1n ( q^2 ) ^+ f_2n ( q^2 ) , [ eqnucleoncur ] where @xmath138 ( @xmath139 ) define , respectively , the dirac and pauli form factors of the proton ( @xmath140 ) and neutron ( @xmath141 ) . in the lf formalism , the two form factors appear in a spin - nonflip ( @xmath142 ) and a spin - flip ( @xmath143 ) transitions @xcite . the calculation of the overlaps ( [ sint ] ) in the holographic model with the interaction ( [ eqvtotal ] ) , where one replaces @xmath52 and @xmath144 by the appropriate initial and final nucleon state modes , in this case the nucleon modes @xmath105 . the leading twist case ( @xmath35 ) , the dirac form factor is determined by the functions @xmath145 and @xmath146 , and the pauli form factor
is determined by overlap of the @xmath147 and @xmath148 components @xcite .
the final expressions for the form factors are then @xcite : f_1n & = & e_n + + & & g_v _
n + + & & _ n , [ eqf1n ] + f_2n & = & _ n , [ eqf2n ] where @xmath149 and @xmath50 is the holographic scale discussed in the previous section . in eqs .
( [ eqf1n])-([eqf2n ] ) , @xmath150 ( @xmath151 , @xmath152 ) and @xmath153 take different values for proton and neutron . in eq .
( [ eqf2n ] ) we can replace @xmath154 by unity , if @xmath155 is a good approximation for the nucleon mass . in an exact @xmath156-flavor model one has @xmath157 .
these parameters can be determined by the proton and neutron bare anomalous magnetic moment @xmath158 and @xmath159 ( we use the upper index @xmath160 to indicate the bare values ) . in a model with no meson cloud
one can write @xmath161 and @xmath162 , assuming @xmath163 .
the holographic results ( [ eqf1n])-([eqf2n ] ) yield the correct perturbative qcd behavior for the nucleon electromagnetic form factors : @xmath164 and @xmath165 , as consequence of @xmath166 . for a more detailed discussion see refs .
@xcite . as previously discussed the non - minimal couplings @xmath153 and @xmath96
are included phenomenologically . as can be seen from eqs .
( [ eqf1n])-([eqf2n ] ) , the minimal coupling only produces the dirac form factor .
the non - minimal couplings generate a non - zero @xmath143 and adds an extra contribution to the dirac form factor ( proportional to @xmath153 ) @xcite .
relations of the type ( [ eqf1n])-([eqf2n ] ) were derived for the first time in ref .
@xcite for the case @xmath167 . to determine the values of @xmath96 and @xmath153 we performed a fit to the data @xmath168 where @xmath169 is the threshold of the data included in the fit .
we considered the cases @xmath170 , 2.0 and 2.5 gev@xmath4 , since the meson cloud effects are in principle suppressed for large @xmath2 , but we do not know a priory the threshold of the suppression .
we use the data from refs .
check refs .
@xcite for a detailed description of the database . from the fits
, we conclude that the results for the neutron electric form factor @xmath171 are very sensitive to the value of @xmath169 .
if the low @xmath2 is included in the fit ( @xmath172 1.0 gev@xmath4 ) the best fit favors a negative @xmath171 for low @xmath2 , in conflict with the measured data .
this result is very pertinent , because it is known that meson cloud effects , in particular the pion cloud , is very important for the description of the @xmath171 at low @xmath2 . since the pion cloud contributions are not included in the parametrization of eqs .
( [ eqf1n])-([eqf2n ] ) , it is not surprising to see that the fit with @xmath173 gev@xmath4 fails to describe the @xmath171 data
. it can be surprising , however , to note that the best description of the nucleon data occurs for @xmath174 gev@xmath4 in the fits with @xmath175 and 2.5 gev@xmath4 , when only a few data points for @xmath171 are considered .
one then concludes that the shape of @xmath171 is determined by the form factors @xmath176 , @xmath177 and @xmath178 .
this result shows the consistence of the fitting procedure .
the parameters obtained for the fits with @xmath170 , 2.0 and 2.5 gev@xmath4 are presented in table [ tabletub ] .
we can conclude that the parameters are sensitive to the data included in the fit .
in particular , @xmath96 depends strongly of the threshold @xmath169 .
@xmath96 , @xmath179 and @xmath180 obtained by the fit of the nucleon form factor data with @xmath14 .
[ cols="^,^,^,^",options="header " , ] the results obtained for the nucleon electric ( @xmath181 ) and magnetic ( @xmath182 form factors using the parameters from table [ tabletub ] are presented in fig . [ figgegm ] .
in the figure , we include a band to represent the interval between the model with @xmath173 gev@xmath4 and @xmath183 gev@xmath4 . in the calculations we used @xmath184 gev , in order to have a good description of the @xmath104 mesons and nucleon masses . in general the estimates from different parametrizations
are very close , except , for the neutron electric form factor .
the function @xmath171 is in fact very sensitive to the parameters @xmath96 , @xmath179 and @xmath180 . in the graph of @xmath171
the lower limit corresponds to @xmath173 gev@xmath4 and the upper limit to the case @xmath185 gev@xmath4 .
the intermediate case ( @xmath175 gev@xmath4 ) is included in order to emphasize the strong dependence of @xmath171 with the threshold @xmath169 used in the fit , and it is represented by the dashed - line . concerning the remaining form factors , one can notice an overestimation of the low @xmath2 data ( @xmath186 gev@xmath4 ) for @xmath177 and @xmath187 .
these results may be interpreted as the manifestation of the pion cloud effects , not included in the formalism associated with eqs .
( [ eqf1n])-([eqf2n ] ) . in the past
@xcite the parameters @xmath153 and @xmath96 have been fixed by the static properties of the nucleon , such as the anomalous magnetic moments and the nucleon electric charge radius .
since those observables depend on the meson cloud , we choose in this work to fix the parameters in a region where the meson cloud is significantly reduced , in order to obtain a more accurate estimate of the bare parameters .
in the recent work the brodsky - teramond model for the nucleon @xcite was improved with the inclusion of the explicit pion cloud contributions ( @xmath188 states ) with a few adjustable parameters .
it was shown that , indeed , the inclusion of the pion cloud contribution ( @xmath189 ) is fundamental for an accurate description of @xmath171 . in order to understand the role of the valence quark degrees of freedom we restrict the present study to the the leading twist contribution ( @xmath35 ) .
once fixed the parameters of the model ( bare parameters ) one can use those parameters to calculate the valence quark contributions for the @xmath1 transition form factors .
we consider now the @xmath1 transition form factors . again omitting the spinors of the nucleon and the roper ( and the electric charge @xmath137 ) ,
the transition current @xmath136 can be expressed as @xcite j^= f_1n^(q^2 ) ( ^- ) + f_2n^(q^2 ) , + [ eqropercur ] where @xmath190 and @xmath191 ( @xmath86 ) are the dirac and pauli transition form factors , respectively . using the holographic wave functions for the nucleon and the roper combined with the interaction ( [ eqvtotal ] ) ,
one concludes that the dirac and pauli form factors associated with the @xmath1 in leading twist @xcite are f_1n^&= & _
n + + & & g_v _
n + + & & _ n , [ eqf1r ] + f_2n^&= & _ n ( ) ^2 + & & , [ eqf2r ] where @xmath192 , @xmath193 , @xmath194 , @xmath195 and @xmath196 . in eq .
( [ eqf2r ] ) we can replace @xmath197 by unity if @xmath198 is a good approximation to the roper mass . in comparison with ref .
@xcite we include an extra factor @xmath199 in order to be consistent with the more usual definition of the transition form factors ( [ eqropercur ] ) .
the results for the nucleon to roper transition form factors associated with the parameters @xmath96 , @xmath179 and @xmath180 discussed previously ( table [ tabletub ] ) , are presented in fig .
[ figroper ] , for the proton target ( @xmath140 ) . as for the nucleon
we use @xmath200 gev . the data presented here are those from clas / jefferson lab for single pion production @xcite and double pion production @xcite , and it is collected in the database @xcite .
we do not discuss the results for the neutron target ( @xmath141 ) because they are restricted to the photon point . in fig .
[ figroper ] , we can see that , contrary to the nucleon case , the results are almost insensitive to the variation of parameters , therefore we consider only the upper and lower cases ( @xmath173 and 2.5 gev@xmath4 ) , delimited by the band . in the figure
one can see that the holographic approach based on eqs .
( [ eqf1r])-([eqf2r ] ) in the leading twist approximation gives a very good description of the large @xmath2 data ( @xmath201 gev@xmath4 ) in the range @xmath202, ...
,1.57 , @xmath203, ... ,0.42 and @xmath204, ... ,@xmath205 . in the above holographic approach one obtains a clear estimate of the valence quark contributions based on framework that includes only one pre - defined parameter : the mass scale @xmath50
. since the coefficients @xmath96 , @xmath179 , @xmath180 are determined by the nucleon form factors , one can consider this estimate of the valence quark contributions for the nucleon to roper form factors as a parameter - free prediction
. the present results for @xmath206 and @xmath207 are consistent with estimates of the form factors based on different frameworks , such as the results of refs .
in particular , the present model and the results of the covariant quark model from refs .
@xcite are very close for @xmath208 gev@xmath4 . the original holographic calculation of the nucleon to roper transition form factors was done in ref .
@xcite . in that work higher fock states ( twist 4 and 5 contributions )
were also taken into account .
however the effective contribution of the higher fock states was determined by the equation for the roper mass , and not by the physics associated with the form factors ( dominance of the leading twist contributions for large @xmath2 ) .
in addition , the overlap between the nucleon and roper states was determined in ref .
@xcite using coefficients adjusted to the form factor data . in the present work we rather prefer to concentrate on deriving a clean estimate of the valence quark contribution using an alternative method of fixing the parameters of the model , instead of trying to estimate the quark - antiquark and gluon contributions using lf holography
since we do not consider higher twist contributions , the overlap between the states is just the result of the overlap between the valence quark states of both baryons with no unknown coefficients .
although we conclude that the leading twist calculation is a good approximation for the form factors , one still notes that it may not be sufficient for a satisfactory estimate for the mass of the roper .
different estimates of the mass suggest , in fact , that higher fock states are crucial for the explanation of the experimental value @xcite . in fig .
[ figroper ] , we also present the estimate of the lf holographic model from teramond and brodsky for @xmath206 @xcite . in their formulation ,
the dirac form factor is expressed by an analytic function dependent on the @xmath104 meson masses @xcite .
the closeness between results is a consequence of the use of the holographic masses instead of the physical masses .
we note at this point , that , the relations ( [ eqf1r])-([eqf2r ] ) are also analytic parametrization of the roper form factors , since the form factors are expressed in terms of functions of @xmath209 , using @xmath210 .
we leave however the comparison between analytic parametrization of the roper form factors to a separated work @xcite . in the graph for @xmath206
one can notice a deviation between our estimate and the data for @xmath186 gev@xmath4 .
this result can be interpreted as the consequence of the meson cloud contributions not included in our leading twist analysis .
similar results were obtained in independent works @xcite .
sizable contributions of the meson cloud effects were found in refs .
@xcite . as for the results for @xmath207
, one can note that the parametrization based on the lf holography gives negative values for small @xmath2 , and are close to the data .
in particular we estimate the change of sign in @xmath207 for @xmath211 gev@xmath4 .
the results for @xmath207 suggest that contrarily to the form factor @xmath206 , where the meson cloud are sizable , for @xmath207 the meson cloud contribution are very small at low @xmath2 . to the best of our knowledge , it is the first time that this fact was observed in the context of a quark model .
there is , nevertheless , a discrepancy between the model and the data in the region between 0.9 and 1.5 gev@xmath4 . concerning the @xmath207 data , one can note that in the region
gev@xmath4 there are only two datapoints associated the two pion electroduction clas data @xcite .
it is important to check if the new data associated with the one pion production confirms the trend from ref .
@xcite . at the moment
one can conclude that the holographic model gives a very good description of the @xmath207 data except for three data points in the range 0.91.5 gev@xmath4 ( underestimation of the data in about 60% ) . summarizing the results presented here for the nucleon and roper form factors
, one can conclude that the leading twist approximation provides a very good estimate for both nucleon and roper form factors .
this result is consistent with the lf formalism , because in the drell - yan - west frame the contributions from higher fock states are small @xcite .
it is expected however that the meson cloud effects provide significant contributions to the @xmath212 transition form factors for some nucleon excitations @xmath5 for small @xmath2 @xcite . in the present work
, the meson cloud effect is observed in particular for the form factors of the roper : @xmath206 , and in the region 0.91.5
gev@xmath4 for @xmath207 .
those meson cloud corrections come from higher twist contributions to the light - front wave functions . for higher mass resonances
it is expected that higher twist corrections also give important contributions to the transition form factors at low @xmath2 . in that case
lf holography can be used to estimate mainly the contribution of the quark core . although the estimate for low @xmath2 may appear rough , since the calculation is based on massless quarks ( and the physical quarks have mass ) , for large @xmath2 the estimate is expected to be accurate due to the dominance of the valence quark degrees of freedom . in the future one can use holography to estimate transition form factors in the leading twist approximation for other @xmath5 states .
those estimates are expected to be accurate at large @xmath2 , since they are based on valence quark degrees of freedom . at low @xmath2
lf holography may fail in leading twist , since the framework provides only the valence quark contributions to the form factors .
that information can be , however , very useful to understand the role of the meson cloud contributions .
on one hand , we can use the comparison with the data to estimate the effect of the meson cloud contribution . on the other hand
, estimates of the bare core can also be used as input to dynamical coupled - channel reaction models in the parametrization of baryon bare core @xcite .
it is worth mentioning that some authors interpret the roper as a dynamically - generated baryon - meson resonance , without an explicit reference to three - quark systems , except for the nucleon and the @xmath6 @xcite . as far as we know
, there are no calculations of @xmath1 transition form factors for intermediate and large @xmath2 , based on dynamically - generated resonance models for the roper .
future lattice qcd simulations can help to understand the role of the baryon - meson contribution for the transition form factors at low @xmath2 @xcite .
in the present work we apply the lf holography in the soft - wall approximation to the study of the electromagnetic structure of the nucleon and nucleon excitations .
more specifically we study the nucleon and @xmath1 transition form factors in the leading twist approximation . since in the leading twist approximation
the transition form factors are determined by the valence quark degrees of freedom , in the present approach we estimate the contribution of valence quarks to the electromagnetic form factors .
the light - front wave functions are determined by an appropriate choice of boundary conditions in the 5d space and by the expected pqcd falloff for the dirac and pauli form factors ( bottom - up approach to qcd ) .
the light - front holography in the soft - wall version was used previously in the study of the nucleon elastic form factors , in leading twist and higher orders .
since in those works the couplings are adjusted to the dressed couplings ( empirical anomalous magnetic moments ) they may describe well the low @xmath2 data , but fail in the description of the large-@xmath2 data . in the present work we fix the free parameters of the model associated with the bare couplings using the large @xmath2 data for the nucleon , where the effect of the meson cloud contributions is significantly reduced .
our expressions for the electromagnetic form factors depend only of the holographic mass scale @xmath50 and of three bare couplings : @xmath96 , @xmath179 and @xmath180 .
once determined the bare couplings by the nucleon data , the model is used to predict the @xmath1 transition form factors .
our results for the @xmath1 transition form factors compare well with the empirical data . for the dirac form factor ,
the deviation observed for @xmath213 gev@xmath4 is compatible with the interpretation that meson cloud contributions are important in that region .
as for the pauli form factor , our estimate is very close to the data , both at low and at large @xmath2 , except for 3 datapoints in the range 0.91.5 gev@xmath4 .
the result at low @xmath2 , suggests that the meson cloud contributions for @xmath207 are small .
as far as we know this is the first time that this effect is observed .
the method used in the present work for the roper can in the future be extended for higher mass nucleon excitations .
the use of the light - front holography in leading twist provides then a natural method to estimate the valence quark contributions for the transition form factors .
the effect of the meson cloud can then be estimated from the comparison with the experimental data .
theoretical estimates of the bare core contributions are very important for the study of the baryon - meson reactions and nucleon electroproduction reactions .
the results from light - front holography may be used as input to dynamical coupled - channel reaction models in the theoretical study of those reactions .
g. r. thanks valery lyubovitskij for useful clarifications about the calculation of the transition form factors .
this work was supported by the universidade federal do rio grande do norte / ministrio da educao ( ufrn / mec ) .
the work of d. m. was also partially supported by the rfbr grant 16 - 01 - 00291 .
i. t. obukhovsky , a. faessler , d. k. fedorov , t. gutsche and v. e. lyubovitskij , phys .
d * 84 * , 014004 ( 2011 ) [ arxiv:1104.0957 [ hep - ph ] ] ; i. t. obukhovsky , a. faessler , t. gutsche and v. e. lyubovitskij , phys .
d * 89 * , 014032 ( 2014 ) [ arxiv:1306.3864 [ hep - ph ] ] .
j. m. maldacena , int .
j. theor .
* 38 * , 1113 ( 1999 ) [ adv .
theor . math .
* 2 * , 231 ( 1998 ) ] [ hep - th/9711200 ] .
s. s. gubser , i. r. klebanov and a. m. polyakov , phys .
b * 428 * , 105 ( 1998 ) [ hep - th/9802109 ] .
e. witten , adv .
theor . math .
* 2 * , 253 ( 1998 ) [ hep - th/9802150 ] .
e. witten , adv .
* 2 * , 505 ( 1998 ) [ hep - th/9803131 ] .
i. r. klebanov and m. j. strassler , jhep * 0008 * , 052 ( 2000 ) [ hep - th/0007191 ] . t. sakai and s. sugimoto , prog .
phys . * 113 * , 843 ( 2005 ) [ hep - th/0412141 ] .
a. dymarsky , d. melnikov and a. solovyov , jhep * 0905 * , 105 ( 2009 ) [ arxiv:0810.5666 [ hep - th ] ] . i. gordeli and d. melnikov , arxiv:1311.6537 [ hep - ph ] . m. berg , m. haack and w. mueck , nucl . phys .
b * 736 * , 82 ( 2006 ) [ hep - th/0507285 ] .
i. gordeli and d. melnikov , jhep * 1108 * , 082 ( 2011 ) [ arxiv:0912.5517 [ hep - th ] ] .
j. arrington , w. melnitchouk and j. a. tjon , phys .
c * 76 * , 035205 ( 2007 ) [ arxiv:0707.1861 [ nucl - ex ] ] .
a. j. r. puckett _ et al .
_ , phys . rev
. lett . * 104 * , 242301 ( 2010 ) [ arxiv:1005.3419 [ nucl - ex ] ] .
m. ostrick _ et al .
_ , phys .
lett . * 83 * , 276 ( 1999 ) ; c. herberg _ et al .
_ , eur .
j. a * 5 * , 131 ( 1999 ) ; d. i. glazier _ et al .
_ , eur .
phys . j. a * 24 * , 101 ( 2005 ) [ arxiv : nucl - ex/0410026 ] ; i. passchier _ et al . _ , phys .
lett . * 82 * , 4988 ( 1999 ) [ arxiv : nucl - ex/9907012 ] ; t. eden _ et al .
_ , phys .
c * 50 * , 1749 ( 1994 ) ; h. zhu _ et al .
_ [ e93026 collaboration ] , phys . rev .
lett . * 87 * , 081801 ( 2001 ) [ arxiv : nucl - ex/0105001 ] ; r. madey _ et al . _
[ e93 - 038 collaboration ] , phys .
lett . * 91 * , 122002 ( 2003 ) [ arxiv : nucl - ex/0308007 ] ; g. warren _
[ jefferson lab e93 - 026 collaboration ] , phys .
lett . * 92 * , 042301 ( 2004 ) [ arxiv : nucl - ex/0308021 ] ; s. riordan _ et al .
_ , phys .
lett . * 105 * , 262302 ( 2010 ) arxiv:1008.1738 [ nucl - ex ] ; r. schiavilla and i. sick , phys .
c * 64 * , 041002 ( 2001 ) [ arxiv : nucl - ex/0107004 ] .
e. bosted , phys .
c * 51 * , 409 ( 1995 ) ; g. kubon _ et al . _ , phys .
b * 524 * , 26 ( 2002 ) [ arxiv : nucl - ex/0107016 ] ; h. anklin _ et al .
_ , phys . lett .
b * 428 * , 248 ( 1998 ) ; h. anklin _ et al .
_ , phys .
b * 336 * , 313 ( 1994 ) ; j. lachniet _ et al .
_ [ clas collaboration ] , phys .
lett . * 102 * , 192001 ( 2009 ) [ arxiv:0811.1716 [ nucl - ex ] ] .
i. g. aznauryan _ et al . _ [ clas collaboration ] , phys .
c * 80 * , 055203 ( 2009 ) [ arxiv:0909.2349 [ nucl - ex ] ] .
v. i. mokeev _ et al . _ [ clas collaboration ] , phys .
c * 86 * , 035203 ( 2012 ) [ arxiv:1205.3948 [ nucl - ex ] ] ; v. i. mokeev _ et al .
_ , phys .
c * 93 * , 025206 ( 2016 ) [ arxiv:1509.05460 [ nucl - ex ] ] . | the structure of the nucleon and the first radial excitation of the nucleon , the roper , @xmath0 , is studied within the formalism of light - front holography .
the nucleon elastic form factors and @xmath1 transition form factors are calculated under the assumption of the dominance of the valence quark degrees of freedom .
contrary to the previous studies , the bare parameters of the model associated with the valence quark are fixed by the empirical data for large momentum transfer ( @xmath2 ) assuming that the corrections to the three - quark picture ( meson cloud contributions ) are suppressed .
the @xmath1 transition form factors are then calculated without any adjustable parameters .
our estimates are compared with results from models based on valence quarks and others .
the model compares well with the @xmath1 transition form factor data , suggesting that meson cloud effects are not large , except in the region @xmath3 gev@xmath4 .
in particular , the meson cloud contributions for the pauli form factor are small . |
after the advent of d - branes @xcite there has been a renewed interest in the study of supergravity black @xmath8-branes , in particular those preserving a fraction of the original supersymmetry .
this is due to their identification with the bps saturated non perturbative states of superstring theory @xcite which has promoted them from classical solutions of the low energy theory to solutions of the whole quantum theory .
therefore they represent an important tool in probing the non perturbative regime of superstring theories .
of particular interest are the regular ones , namely those having a non vanishing bekenstein hawking entropy .
in this paper we deal with bps static black hole solutions of @xmath9 supergravity preserving 1/8 of the original @xmath0 supersymmetry , completing a program started in @xcite and continued in @xcite .
let us recall that in the context of toroidally compactified type ii supergravity the only regular black hole solutions are the 1/8 supersymmetry preserving ones .
the 1/2 and 1/4 black holes , whose general form has been completely classified in @xcite , have a vanishing horizon area . more precisely , the only 4dimensional _ regular _ black hole configurations are those preserving 4 supersymmetry charges , irrespectively of their higher dimensional origin .
one of the results of @xcite was to show that the most general @xmath10 black hole solution of @xmath0 supergravity can be related , through a @xmath4duality transformation , to a solution of a suitable consistent truncation @xmath11 of the supergravity theory ( corresponding to a calabi - yau compactification ) .
@xmath4duality in supergravity denotes the largest global symmetry of the field equations and bianchi identities .
we shall recall the main facts about @xmath4duality orbits of bps black holes in section 2 and it will be shown that the problem of studying the most general @xmath10 bps black hole in the @xmath0 theory can be reduced to that of finding the so called _ generating solution _ , that is the most general one modulo @xmath4duality transformations , which is a solution of the simpler @xmath1 truncation .
once this solution is found , acting on it with the maximal compact subgroup @xmath12 of the @xmath4duality group one generates the general charged black hole and then acting with the whole @xmath4duality group @xmath13 one generates the most general solution , namely that with fully general asymptotic values of the scalar fields . in order to define the action of the @xmath0 @xmath4duality group on a solution of the aforementioned @xmath1 truncation , the embedding of the latter in the @xmath0 theory has to be defined in a precise group theoretical fashion .
this was done in @xcite and @xcite using the solvable lie algebra formalism @xcite .
the @xmath1 truncation mentioned above is called the @xmath2 model ( first studied in @xcite ) and essentially is a @xmath1 supergravity coupled to 3 vector multiplets interacting through the special khler manifold @xmath14 ^ 3 $ ] .
it has therefore @xmath15 real scalars ( 3 dilaton like and 3 axion like ) and four vector fields .
a generic static , spherically symmetric black hole solution of this model is then characterized by @xmath16 electric ( @xmath17 ) and @xmath16 magnetic ( @xmath18 ) quantized charges and a point @xmath19 in the scalar manifold representing the asymptotic behavior of the scalar fields at infinite radial distance from the center of the solution
. the generating solution is obtained by fixing the @xmath20 ^ 3 $ ] duality action on a generic bps black hole solution , which is achieved by first choosing a particular point @xmath19 of the scalar manifold and then fixing the action of the isotropy group @xmath21 ^ 3 $ ] on the solution , which amounts to imposing three suitable conditions on the quantized charges . these conditions in turn allow for three conditions on the evolving scalar fields of the solution , compatible with the field equations and the bps requirement .
the generating solution therefore will be described by @xmath22 charges and three independent scalars .
the mathematical framework in which to construct such a solution has been defined in @xcite , however an explicit solution was given only in a simplified case where the 3 dilatons @xmath23 and the 3 axions @xmath24 were separately set equal to a single dilaton field @xmath25 and a single axion field @xmath26 .
this solution , if properly gauge fixed , would depend only on @xmath27 parameters .
the aim of the present paper is to relax these restrictions and compute the @xmath22parameter generating solution within the @xmath2 model .
other regular solutions of the same theory , different from the generating one , have been considered in other papers @xcite .
as pointed out in @xcite and @xcite , the generating solution represents a pure ns
ns black hole , that is a black hole whose 10 dimensional microscopic configuration is made only of ns branes .
indeed , the generating solution for black holes of the toroidally compactified type ii string theory is the same as the one of the toroidally compactified heterotic string and it was already constructed in @xcite .
our point is that , as noticed in @xcite , as far as the microscopic entropy counting is concerned it is difficult to count the states of a pure ns
ns configuration .
therefore it is essential to be able to reproduce r
r configurations which are those represented in terms of intersecting d branes only .
this in principle would not be a problem since @xmath4duality does not make any difference between r
r and ns ns fields and one should be able to move from a solution to any other acting via @xmath4duality transformations .
however , as previously pointed out , in order for this to be possible one has not only to build up the generating solution itself but also to know how the latter is embedded in the full theory .
the paper is organized as follows : in section 2 we describe the structure of the @xmath4duality orbits for @xmath10 bps black holes in terms of @xmath22 independent invariants and characterize the generating solution as a solution depending on @xmath22 independent charge parameters , in terms of which the @xmath22 invariants can be expressed as independent functions on the chosen point of the moduli space at infinity . in section 3
we explicitly construct the 5 parameters generating solution by solving both the first and second order differential equations , after performing the aforementioned @xmath28 parameter gauge fixing .
the need to check that a particular solution of the first order equations representing the bps condition fulfill the second order field equations as well is due to a known feature of solitonic solutions in supergravity that the former equations in general do nt imply the latter .
most of the technical aspects concerning both the derivation of the first and second order differential equations ( the former representing the bps conditions , the latter being the equations of motion ) and the proper formalism needed in order to embed the @xmath2 model solution in the @xmath0 theory properly , have already been given in @xcite and @xcite .
hence we shall skip most technical details and make use of the results already obtained in @xcite .
we end in section 4 with few comments and some concluding remarks .
let us recall the main facts about the @xmath4duality orbits of 1/8 bps black holes in the @xmath0 classical supergravity theory @xcite,@xcite .
it is well known that the equations of motion and the bianchi identities of the @xmath0 classical supergravity theory in 4dimensions are invariant with respect to the @xmath4duality group @xmath29 .
this invariance requires the group @xmath29 to act simultaneously on both the 70 scalar fields @xmath30 spanning the manifold @xmath31 and on the vector @xmath32 consisting of the 28 electric and 28 magnetic quantized charges .
the @xmath4duality group acts on the scalar fields as the isometry group of @xmath33 and on @xmath32 in the @xmath34 ( symplectic ) representation .
a static , spherically symmetric bps black hole solution is characterized in general by the vector @xmath35 and a particular point @xmath19 on the moduli space of the theory whose @xmath36 coordinates @xmath37 are the values of the scalar fields at infinity ( @xmath38 ) .
acting on a black hole solution ( @xmath39 ) by means of a @xmath4duality transformation @xmath40 one generates a new black hole solution ( @xmath41 ) : @xmath42 the bps black hole solutions fill therefore @xmath4duality orbits .
as far as the @xmath10 bps black holes are concerned these orbits turn out to be parameterized by @xmath22 functions @xmath43 ( @xmath44 ) which are invariant under the duality transformations in ( [ uduality ] ) .
these invariants are expressed in terms of the @xmath45 anti symmetric central charge matrix @xmath46 ( the antisymmetric couple @xmath47 , as @xmath48 and @xmath49 run from 1 to 8 , labels the representation @xmath50 of @xmath51 ) in the following way ( @xcite ) : @xmath52 where @xmath53 denotes the matrix @xmath54 and the convention @xmath55 is adopted . among the @xmath56 a particular role is played by the _ moduli independent _ invariant @xmath57 that is the _ quartic invariant _
( which will be denoted in the sequel also by @xmath58 in order to refer to its group theoretical meaning ) of @xmath29 whose value is related to the entropy of the black hole @xcite,@xcite . as pointed out in @xcite , @xmath59 must be non negative in order for the solution to be bps ( @xmath60 ) .
for a fixed value of @xmath57 the inequivalent orbits are parameterized by the remaining four invariants @xmath61 , ( @xmath62 ) .
the behavior of the scalars describing the regular solutions with fixed entropy is schematically represented in figure @xmath63 in which the scalar fields flow from their boundary values @xmath19 at infinity which span @xmath64 ( the disk ) to their fixed values @xmath65 at the horizon @xmath66 @xcite .
it should be understood , of course , that the @xmath67 axis is a @xmath68dimensional space , where @xmath68 is the dimension of @xmath64 .
the invariants @xmath56 turn out to be independent functions of the quantized charges in any generic point @xmath19 of the moduli space except for some `` singular '' points where the number of truly independent invariants could be less than five .
this is the case , for example , of the point @xmath69 parameterized by the fixed values of the scalar fields at the horizon . in this point
the _ only _ independent invariant is the moduli
independent one , @xmath70 .
we will come back on this at the end of this section .
epsf the generating solution may be characterized as the solution depending on the the minimal number of parameters sufficient to obtain all possible @xmath22plets of values for the @xmath22 invariants on a particular point @xmath71 of the moduli
space ( a possible vacuum of the theory ) . from the above characterization
it follows that the whole @xmath4duality orbits of @xmath10 bps black hole solutions may be constructed by acting by means of @xmath29 transformations ( [ uduality ] ) on the generating one .
in particular , if we focus on @xmath10 bps black hole solutions having a fixed value of the entropy ( proportional to the square root of @xmath72 ) and on a particular bosonic vacuum for the theory specified by a point @xmath19 in the moduli space , by acting only on the charges of the generating solution with the @xmath4duality group , it would be possible to construct the whole spectrum of @xmath10 bps solutions of the theory realized in the chosen vacuum @xmath19 ( see figure 2 ) . since in a particular point
@xmath73 on @xmath64 the minimum number of parameters a solution should depend on in order to reproduce all the @xmath22plets of values for the independent invariants is obviously five , we expect the charge vector @xmath32 of the generating solution @xmath74 to depend on five independent charges .
epsf in order to motivate , in brief , the result obtained in @xcite according to which the generating solution for @xmath10 bps black holes in the @xmath0 theory is described within a suitable @xmath1 truncation of the theory ( the @xmath2 model ) , let us notice that the @xmath22 quantities in eq .
( [ inv5 ] ) are invariant with respect to the action of @xmath51 on @xmath75 . in particular , by means of a @xmath76parameter @xmath51 transformation , the central charge matrix can be brought to its _ normal form _ in which it is skew
symmetrized with complex eigenvalues @xmath77 , @xmath78 ( @xmath79 ) . as a consequence of this rotation ( see @xcite and @xcite for a detailed discussion ) the central charge eigenvalues end up depending on just @xmath15 ( dynamical ) scalar fields and @xmath80 quantized charges which characterize the solutions of an @xmath2 model suitably embedded in the original theory . within this truncation , the @xmath15 scalar fields ( @xmath28 dilatons @xmath23 and @xmath28 axions @xmath24 ) belong to @xmath28 vector multiplets and span a manifold @xmath81 ^ 3 $ ] , while the @xmath16 electric charges @xmath17 and @xmath16 magnetic charges @xmath18 ( @xmath82 ) transform in the @xmath83 of @xmath20 ^ 3 $ ] . in the framework of the @xmath2 model , the central charge eigenvalues @xmath84 and @xmath85 ( @xmath86 ) are , respectively , the local realization on moduli space @xmath87 of the @xmath1 supersymmetry algebra central charge @xmath88 and of the @xmath28 _ matter _ central charges associated with the @xmath28 matter vector fields ( which are related to the central charge via the following relation : @xmath89 , @xmath90 being the khler metric on @xmath87 and @xmath91 ) . on the @xmath10 bps
black hole solutions these four eigenvalues are in general independent in a generic point of the moduli space and the bps condition reads : @xmath92 since the @xmath51 transformation used to define the @xmath2 truncation of the original theory did not affect the values of the @xmath22 invariants in eq .
( [ inv5 ] ) , the latter are expected to assume all possible @xmath22plets of values on bps solutions of this theory . from this
we conclude that _ the generating solution for @xmath10 bps black holes in the @xmath0 theory is a solution of the @xmath2 truncation as well . _ in the framework of the @xmath2 model the five invariants in eq .
( [ inv5 ] ) are rewritten in the following form : @xmath93 however in this model there is still a residual invariance of the above quantities represented by the @xmath28 parameter group @xmath21 ^ 3 $ ] , isotropy group of the scalar manifold @xmath87 and subgroup of @xmath51 .
it acts on the four phases @xmath94 of the central charge eigenvalues @xmath95 leaving the overall phase @xmath96 invariant .
the generating solution is obtained by fixing this gauge freedom and therefore it depends , consistently with what stated above , on @xmath22 parameters represented by the four norms of the central charge eigenvalues @xmath97 plus the overall phase @xmath98 .
these quantities are @xmath4duality invariants as well .
it can be shown indeed that the norms @xmath97 may be expressed in terms of the four invariants @xmath99 ( @xmath100 ) while the overall phase is contained in the expression of the pfaffian in @xmath101 and thus is an invariant quantity as well which is expressed in terms of all the five @xmath102 .
indeed , see eq.s ( [ inv5 ] ) and ( [ invstu ] ) : @xmath103 the moduli independent invariant @xmath70 , computed in the @xmath2 model , is the quartic invariant of the @xmath83 of @xmath20 ^ 3 $ ] and it is useful to express it in a form which is intrinsic to this representation .
we may indeed represent the vector @xmath104 as a tensor @xmath105 where @xmath106 are the indices of the @xmath107 of each @xmath108 factor .
the invariants are constructed by contracting the indices of an even number @xmath109 of @xmath105 with @xmath110 invariant matrices @xmath111 .
this contraction gives zero for @xmath112 odd while for @xmath112 even one finds : @xmath113 it can be shown rigorously that the quartic invariant written above is the _ only _ independent invariant of this representation , that is any other invariant may be expressed as powers of it .
the square root of @xmath59 is proportional to the entropy of the solution and its expression is consistent with the result of @xcite . in terms of the @xmath80
quantized charges the components of @xmath105 are : @xmath114 according to the above characterization of the generating solution , it is apparent therefore that the black hole found in @xcite represents just a particular ( regular ) solution characterized by just @xmath27 invariant parameters ( @xmath115 and @xmath116 ) , although depending on three charges .
that solution was obtained by setting @xmath117 and @xmath118 , @xmath119 , @xmath120 .
therefore , acting on it by means of an @xmath4duality transformation , it is clear from our previous discussion that it would be possible to span only a @xmath27parameter sub orbit of the whole @xmath4duality orbit .
the same argument holds for all the regular @xmath10 bps black holes characterized by less than @xmath22 independent invariant parameters , in particular for the double fixed solution found in @xcite .
this is in fact a very particular solution in which the scalar fields do not evolve and their values at infinity coincide with their fixed ones at the horizon @xmath69 . at this point
the matter central charges vanish ( @xmath121 , @xmath86 ) and the central charge @xmath122 becomes proportional to @xmath123 which represents therefore the only invariant parameter characterizing the spectrum of @xmath10 bps solutions on this particular point of the moduli space .
the @xmath22plet of invariants @xmath124 on this solution have the specific simple form @xmath125 , @xmath26 being a constant depending on the @xmath80 charges and proportional to the entropy .
summarizing , the three main facts reviewed in the present section are : * the @xmath4duality orbits of @xmath10 bps black hole solutions are characterized by five @xmath4duality invariants @xmath102 .
* the generating solution @xmath126 is characterized by a generic point on the moduli space at infinity @xmath127 ( in which the five invariants are actually independent functions of the charges @xmath35 ) and by five independent charge parameters @xmath35 such that , by varying the latter one obtains all possible combinations of values of @xmath102 consistent with the bps condition ( @xmath128 ) . * the generating solution of 1/8 susy preserving black holes of @xmath0 supergravity can be thought of as a 1/2 preserving solution of an @xmath2 model suitably embedded in the original theory @xcite . in this framework , which is that we will deal with , the five invariants @xmath102 can be expressed as proper combinations of the norms of the four central charges @xmath129 ( the supersymmetry and the three matter ones corresponding to @xmath130 and @xmath131 respectively ) and their overall phase @xmath98 , according to eqs .
( [ invstu ] ) and ( [ pf ] ) .
in the present section we shall compute the @xmath22parameter generating solution as a @xmath132 bps black hole solution of the @xmath2 model . the property a bps saturated state has of preserving a fraction of the original supersymmetries can be characterized , on a bosonic background , by requiring that the supersymmetry transformations of all the fermion fields vanish along a suitable direction in the supersymmetry parameter space ( killing spinor ) : @xmath133 the first of these equations is equivalent to a system of first order differential equations on the background fields .
in particular the system of equations in the scalar fields has a fixed point , which will be denoted by @xmath134 , towards which the solution @xmath135 flows to at the horizon @xmath136 @xcite . the way eq .
( [ killing ] ) has been written expresses the reality condition for @xmath137 and it amounts to fix one of the three @xmath138 gauge symmetries of @xmath139 already giving therefore a condition on the @xmath80 charges and the scalar fields . in general however , as pointed out in @xcite , one should in fact consider a more general form for the killing spinor condition , namely : @xmath140 and , as noticed by g. moore , imposing the reality of the central charge could in principle imply that some topologically non trivial solutions are disregarded .
nevertheless studying such a special class of solutions is not among the purposes of our present investigation and therefore we shall choose the supersymmetry central charge to be real ( @xmath141 ) .
let us consider the gauge fixing procedure in detail .
the four central charges @xmath142 of the @xmath2 model , depending on the asymptotic values of the six scalars @xmath143 and @xmath80 charges @xmath104 , transform under @xmath20 ^ 3 $ ] duality ( [ uduality ] ) as follows : @xmath144 ^ 3&&z_k(\phi^g,\vec{q}^g)\,=\,h_g\cdot z_k(\phi,\vec{q})\nonumber\\ h_g\in so(2)^3 & & h_g\cdot z_k\equiv e^{{\rm i}\delta_k^g}z_k \label{phases}\end{aligned}\ ] ] hence an @xmath20 ^ 3 $ ] duality transformation on the moduli at infinity and on the quantized charges amounts to an @xmath21 ^ 3 $ ] phase transformation on the four central charges .
this holds true in particular if we consider @xmath145 ^ 3 $ ] .
it follows that the @xmath21 ^ 3 $ ] gauge fixing may be achieved by either imposing three suitable conditions on the phases of the central charges , or alternatively fixing the @xmath21 ^ 3 $ ] action on @xmath35 on a chosen point @xmath146 of the moduli space at infinity
. we shall pursue the latter way which amounts to impose three suitable conditions on the quantized charges .
these conditions will not be derived with group theoretical arguments ; it will suffice , as it has been pointed out in the last section , to show that the five invariants , computed in @xmath147 , are independent functions of the remaining five charges .
since , on one hand , the two fold action of a duality transformation ( and in particular of a @xmath21 ^ 3 $ ] transformation ) on both the quantized charges and the scalar fields is an invariance of the equations of motion , and on the other hand the charges @xmath35 are `` constants of motion '' ( with respect to the @xmath148evolution ) , we expect that _ the three gauge fixing conditions on the electric and magnetic charges have a counterpart in three @xmath148independent conditions on the fields @xmath149 , such that the restricted system of scalar fields and vector fields is still a solution of the field equations .
_ after the gauge fixing procedure therefore we would expect the solution to be described by five independent charges , three real scalar fields and the metric function @xmath150 .
the evolution of the latter four fields will be described in terms of four different harmonic functions .
let us now recall the main positions on the background fields used to derive the first order equations from equation ( [ killing ] ) .
the first of eqs .
( [ killing ] ) can be specialized to the supersymmetry transformations of the gravitino and gaugino within the @xmath2 model in the following way : @xmath151 where @xmath86 labels the three matter vector fields , @xmath152 are the @xmath153 r - symmetry indices and @xmath154 and @xmath155 are the graviphoton and matter field strengths respectively ( the @xmath156 sign stands for the anti self dual part ) . according to the procedure defined in @xcite , we adopt the following anstze for the vector fields : @xmath157 where @xmath158 the moduli independent quantized charges @xmath159 and the moduli dependent electric charges @xmath160 @xcite are obtained by the following integrations : @xmath161 where @xmath162 and @xmath163 denote the spheres centered in @xmath66 of radius @xmath148 and @xmath164 respectively .
the expression of the moduli dependent charges @xmath160 in terms of the scalars @xmath165 and the charges @xmath159 is given in the appendix .
as far as the metric @xmath166 , the scalars @xmath167 , parameterizing @xmath14 ^ 3 $ ] , and the killing spinors @xmath168 are concerned , the anstze we adopt are the following : @xmath169 substituting the above anstze in eqs .
( [ newkill ] ) , after some algebra , one obtains an equivalent system of first order differential equations on the background fields of the form : @xmath170 where the supersymmetry central charge
@xmath171 has the following expression : @xmath172 the vector @xmath173 is the covariantly holomorphic section on the symplectic bundle defined on the special khler manifold @xmath87 .
as already stressed , in order to find a proper solution we need also the equations of motion that must be satisfied together with the first order ones .
the former can be derived from an @xmath1 pure supergravity action coupled to 3 vector multiplets ( see @xcite for notation ) : @xmath174+h_{ij^\star}(z,\bar z)\partial_\mu z^i \partial^\mu \bar{z}^{j^\star } + \left({\rm im}{\cal n}_{\lambda\sigma}f^{\lambda}_{\cdot\cdot } f^{\sigma\vert\cdot\cdot}+{\rm re } { \cal n}_{\lambda\sigma}f^{\lambda}_{\cdot\cdot}\tilde{f}^{\sigma\vert\cdot\cdot}\right)\end{aligned}\ ] ] it was shown in @xcite that the maxwell equations are automatically satisfied by the anstze ( [ strenghtsans ] ) .
what really matters then are the scalar and einstein equations which should be fulfilled by our anstze .
their explicit form have been computed in @xcite and we report them in the appendix . the generating solution we are going to define is characterized by the following conditions on the quantized charges and on the scalar fields : + @xmath175 @xmath176 @xmath177 the relation ( [ imz0 ] ) on the axions derives from the reality condition on the supersymmetry central charge @xmath171 : @xmath178 once the conditions ( [ charcond ] ) on the charges and the two conditions on the scalars in eqs ( [ scalcond1 ] ) are taken into account . one may check that the positions ( [ charcond]),([scalcond1 ] ) and ( [ imz0 ] ) are indeed consistent with the field equations and with the system of first order equations ( [ mammamia ] ) .
in particular they are fulfilled by the fixed point values of the scalar fields @xmath179 : @xmath180 where @xmath181 was defined in eqs .
( [ invariants ] ) . a solution consistent with ( [ charcond]),([scalcond1 ] ) and ( [ imz0 ] ) will be described by four independent fields , say @xmath182 and five independent parameters , say @xmath183 .
as previously pointed out , in order for our solution to be bps it is necessary that @xmath184 which provides an inequality condition on the five parameters @xmath185 . to check that conditions ( [ charcond]),([scalcond1 ] ) and ( [ imz0 ] ) actually fix
the @xmath21 ^ 3 $ ] gauge , let us compute the five invariants @xmath186 in a suitable point of the moduli space at infinity @xmath147 which will characterize the asymptotic behavior of our generating solution . for simplicity we shall choose : @xmath187 it is useful to express the invariants in terms of the following quantities : @xmath188 where @xmath189 denotes the three cyclic permutations of the indices @xmath190 .
it is straightforward to show that the five invariants have the following expressions in terms of the above quantities : @xmath191 where we have used the property that @xmath192 , @xmath193 and @xmath194 have to fulfill the same linear relation ( [ imz0 ] ) as the axions @xmath24 .
an important feature of the system ( [ inva ] ) is to have a real solution in terms of @xmath195 for _ any _ @xmath196 of values of the invariants .
particular care , however , has to be taken when dealing with the case in which one or more of the matter central charges vanish .
the corresponding solution of the system ( [ inva ] ) is obtained by defining suitable limits of the quantities @xmath195 in which it can be shown that the generating solution is still regular .
the case for which @xmath197 , @xmath86 is easily solved by setting @xmath198 , and corresponds to the double fixed solution .
now let us compute the invariants in ( [ inva ] ) on our solution , which is characterized by the above defined point @xmath147 in the moduli space , eq .
( [ inphi ] ) , and by the five independent parameters @xmath199 left over after applying the conditions ( [ charcond ] ) on the @xmath80 quantized charges . to show that the five invariants in @xmath147 are actually independent functions of @xmath199 , we simply need to show that the quantities @xmath195 are . specializing equations ( [ xywv ] ) to the point @xmath147 in ( [ inphi ] )
one obtains : @xmath200 the above system has clearly a solution in terms of @xmath199 for any combinations of real values for @xmath201 .
therefore we have shown that by varying the five parameters @xmath199 , one may cover the whole spectrum of values for the five invariants in ( [ inva ] ) ( or equivalently @xmath102 , @xmath202 defined in section @xmath27 ) describing all possible @xmath4duality inequivalent solutions . implementing the conditions ( [ charcond]),([scalcond1 ] ) and ( [ imz0 ] ) , the system of first order equations ( [ mammamia ] ) simplifies dramatically and reduces to the following form : @xmath203 to solve the above equations it is useful to start defining the function @xmath204 whose equation is easily solved : @xmath205 in what follows we shall put @xmath206 .
the equations for the independent axions @xmath207 may then be rewritten as follows : @xmath208 the solution of the above equations is : @xmath209 as far as the equation for @xmath25 is concerned , substituting the expressions of @xmath210 in @xmath211 one obtains an equation for @xmath212 which is easily solved by introducing a new harmonic function @xmath213 .
once @xmath212 is known , from the expression of @xmath214 it is straightforward to derive the solution for @xmath150 .
the result for the independent evolving fields @xmath215 is summarized below in terms of the four different harmonic functions @xmath216 : @xmath217^{-1/4 } \cr m = -\frac{q_3}{\alpha p_3}\;\;,\;\ ; n = \alpha^2 a_2^{fix } \cr h_4(r)\ , = \,1+\frac{k_4}{r}\;\;,\;\ ; k_4\,=\,k \,b_{fix}^2 } \label{alltogether}\end{aligned}\ ] ] it can be shown that the above functional expressions for the fields @xmath218 and @xmath219 fulfill the field equations and therefore are a solution of the theory .
eq.([alltogether ] ) , together with eq.s ( [ charcond])-([imz0 ] ) , represents our 5 parameter generating solution .
let us consider now the _ near horizon _ limit of the function @xmath220 : @xmath221 substituting this limit in the anstze for the metric ( [ fishlicense ] ) we obtain the known result that the near horizon geometry of a regular bps black hole is described by a bertotti
robinson metric : @xmath222 the entropy is proportional to the area of the horizon ( @xmath223 ) according to the bekenstein hawking formula : @xmath224 where we have expressed @xmath225 as the ratio @xmath226 . as a final remark , we would like to comment on the difference between the form of the generating solution in ( [ alltogether ] ) and the one described in _ statement 3.1 _ of @xcite , which was characterized by two double fixed axions .
the functional expression of the generating solution obviously depends on the gauge fixing procedure adopted and , for computational convenience , in the present work we have fixed the @xmath21 ^ 3 $ ] gauge using a prescription which is different from that suggested in @xcite .
in this paper we have constructed the generating solution for bps saturated static black holes of @xmath0 supergravity ( that is either m theory compactified on @xmath227 , or equivalently type ii string theory compactified on @xmath228 ) .
this solution preserves 4 supercharges and can be seen either as 1/8 preserving in the context of @xmath0 theory or as 1/2 preserving in a type ii calabi
yau compactification ( or eventually 1/4 supersymmetry preserving in heterotic on @xmath228 or type ii on @xmath229 ) . actually , among compactification of string theory down to four dimensions , is the only regular one , modulo @xmath4-duality transformation , irrespectively of the original ten dimensional theory . in order for it to act as a generating solution one has to embed it in the different @xmath4duality groups ( @xmath29 for toroidal type ii compactifications and @xmath230 for heterotic ones ) .
our 5 parameter solution should be related via @xmath4duality transformations to that found in @xcite . while both these solutions carry only ns
ns charges , the proper group theoretical embedding of our @xmath2 model generating solution in the @xmath0 theory ( @xcite ) allows one to obtain , in principle , the macroscopic description of pure r
r black holes which can be interpreted microscopically in terms of d branes only @xcite .
there has been an intense study in giving the precise correspondence between macroscopic and microscopic black hole configurations in the last couple of years .
this has been investigated both in the context of @xmath0 compactifications ( see for example @xcite-@xcite ) and of @xmath1 compactifications ( see for example @xcite@xcite ) .
however , all these solutions were somewhat particular under one circumstance or another .
what we mean is that a precise and general recipe to give this correspondence for _ any _ macroscopic configuration is still lacking . on the contrary , if we know how to transform the generating solution into a generic one , in particular to those whose microscopic interpretation is known , then we can derive the microscopic stringy description of any geometric macroscopic solution . and this could shed light even on some still not understood aspects of black hole physics as , for example , the very conceptual basis of the microscopic entropy counting ( for recent work in this direction see for instance @xcite ) .
while most of the group theoretical machinery necessary in order to make the embedding has been already constructed in @xcite , it is our aim , in a forthcoming paper , to give some explicit examples of the use of @xmath4duality transformations in moving from a given solution to other ones @xcite .
for example , one could find the macroscopic description corresponding to the five parameter r
r configuration of @xcite which , to our knowledge , is the only pure r
r microscopic configuration depending on five parameters present in the literature .
* acknowledgments * m. b. and m. t. would like to thank the theoretical physics department of turin university for the kind hospitality during some stages of this work . we also thank g. bonelli , m. s. gelli and c. leone for various fruitful discussions .
setting @xmath167 eqs.([eqs122 ] ) can be rewritten in the form : @xmath231 \nonumber\\ \frac{db_1}{dr}\,&=&\,\pm \frac{e^{{\cal u}(r)}}{r^2}\sqrt{- \frac{b_1}{2b_2b_3 } } [ { a_1q_1 } + { a_2q_2 } + { a_3q_3 } + \left ( { a_1}\,{a_2}\,{a_3 } + { a_3}\,{b_1}\,{b_2 } + { a_2}\,{b_1}\,{b_3 } - { a_1}\,{b_2}\,{b_3 } \right ) \ , { p^0 } + \nonumber \\ & & + \left ( -\left ( { a_2}\,{a_3 } \right ) + { b_2}\,{b_3 } \right ) \,{p^1 } - \left ( { a_1}\,{a_3 } + { b_1}\,{b_3 } \right ) \,{p^2 } - \left ( { a_1}\,{a_2 } + { b_1}\,{b_2 } \right ) \ , { p^3 } + { q_0 } ] \nonumber\\ \frac{da_2}{dr}\,&=&\ , ( 1,2,3 ) \rightarrow ( 2,1,3 ) \nonumber\\ \frac{db_2}{dr}\,&=&\ , ( 1,2,3 ) \rightarrow ( 2,1,3 ) \nonumber\\ \frac{da_3}{dr}\,&=&\ , ( 1,2,3 ) \rightarrow ( 3,2,1 ) \nonumber\\ \frac{db_3}{dr}\,&=&\ , ( 1,2,3 ) \rightarrow ( 3,2,1 ) \nonumber\\ \frac{d\cal { u}}{dr}\,&=&\ , \pm \frac{e^{{\cal u}(r)}}{r^2}\frac{1}{2\sqrt{2 } ( - b_1b_2b_3)^{1/2}}[{a_1q_1 } + { a_2q_2 } + { a_3q_3 } + \left ( { a_1}\,{a_2}\,{a_3 } - { a_3}\,{b_1}\,{b_2 } - { a_2}\,{b_1}\,{b_3 } - { a_1}\,{b_2}\,{b_3 } \right ) \ , { p^0 } + \nonumber \\ & & - \left ( { a_2}\,{a_3 } - { b_2}\,{b_3 } \right ) \ , { p^1 } - \left ( { a_1}\,{a_3 } - { b_1}\,{b_3 } \right ) \ , { p^2 } - \left ( { a_1}\,{a_2 } - { b_1}\,{b_2 } \right ) \ , { p^3 } + { q_0 } ] \nonumber\\ 0\,&=&\,{b_1q_1 } + { b_2q_2 } + { b_3q_3 } + \left ( { a_2}\,{a_3}\,{b_1 } + { a_1}\,{a_3}\,{b_2 } + { a_1}\,{a_2}\,{b_3 } - { b_1}\,{b_2}\,{b_3 } \right ) \ , { p^0 } - \left ( { a_3}\,{b_2 } + { a_2}\,{b_3 } \right ) \ , { p^1}\nonumber\\ & & - \left ( { a_3}\,{b_1 } + { a_1}\,{b_3 } \right ) \ , { p^2 } - \left ( { a_2}\,{b_1 } + { a_1}\,{b_2 } \right ) \ , { p^3}\end{aligned}\ ] ]
the explicit form of the equations of motion for the most general case is : @xmath232 \nonumber\\ \left(b_1^{\prime\prime}+2\frac{b_1^{\prime } } { r}+ \frac{(a_1^{\prime 2}-b_1^{\prime2})}{b_1}\right)\,&=&\ , -\frac{{e^{2\,u}}}{{b_2}\,{b_3}\,{r^4}}\ , [ - ( { { { a_1}}^2}\,{{{b_2}}^2}\ , { { { b_3}}^2}\,{{{p^0}}^2 } ) + { { { b_1}}^2}\,{{{b_2}}^2}\,{{{b_3}}^2}\ , { { { p^0}}^2 } + 2\,{a_1}\,{{{b_2}}^2}\ , { { { b_3}}^2}\,{p^0}\,{p^1 } + \nonumber \\ & & - { { { b_2}}^2}\,{{{b_3}}^2}\,{{{p^1}}^2 } + { { { b_1}}^2}\,{{{b_3}}^2}\,{{{p^2}}^2 } + { { { b_1}}^2}\,{{{b_2}}^2}\,{{{p^3}}^2 } + { { { a_1}}^2}\,{{{b_2}}^2}\,{{{b_3}}^2}\ , { { { \ell ( r)_0}}^2 } + \nonumber \\ & & - { { { b_1}}^2}\,{{{b_2}}^2}\ , { { { b_3}}^2}\,{{{\ell ( r)_0}}^2 } + { { { a_3}}^2}\,{{{b_1}}^2}\,{{{b_2}}^2}\ , ( { { { p^0}}^2 } - { { { \ell ( r)_0}}^2 } ) + { { { a_2}}^2}\,{{{b_1}}^2}\,{{{b_3}}^2}\nonumber \\ & & ( { { { p^0}}^2 } - { { { \ell ( r)_0}}^2 } ) - 2\,{a_1}\,{{{b_2}}^2}\,{{{b_3}}^2}\,{\ell ( r)_0}\ , { \ell ( r)_1 } + { { { b_2}}^2}\,{{{b_3}}^2}\ , { { { \ell ( r)_1}}^2 } + \nonumber \\ & & - { { { b_1}}^2}\,{{{b_3}}^2}\ , { { { \ell ( r)_2}}^2 } + 2\,{a_2}\,{{{b_1}}^2}\ , { { { b_3}}^2}\ , ( - ( { p^0}\,{p^2 } ) + { \ell ( r)_0}\,{\ell ( r)_2 } ) + \nonumber \\ & & - { { { b_1}}^2}\,{{{b_2}}^2}\,{{{\ell ( r)_3}}^2 } + 2\,{a_3}\,{{{b_1}}^2}\,{{{b_2}}^2}\ , ( - ( { p^0}\,{p^3 } ) + { \ell ( r)_0}\,{\ell ( r)_3 } ) ] \ , \nonumber\\ \left(a_2^{\prime\prime}-2\frac{a_2^{\prime}b_2^{\prime}}{b_2}+2\frac{a_2^{\prime}}{r}\right)\,&=&\ , ( 1,2,3 ) \rightarrow ( 2,1,3 ) \nonumber\\ \left(b_2^{\prime\prime}+2\frac{b_2^{\prime } } { r}+ \frac{(a_2^{\prime 2}-b_2^{\prime2})}{b_2}\right)\,&=&\ , ( 1,2,3 ) \rightarrow ( 2,1,3 ) \nonumber\\ \left(a_3^{\prime\prime}-2\frac{a_3^{\prime}b_3^{\prime}}{b_3}+2\frac{a_3^{\prime}}{r}\right)\,&=&\ , ( 1,2,3 ) \rightarrow ( 3,2,1 ) \nonumber\\ \left(b_3^{\prime\prime}+2\frac{b_3^{\prime } } { r}+ \frac{(a_3^{\prime 2}-b_3^{\prime2})}{b_3}\right)\,&=&\ , ( 1,2,3 ) \rightarrow ( 3,2,1 ) \nonumber \\
\underline{\mbox{einstein equations : } } \quad\quad\quad\quad & & \nonumber\\ { \cal u}^{\prime\prime}+\frac{2}{r}{\cal u}^\prime\,&=&\,-2e^{-2{\cal u } } s_{00 } \nonumber\\ ( { \cal u}^\prime)^2+\sum_i\frac{1}{4b_i^2}\left((b_i^\prime)^2+(a_i^\prime)^2\right)\,&=&\ , -2e^{-2{\cal u}}s_{00 } \label{porcodue}\end{aligned}\ ] ] where the quantity @xmath233 on the right hand side of the einstein eqs .
has the following form : @xmath234 the explicit expression of the @xmath235 charges in terms of the quantized ones is : @xmath236 k. stelle , _
`` @xmath237 branes in supergravity '' _ , based on lectures given at the ictp summer school in 1996 and 1997 ( hep - th/9803116 ) ; r. dauria and p. fr , _ `` black holes in supergravity '' _ , lecture notes for the sigrav graduate school in contemporary relativity , villa olmo , como , april 1998 , hep - th/9812160 . m.j .
duff , james t. liu and j. rahmfeld , nucl . phys .
* b459 * ( 1996 ) 125 ( hep - th/9508094 ) ; r. kallosh , m. shmakova and w. k. wong , phys . rev . * d54 * ( 1996 ) 6284 ( hep - th/9607077 ) ; k. behrndt , r. kallosh , j. rahmfeld , m. shmakova and w.k .
wong , phys .
* d54 * ( 1996 ) 6293 ( hep - th/9608059 ) ; m. shmakova , phys . rev . * d56 * ( 1997 ) 540 ( hep - th/9612076 ) .
m. cvetic and d. youm , phys . rev .
* d53 * ( 1996 ) 584 ( hep - th/9507090 ) ; nucl . phys . * b472 * ( 1996 ) 249 ( hep - th/9512127 ) ; phys . rev .
* d54 * ( 1996 ) 2612 ( hep - th/9603147 ) ; m. cvetic , a. tseytlin , phys . rev . *
d53 * ( 1996 ) 5619 ; erratum - ibid .
* d55 * ( 1997 ) 3907 ( hep - th/9512031 ) . c. m. hull and p. k. townsend , nucl
b451 * ( 1995 ) 525 ( hep - th/9505073 ) ; e. witten , nucl .
* b443 * ( 1995 ) 85 ( hep - th/9503124 ) ; c. m. hull and p. k. townsend , nucl . phys . * b438 * ( 1995 ) 109 ( hep - th/9410167 ) .
s. ferrara , r. kallosh and a. strominger , phys .
* d52 * ( 1995 ) 5412 ( hep - th/9508072 ) ; a. strominger , phys .
lett . * b383*(1996 ) 39 ( hep - th/9602111 ) ; s. ferrara and r. kallosh , phys . rev . * d54 * ( 1996 ) 1525 ( hep - th/9603090 ) .
l. andrianopoli , r. dauria and s. ferrara , phys . lett .
* b403 * ( 1997 ) 12 ( hep - th/9703156 ) ; for a recent discussion on supergravity central charges in any dimension see r. dauria , s. ferrara and m. a. lled , _ `` on cenatral charges and hamiltonians for 0-branes dynamics '' _ ( hep - th/9903089 ) .
l. andrianopoli , m. bertolini , a. ceresole , r. dauria , s. ferrara , p. fr , nucl . phys .
* b476 * ( 1996 ) 397 ( hep - th/9603004 ) ; l. andrianopoli , m. bertolini , a. ceresole , r. dauria , s. ferrara , p. fr , t. magri , j.geom.phys . * 23 * ( 1997 ) 111 ( hep - th/9605032 ) .
m. bertolini , p. fr , r. iengo and c. a. scrucca , phys . lett .
* b431 * ( 1998 ) 22 ( hep - th/9803096 ) ; m. bertolini , p. fr , r. iengo and c. a. scrucca , _
d3-branes dynamics and black holes _ , talk presented by m. bertolini at the tmr , ecc contract meeting held in corf , september 1998 , hep - th/9810150 . | in this paper we construct the 5 parameter _ generating solution _ of @xmath0 bps regular supergravity black holes as a five parameter solution of the @xmath1 @xmath2 model .
our solution has a simpler form with respect to previous constructions already appeared in the literature and moreover , through the embedding @xmath3 ^ 3\subset su(3,3)\subset e_{7(7)}$ ] discussed in previous papers , the action of the @xmath4duality group is well defined .
this allows to reproduce via @xmath4duality rotations any other solution , like those corresponding to r
r black holes whose microscopic description is given by intersecting d branes .
= cmss10 = cmss10 at 7pt # 1#2#3#4 ( ) = cmss10 = cmss10 at 7pt # 1#2\{#1;#2 } = cmss10 = cmss10 at 7pt c = [ ] = = | # 1 sissa ref 55/99/ep swat/227 * the generating solution of regular n=8 bps black holes * m. bertolini@xmath5 , p. fr@xmath6 and m. trigiante@xmath7 _ @xmath5international school for advanced studies isas - sissa and infn + sezione di trieste , via beirut 2 - 4 , 34013 trieste , italy _
_ @xmath6dipartimento di fisica teorica , universit di torino and infn + sezione di torino , via p. giuria 1 , 10125 torino , italy _ _ @xmath7department of physics , university of wales swansea , singleton park + swansea sa2 8pp , united kingdom _
e - mail : [email protected] , [email protected] , [email protected] ' '' '' width 3.cm 0.2 cm supported in part by eec under tmr contracts erbfmrx ct960045 and erbfmrx |
the introduction of effective chiral lagrangians to account for the basic symmetries of qcd and its application through @xmath6 to the study of meson meson interaction @xcite or meson baryon interaction @xcite has brought new light into these problems and allowed a systematic approach . yet , @xmath6 is constrained to the low energy region , where it has had a remarkable success , but makes unaffordable the study of the intermediate energy region where resonances appear . in recent years , however , the combination of the information of the chiral lagrangians , together with the use of nonperturbative schemes , have allowed one to make prediction beyond those of the chiral perturbation expansion .
the main idea that has allowed the extension of @xmath6 to higher energies is the inclusion of unitarity in coupled channels . within the framework of chiral dynamics
, the combination of unitarity in coupled channels together with a reordering of the chiral expansion , provides a faster convergence and a larger convergence radius of a new chiral expansion , such that the lowest energy resonances are generated within those schemes .
a pioneering work along this direction was made in @xcite where the lippmann schwinger equation in coupled channels was used to deal with the meson baryon interaction in the region of the @xmath4 and @xmath7 resonances .
similar lines , using the bethe salpeter equation in the meson meson interaction , were followed in @xcite and a more elaborated framework was subsequently developed using the inverse amplitude method ( iam ) @xcite and the n / d method @xcite .
the iam method was also extended to the case of the meson baryon interaction in @xcite and in @xcite where a good reproduction of the @xmath8 resonance was obtained using second order parameters of natural size .
the n / d method has also been used for @xmath9 scattering @xcite and in the @xmath10 and coupled channels system in @xcite .
a review of these unitary methods can be seen in @xcite .
the consideration of coupled channels to study meson baryon interactions at intermediate energies has also been exploited in @xcite using the k - matrix approach , although not within a chiral context .
the bethe salpeter equation was also used in the study of the the meson baryon interaction in the strangeness @xmath11 sector in @xcite and in the @xmath0 sector around the @xmath4 region in @xcite . in this latter work , aimed at determining the @xmath12 coupling ,
only the vicinity of the resonance was studied and no particular attention was given to the region of lower energies .
subsequently a work along the same lines using the bethe salpeter equation , but considering all the freedom of the chiral constraints , was done in @xcite and a good reproduction of the experimental observables was obtained for the @xmath13 sector .
those works considered only states of meson baryon in the coupled channels and both in @xcite and @xcite the @xmath2 channel was omitted .
this channel plays a moderate role in the @xmath13 sector @xcite but , as we shall see , it plays a crucial role in the @xmath3 sector .
our aim in the present work is to extend the chiral unitary approach to account for the @xmath2 channel , including simultaneously some other corrections inspired by vector meson dominance ( vmd ) which finally allow one to have a reasonable description of the meson baryon interaction up to meson baryon energies of around 1600 mev .
the @xmath4 resonance is generated dynamically in this approach and the mass , width and branching ratios are obtained in fair agreement with the experiment .
the phase shifts and inelasticities for @xmath9 scattering in that region are also evaluated and good agreement with experiment is also found both in the @xmath13 and @xmath3 sectors . in addition some trace of the @xmath5 is found , linked to the introduction of the @xmath2 channel , with a pole in the second riemann sheet with the right energy , albeit a large width .
a side effect of the calculations is that we determine the s - wave part of the @xmath14 transition amplitude , revising previous determinations in @xcite and @xcite , and together with the p - wave amplitudes previously determined , we obtain a good reproduction of the cross sections for these reactions .
in this section we study the meson baryon scattering in s - wave in the strangeness @xmath0 sector .
we shall make use of the bethe salpeter equation in coupled channels considering states of a meson of the @xmath15 octet and a baryon of the @xmath16 octet , as required by the su(3 ) chiral formalism .
for total zero charge we have six channels , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , and @xmath22 .
the bethe salpeter equation is given by @xmath23 where @xmath24 is the product of the meson and baryon propagators .
the diagrammatic expression of the @xmath25 matrix is shown in fig.[fig : bseq ] .
following @xcite we take the kernel ( potential ) of the bethe salpeter equation from the lowest order chiral lagrangian involving mesons and baryons @xmath26 \label{eqn : lagrangian}\ ] ] where @xmath27 and @xmath28 are the su(3 ) matrices for the octet baryon field and the octet meson field respectively , and @xmath29 is the weak decay constant of the meson . from this lagrangian ,
the transition potentials between our six channels are given by @xmath30 with the initial(final ) baryon spinor @xmath31 ( @xmath32 ) and the initial(final ) meson momentum @xmath33 ( @xmath34 ) .
the coefficients @xmath35 , reflecting the su(3 ) symmetry of the problem , are obtained from eq .
( [ eqn : lagrangian ] ) and shown in table [ tbl : cij ] . .@xmath35 coefficient in the potential . @xmath36 [ cols="^,^,^,^,^,^,^",options="header " , ] the lower part of table [ tbl : coupling ] shows the same quantities obtained from a breit - wigner fit of the real energy scattering amplitudes @xmath37 .
we fit them by the breit - wigner form together with a background @xmath38 at @xmath39 mev .
the unknown parameters @xmath40 , @xmath41 and @xmath42 are determined by the method of least squares .
we obtain the values of @xmath43 from the @xmath44 corresponding to the @xmath19 final state amplitudes .
their absolute value agree fairly well with the values obtained from the pole residues and gives us confidence about our numerical evaluation of the couplings .
for instance , the @xmath9 , @xmath19 and @xmath2 branching ratios are now 24 % , 67 % and 9 % respectively .
this latter analysis allows us to determine the sign of the couplings .
the signs given are relative to that of @xmath45 .
it is instructive to decompose the resonance in the su(3 ) representations .
in fact , our result , ignoring the coupling to the @xmath2 channel , leads to @xmath46 and tells us that the @xmath4 resonance is almost an equal weight mixture of the r - parity even su(3 ) octet @xmath47 and the r - parity odd su(3 ) octet @xmath48 .
it would be interesting to compare these results with the results from other models or lattice qcd simulations .
we have studied the s - wave @xmath9 scattering , together with that of coupled channels , in a chiral unitary model in the region of center of mass energies from threshold to 1600 mev .
we calculated the t matrix using the bethe - salpeter equation in the eight coupled channels including six meson - baryon channels and two @xmath2 channels .
we took the transition potentials between the meson - baryon systems from the lowest order chiral lagrangian and improved them taking into account the vector meson dominance hypothesis .
then we introduced the appropriate @xmath49 transition potentials which influence both the elastic scattering and the pion production processes . in the present model the renormalization due to higher order contributions
is included by means of subtraction constants in the real part of the propagators of the two or three - body systems , which are taken as free parameters and determined through comparison with the t matrix of the data analysis .
the imaginary part of the meson - baryon or @xmath2 propagators is fixed and ensures unitarity in the s matrix .
a realistic t matrix is obtained with a few free parameters for energies up to 1600 mev .
the phase - shifts and the inelasticities are well reproduced in both isospin 1/2 and 3/2 .
we find that the correction of the chiral coefficient and the @xmath2 channels are important to obtain an accurate t matrix , especially in isospin 3/2 .
our analysis allowed us to determine the s - wave amplitudes for @xmath50 and we found that the isospin 3/2 @xmath49 amplitude is different from the two previous empirical ones . the resonance @xmath4 is generated dynamically and qualifies as a quasi bound state of meson and baryon .
the corresponding pole is seen in the t matrix on the complex plane .
we calculate the total and partial decay width of the resonance .
the total width obtained , about 80 mev , is smaller than the pdg estimation , but agrees with the new data from bepc .
also the large @xmath19 branching ratio observed in the data is reproduced .
the present study has served to show the potential of the chiral unitary approach extending the predictions to higher energies than it would be possible with the use of @xmath6 . yet
, we also saw that improvements in the basic information of the lowest order chiral lagrangians to account for phenomenology of vmd are welcome . on the other hand we found mandatory the inclusion of the @xmath51 channels in order to find an accurate reproduction of the data , particularly those in the isospin 3/2 sector .
we also found that the introduction of these channels , forcing them to reproduce the inelasticities and other data , has as an indirect consequence that the @xmath5 resonance appears then as a pole in the complex plane indicating a large mixing of this resonance with @xmath51 states .
this interpretation would be consistent with the large experimental coupling of this resonance to the @xmath51 channel .
we would like to acknowledge discussions with j.c . ncher .
one of us , t. i. , would like to thank j.a .
oller and a. hosaka for useful discussions .
this work has been partly supported by the spanish ministry of education in the program `` estancias de doctores y tecnlogos extranjeros en espaa '' , by the dgicyt contract number bfm2000 - 1326 and by the eu tmr network eurodaphne , contact no .
erbfmrx - ct98 - 0169
. 99 j. gasser , h. leutwyler , nucl .
b250 ( 1985 ) 465 u.g .
meissner , rep .
56 ( 1993 ) 903 v. bernard , n. kaiser and u.g .
meissner , int .
e4 ( 1995 ) 193 a. pich , rep .
58 ( 1995 ) 563 g. ecker , prog .
35 ( 1995 ) 1 n. kaiser , p.b .
siegel and w. weise , phys .
lett . b362 ( 1995 ) 23 n. kaiser , p.b . siegel and w. weise ,
a594 ( 1995 ) 325 n. kaiser , t. waas and w. weise , nucl .
a 612 ( 1997 ) 297 j.a .
oller and e. oset , nucl .
a 620 ( 1997 ) 438 ; erratum , nucl .
phys . a 652 ( 1999 ) 407 j.a .
oller , e. oset and j.r .
pelez , phys .
d59 ( 1999 ) 074001 ; erratum phys .
d60 ( 1999 ) 099906 j.a .
oller and e. oset , phys .
d60 ( 1999 ) 074023 a. gmez nicola and j.r .
pelez , phys .
d62 ( 2000 ) 017502 a. gmez nicola , j. nieves , j.r .
pelez and e. ruiz arriola , phys .
b486 ( 2000 ) 77 u.g .
meissner and j.a .
oller , nucl .
a673 ( 2000 ) 311 j.a oller and u.g .
meissner , phys .
b500 ( 2000 ) 263 j.a .
oller , e. oset and a. ramos , prog .
45 ( 2000 ) 157 t. p. vrana , s.a .
dytman and t.s.h .
lee , phys .
328 ( 2000 ) 181 e. oset and a. ramos , nucl .
a635 ( 1998 ) 99 j.c .
ncher , a. parreo , e. oset , a. ramos , a. hosaka and m. oka , nucl .
a678 ( 2000 ) 187 j. nieves and e. ruiz arriola , hep - ph/0104307 , phys .
d in print a.m. green and s. wycech , phys .
c60 ( 1999 ) 035208 f. mandl and g. show , quantum field theory , john wyley and sons , 1984 d.m .
manley , phys .
d30 ( 1984 ) 536 h. burkhardt and j.lowe , phys .
67 ( 1991 ) 2622 center of nuclear study , http://gwdac.phys.gwu.edu/ g. ecker , j. gasser , a. pich and e. de rafael , nucl .
phys . b321 ( 1989 ) 311 j. nieves and e. ruiz arriola , phys .
b455 ( 1999 ) 30 j. nieves and e. ruiz arriola , nucl .
a679 ( 2000 ) 57 m. urban , m. buballa and j. wambach , nucl .
a641 ( 1998 ) 433 f. klingl , n. kaiser and w. weise , z. phys .
a356 ( 1996 ) 193 r.aaron , phys .
rev d16 ( 1977 ) 50 e. oset and m.j .
vicente vacas , nucl .
phys . a446 ( 1985 ) 584 v. sossi , n. fazel , r.r .
johnson and m.j .
vicente vacas , phys .
b298 ( 1993 ) 287 o. jaekel , m. dillig and c.a.z .
vasconcellos , nucl .
phys . a 541 ( 1992 )
675 v. bernard , n. kaiser and u.g .
meissner , nucl .
b457 ( 1995 ) 147 v. bernard , n. kaiser and u.g .
meissner , nucl .
a619 ( 1997 ) 261 t.s .
jensen and a.f .
miranda , phys .
c55 ( 1997 ) 1039 j. caro ramon , n. kaiser , s. wetzes and w. weise , nucl .
a672 ( 2000 ) 249 m.batinic , i.slaus and a.svarc , phys .
c52 ( 1995 ) 2188 h.c .
chiang , e. oset and l.c .
liu , phys .
c44 ( 1991 ) 738 r.s.hyano , s.hirenzaki and a.grillitzer , eur .
phys . j. a6 ( 1999 ) 99 h .- ch .
schrder , , phys .
b469 ( 1999 ) 25 particle data group , eur .
j. c 15 ( 2000 ) 1 j.z .
bai , phys .
b510 ( 2001 ) 75 | we study the s - wave interaction of mesons with baryons in the strangeness @xmath0 sector in a coupled channel unitary approach .
the basic dynamics is drawn from the lowest order meson baryon chiral lagrangians .
small modifications inspired by models with explicit vector meson exchange in the @xmath1channel are also considered .
in addition the @xmath2 channel is included and shown to have an important repercussion in the results , particularly in the @xmath3 sector .
the @xmath4 resonance is dynamically generated and appears as a pole in the second riemann sheet with its mass , width and branching ratios in fair agreement with experiment .
a @xmath5 resonance also appears as a pole at the right position although with a very large width , coming essentially from the coupling to the @xmath2 channel , in qualitative agreement with experiment . |
fig . 1 . : : flow chart of the present procedure for gamow factor and/or coulomb wave function , which includes momentum resolution by generating the gaussian random numbers . fig . 2 . : : results of @xmath64 fit for 8 mkm target of @xmath66 reaction with @xmath31 mev / c by eq .
( [ eqn : a6 ] ) .
3 . : : results of the @xmath64 fit for ref .
@xcite of @xmath66 reaction with @xmath31 mev / c by eq .
( [ eqn : a9 ] ) .
4 . : : results of the @xmath64 fits for @xmath67 reactions with @xmath31 mev / c by eq .
( [ eqn : a9 ] ) : @xmath68 8 mkm target ; @xmath69 1.4 mkm target . fig . 5 .
: : physical picture of the obtained source size based on the assumed resonance effects . | we propose a new method for the coulomb wave function correction including the momentum resolution for charged hadron pairs and apply it to the precise data on @xmath0 correlations obtained in @xmath1 reaction at 70 gev / c .
it is found that interaction regions of this reaction ( assuming gaussian source function ) are @xmath2 and @xmath3 fm for the thicknesses of the target @xmath4 and @xmath5 microns , respectively .
the physical picture of the source size obtained in this way is discussed .
= -1 cm = 0 cm = 0 cm = 23.7 cm = 16 cm * 1 .
introduction.*recently we have obtained the new formulae for the coulomb wave function correction for charged hadron pairs @xcite . in particular
we have applied them ( in @xcite ) to data on @xmath0 correlation obtained in @xmath6 reaction at 70 gev / c @xcite ( which were originally corrected by usual gamow factor only ) . however , as it was pointed out to us by one of the author of @xcite , we did not take into account their published finite momentum resolution @xcite .
in fact , our formulae can not be applied directly to experimental data in which such momentum resolution is accounted for .
therefore in the present letter we would like to extend our method for the coulomb wave correction provided in @xcite to include also the momentum resolution case and to re - analyse data of @xcite and also to analyse the new , preliminary data of @xcite obtained with two kinds of thickness of ta target : 8 microns ( 8 mkm ) and 1.4 microns ( 1.4 mkm ) .
+ in the next paragraph we first reconstruct ( for the sake of completeness ) the analysis performed in ref .
@xcite and then , in the third paragraph , we propose our new method for the coulomb wave correction including this time also the momentum resolution .
the final part contains our concluding remarks .
+ * 2 . reconstruction of the analysis of @xmath0 correlation data performed by using the gamow factor with momentum resolution.*it has been stressed in @xcite that relative momenta of @xmath0 pairs observed by them have some finite resolutions .
the averaged correlation function , defined as @xmath7 depends therefore on this momentum resolution , where @xmath8 stands for non - coulomb correlation factor . to account for it
the following random number method has been proposed in ref .
@xcite in order to obtain the corresponding averaged quantities in analysis of the correlation data .
+ first of all , the relative momentum of the measured pair , @xmath9 , is decomposed into its longitudinal and transverse components : @xmath10 and @xmath11 , respectively , by making use of the uniform random number @xmath12 ( it is worthwhile to notice at this point that the transverse components @xmath11 in data of @xcite are smaller than @xmath13 mev / c ) .
one uses the following scheme here : @xmath14 at the next step , the gaussian random numbers for @xmath10 and @xmath11 are generated in the following way : @xmath15 in the above equations @xmath16 stands for the standard gaussian random number ] ] whereas @xmath17 and @xmath18 are longitudinal and transverse setup resolutions for the corresponding components , which are equal to ( values used in @xcite ) : @xmath19 mev / c ; @xmath20 mev / c ( for target of the thickness @xmath4 mkm ) and @xmath21 mev / c ( for the 1.4 mkm target ) . + finally , using the randomized number @xmath22 one calculates the corresponding randomized gamow factor correction : @xmath23 where @xmath24 . the full flow chart for this procedure is shown in fig .
1 . calculating now the average value of @xmath25 in 100 k events one can estimate the gamow factor with this finite momentum resolution , @xmath26 where @xmath27 is a free parameter .
it is understood ( or , rather , implicitly assumed ) that essentially all unlike sign pions one deals with here originate from decays of long lived particles like @xmath28 , @xmath29 , @xmath30 , and so on .
figure 2 shows the results of analysis of new data ( for 8 mkm target ) @xcite for region @xmath31 mev / c using this method .
+ * 3 .
proposal of the new method.*we would like to propose now a new method of coulomb wave function correction with a source function @xmath32 instead of the gamow factor , in order to analyse the same data . as usual
we decompose the wave function of unlike charged bosons with momenta @xmath33 and @xmath34 into the wave function of the center - of - mass system ( c.m . ) with total momentum @xmath35 and the inner wave function with relative momentum @xmath36 .
this allows us to express coulomb wave function @xmath37 in terms of the confluent hypergeometric function @xmath38 @xcite : @xmath39 where @xmath40 and the parameter @xmath41 .
assuming factorization in the source functions , @xmath42 ( here @xmath43 ) , we obtain the expression for coulomb correction for the system of @xmath0 pairs identical ( modulo the sign ) as in @xcite : @xmath44\ : , \label{eqn : a8}\end{aligned}\ ] ] where @xmath45 for the specific choice of gaussian source distribution , @xmath46 , we have @xmath47 whereas exponential source function , @xmath48 , leads to @xmath49 using now the same method of gaussian random numbers as in the previous paragraph , we can analyse the old and the new data on @xmath0 pairs @xcite using the following formula : @xmath50 figs .
3 and 4 show results of our analysis of the old and the new data , respectively .
table [ tbl : a1 ] show our results obtained using eq .
( [ eqn : a8 ] ) applied to old and new data with @xmath51mev / c . + * 4 . concluding remarks.*we have proposed the new method for the coulomb wave function correction with momentum resolution and applied it to the analysis of the precise data provided by @xcite .
authors of ref .
@xcite have analysed their @xmath0 correlation data using gamow factor for coulomb corrections together with the random numbers method to account for final momentum resolution .
we have repeated this analysis replacing gamow factor by the coulomb wave function but following the same method for correction for the momentum resolution effect ( cf .
eq .
( [ eqn : a8 ] ) ) . as a result
we have obtained the following ranges of interaction for the gaussian source function : @xmath52 to get a correct physical picture of the source size , we should calculate the root mean squared size , which is equal to : @xmath53 the present study of @xmath0 pair correlations has shown therefore that one can estimate the interaction region even from the @xmath0 correlation data .
it can be compared with the size of the ta nucleus , which is given by : @xmath54 as one can see , @xmath55 is significantly bigger than @xmath56 .
we attribute this difference to a physical picture shown in fig . 5 , i.e. , to the fact that unlike - sign pions are mostly ( if not totally ) emerging from the long - lived resonances shown there .
( in a future one should consider also a possibility of more direct estimation of the parameter @xmath27 and its role in determining the source size parameter correlation data , a similar function @xmath57 is introduced : @xmath58\:,\ ] ] where @xmath59 and @xmath60 are the degree of coherence and an exchange function due to the bose - einstein effect .
@xmath57 is attributed the resonances effect ; @xmath61 depends on the monte carlo programs @xcite . ] ) . + for completeness we have also tried to analyse the same data using exponential source function instead of gaussian .
as is shown in table [ tbl : a2 ] this leads to errors on @xmath62 of the order of @xmath63% , i.e. , with this type of source function we can not estimate the source size ( therefore it has to be discarded ) . +
* acknowledgements:*authors are grateful to dr .
l. g. afanasev for his kind correspondences and for providing us with the new data on @xmath0 correlation prior to publication .
this work is partially supported by the grant - in - aid for scientific research from the ministry of education , science and culture of japan ( # 06640383 ) , and the japan society of promotion of science ( jsps ) .
one of authors ( i. a. ) is also partially supported by russian fund of fundamental research ( grant 96 - 02 - 16347a ) . 99 m. biyajima , t. mizoguchi , t. osada and g. wilk , _ phys .
lett . _
* b353 * ( 1995 ) 340 .
m. biyajima , t. mizoguchi , t. osada and g. wilk , _ phys .
lett . _ * b366 * ( 1996 ) 394 .
l. g. afanasev et al .
, _ sov .
j. nucl .
phys . _ * 52 * ( 1990 ) 666 ( russian ed . : _ jad .
fiz . _
* 52 * ( 1990 ) 1046 ) .
l. g. afanasev , private communications ( oct . , 1995 ) .
l. g. afanasev , private communications ( dec . , 1995 ) ; l. g. afanasev et al .
, in preparation .
m. g. bowler , _ phys .
lett . _
* b270 * ( 1991 ) 69 .
l. i. schiff , _ quantum mechanics _ , 2nd ed .
( mcgraw - hill , new york , 1955 ) , p. 117 .
e. a. de wolf , proceeding of xxivth int .
sym . on multiparticle dynamics ( vietri sul mare , sept .
1994 ) .
[ tbl : a1 ] .results of the @xmath64 fits of @xmath65 for gaussian source by eqs .
( [ eqn : a6 ] ) and ( [ eqn : a9 ] ) . [ cols="^,^,^,^,^",options="header " , ] |
in quantum field theory , supersymmetry ( susy ) generates transformations between boson and fermion , supplying us with a global framework to describe both kinds of particles using the same supersymmetric multiplet @xcite .
witten was the first people who considered the simplest supersymmetric system composed of a particle of spin @xmath0 moving in one dimension @xcite , defining thus the algebra that the charge operators ( also called supercharges ) must meet and the way in which the supersymmetric hamiltonian is expressed in terms of them .
these algebraic relations define what nowadays is called supersymmetric quantum mechanic ( susy qm ) .
one of the simplest systems realizing susy qm is the so - called supersymmetric harmonic oscillator .
it has been employed to illustrate how supersymmetry works , and it has been used also to introduce the supercoherent states , a supersymmetric generalization of the standard coherent states @xcite . in order to perform the supercoherent states analysis , it is important to study the algebraic commutation relations between the hamiltonian @xmath1 and the creation @xmath2 and annihilation operators @xmath3 of the system .
in fact , one starts by defining the susy annihilation operators ( sao ) by imposing a natural commutation relationship with @xmath1 .
however , as it was shown in @xcite , the form of these operators is not unique . in a recent work , kornbluth and
zypman @xcite have found a very general expression for the operator @xmath3 , as well as the explicit form for its eigenvectors @xmath4 with complex eigenvalues @xmath5 , the so - called supercoherent states . the last turn out to be expressed in terms of standard coherent states @xcite , defined for the one - dimensional harmonic oscillator through the heisenberg - weyl algebra .
it is possible to consider also a class of deformations preserving the structure given in @xcite . in this work
we will analyze a particular deformation of the sao , the corresponding operator being labeled as @xmath6 .
proceeding in a similar way as kornbluth and zypman , we are going to find the explicit form of its eigenvector @xmath7 with complex eigenvalue @xmath8 . due to the deformation employed
, we will see that the eigenstates of @xmath6 are expressed now in terms of the corresponding nonlinear coherent states ( nlcs ) , which have been studied previously for nonlinear deformations of lie algebras @xcite .
therefore , the states @xmath7 will be called nonlinear supercoherent states ( nlscs ) .
on the other hand , due to the equidistant nature of its hamiltonian spectrum , the supersymmetric harmonic oscillator executes an evolution loop , a closed dynamical process such that the evolution operator becomes the identity ( up to a global phase factor ) at a certain time ( the so - called loop period , @xmath9 in our case ) .
the evolution loops were introduced as the basis to produce arbitrary unitary operations , which is the subject of the dynamical manipulation problem @xcite , and they appear for systems ruled either by time - dependent or by time - independent hamiltonians @xcite .
it is important to note that , for systems executing an evolution loop , any quantum state turns out to be cyclic .
starting from the work of berry @xcite , it has been realized that to any cyclic state one can associate a geometric phase @xmath10 , which characterizes global curvature effects of the space of physical states .
in fact , the geometric phase turns out to be the holonomy of the horizontal lifting of the closed trajectory in projective hilbert space @xcite ( for a recent collection of articles about geometric phases see @xcite ) . as a consequence
, the linear and nonlinear supercoherent states of the supersymmetric harmonic oscillator become cyclic states , and it would be important to evaluate their associated geometric phases . this paper is organized as follows . in section [ sec : level2 ] the standard and the simplest nonlinear coherent states will be quickly reviewed . in section
[ sec : level3 ] the hamiltonian of the supersymmetric harmonic oscillator , its eigenstates , eigenvalues , the annihilation and creation operators recently introduced by kornbluth and zipman , and its evolution loop will be discussed . in section [ sec : level4 ] a simple deformation of the previous ladder operators and the associated nonlinear supercoherent states will be constructed , while in section [ sec : level5 ] the corresponding geometric phases will be studied .
our conclusions will be presented in section [ sec : level6 ] .
before addressing the nonlinear coherent states , let us quickly review the standard harmonic oscillator ones . a standard harmonic oscillator coherent state @xmath11 can be defined by the relation @xmath12 _ i.e. _ , it is a normalized eigenstate of the annihilation operator @xmath13 with complex eigenvalue @xmath14 .
in addition , other definitions can be used to build this type of quantum states , which for the harmonic oscillator are equivalent to each other : * they satisfy the minimum heisenberg uncertainty relation , @xmath15 , where @xmath16 and @xmath17 are the uncertainties associated with the position and momentum operators , respectively . *
they appear of applying the unitary displacement operator @xmath18 on the ground state , @xmath19 ; hereafter @xmath20 will denote the complex conjugate of @xmath21 . in the fock representation ,
the normalized coherent states acquire the form @xmath22 from this explicit form , the mean values for the position @xmath23 and momentum @xmath24 operators as well as their squares become ( for @xmath25 ) : @xmath26 ^ 2 , \\
\langle\hat{p}\rangle & = & \sqrt{2}\mathbf{im}(\alpha ) , \\
\langle\hat{p}^2\rangle & = & \frac{1}{2}+2[\mathbf{im}(\alpha)]^2 .
\end{aligned}\ ] ] hence , the heisenberg uncertainty relation turns out to be given by @xmath27 .
the nonlinear coherent states ( nlcs ) have attracted a lot of interest because they exhibit nonclassical behavior . besides , many states in quantum optics can be viewed as a type of nlcs .
some authors @xcite define a nonlinear coherent state @xmath28 as an eigenstate of the deformed annihilation operator @xmath29 , so that @xmath30 where @xmath31 is a well behaved real function of the number operator @xmath32 and @xmath5 is the complex eigenvalue .
note that these deformations give place to a kind of anharmonic oscillators , called nonlinear @xmath33-oscillators , whose frequency depends on the energy @xcite .
moreover , a classification of potentials with energy spectra equivalent to @xmath33-oscillators with different kinds of nonlinearity has been also performed @xcite .
let us remark that the nonlinear coherent states @xmath34 depend strongly on the action of the operator @xmath35 on the fock states @xmath36 , which in turn depend on the explicit form of the function @xmath31 , namely , @xmath37 in order to make this fact clear , let us suppose first that @xmath38 , @xmath39 @xmath40 an explicit calculation , by defining @xmath41 , leads to @xmath42!\}^2}\right]^{-1/2}\sum_{n=0}^{\infty}\frac{z^n}{\sqrt{n!}[g(n)]!}|n\rangle,\ ] ] where @xmath43 ! = \begin{cases } 1 & \mbox{for } n \equiv 0 \\ g(n)\cdots
g(1 ) & \mbox{for } n > 1 \end{cases}\ ] ] on the other hand , if @xmath44 and @xmath38 for @xmath45 , then we get : @xmath46!\}^2}\right]^{-1/2}\sum_{n=0}^{\infty}\frac{z^n}{\sqrt{(n+1)!}[f(n)]!}|n+1\rangle.\ ] ] next , let us analyze in more detail two particular forms of @xmath31 , leading us to two kinds of coherent states which are quite similar in form ( see eqs .
( [ 15.01 ] ) and ( [ 15.02 ] ) )
although their physical properties are really different . first of all , let us assume that @xmath47 in eq .
( [ 15 ] ) so that @xmath48 @xmath39 @xmath40 .
therefore , the corresponding eigenvalue equation takes the form @xmath49 _ i.e. _ , @xmath50 is an eigenvector of the operator @xmath51 with eigenvalue @xmath14 .
thus , the explicit expression for the nonlinear coherent states of eq .
( [ 15.01 ] ) becomes now : @xmath52^{-1/2}\sum_{n=0}^\infty\frac{\alpha^n}{\gamma(n+1)\sqrt{\gamma(n+1)}}|n\rangle,\ ] ] where @xmath53 and @xmath54 is the generalized hypergeometric function defined by @xmath55 knowing the explicit form of the nonlinear coherent states @xmath50 , the mean values of the operators @xmath23 , @xmath24 and their squares become : @xmath56 ^ 2\beta(r)+[\mathbf{im}(\alpha)]^2\sigma(r ) , \label{11b}\\ \langle\hat{p}\rangle_{\mathrm{nl } } & = & \sqrt{2}\mathbf{im}(\alpha)\frac{_0f_2(1,2;r^2)}{_0f_2(1,1;r^2 ) } , \label{11c}\\ \langle\hat{p}^2\rangle_{\mathrm{nl } } & = & \frac{1}{2}+[\mathbf{re}(\alpha)]^2\sigma(r)+[\mathbf{im}(\alpha)]^2\beta(r ) , \label{11d } \end{aligned}\ ] ] where @xmath57 then , the explicit expression for the heisenberg uncertainty relation is ( see its plot in fig .
[ fig : nl ] ) : @xmath58 ^ 2\tau(r)+[\mathbf{im}(\alpha)]^2\sigma(r)}\sqrt{\frac{1}{2}-[\mathbf{im}(\alpha)]^2\tau(r)+[\mathbf{re}(\alpha)]^2\sigma(r)},\ ] ] where @xmath59 ^ 2-\beta(r).\ ] ] note that in the limit @xmath60 , the uncertainty relation takes a minimum value equal to @xmath0 , which is consistent with the fact that @xmath61 .
now , let us assume that @xmath62 in eq .
( [ 15 ] ) , so that @xmath44 .
similarly as in the previous section , if @xmath63 label the eigenvectors of @xmath64 , the eigenvalue equation becomes @xmath65 thus , the nonlinear coherent states of eq .
( [ 15.02 ] ) are expressed as @xmath66^{-1/2}\sum_{n=0}^\infty\frac{\alpha^{n}}{\gamma(n+1)\sqrt{\gamma(n+2)}}|n+1\rangle.\ ] ] once again , the mean values of the operators @xmath23 , @xmath24 and their squares in the state @xmath63 are : @xmath67 ^ 2\,\frac{_0f_2(2,3;r^2)}{_0f_2(1,2;r^2 ) } , \label{17b } \\ \langle\hat{p}\rangle_{\mathrm{nl } } & = & \sqrt{2}\mathbf{im}(\alpha)\frac{_0f_2(2,2;r^2)}{_0f_2(1,2;r^2 ) } , \label{17c}\\ \langle\hat{p}^2\rangle_{\mathrm{nl } } & = & \frac{3}{2}+[\mathbf{im}(\alpha)]^2\,\frac{_0f_2(2,3;r^2)}{_0f_2(1,2;r^2)}. \label{17d } \end{aligned}\ ] ] then , the expression for the heisenberg uncertainty relation is now @xcite ( see also fig . [
fig : nl1 ] ) : @xmath68 ^ 2\rho(r)}\sqrt{\frac{3}{2}-[\mathbf{im}(\alpha)]^2\rho(r)},\ ] ] where @xmath69 ^ 2-\frac{_0f_2(2,3;r^2)}{_0f_2(1,2;r^2)}.\ ] ] in the limit @xmath60 , the uncertainty relation takes now a maximum value equal to @xmath70 , which is also consistent with the fact that @xmath71 .
the supercoherent states are defined as eigenstates of the annihilation operator of the supersymmetric harmonic oscillator which hamiltonian @xmath1 , expressed in terms of the standard creation and annihilation operators with @xmath72 , is @xcite : @xmath73 its corresponding energy levels and associated eigenstates are : @xmath74 where @xmath75 and @xmath76 .
note that the eigenspace associated to @xmath77 for @xmath78 is two - dimensional , _
i.e. _ , @xmath77 is doubly degenerate for such an @xmath79 . by definition ,
a supersymmetric annihilation operator @xmath80 satisfies the following commutation relation with @xmath1 @xmath81=-\omega \hat{x}.\ ] ] as discussed in @xcite , the sao associated to the hamiltonian @xmath1 of the susy harmonic oscillator is not unique .
a very general form , which includes most of the previous proposals , has been considered recently @xcite , namely , @xmath82 where @xmath83 are arbitrary complex numbers .
for some particular values of the parameters @xmath83 , the annihilation operators given in @xcite are recovered ( see also @xcite ) .
indeed , for @xmath84 and @xmath85 the corresponding annihilation operator becomes @xmath86 while for @xmath87 and @xmath88 , it takes the form @xmath89 the supercoherent states are defined in the usual way @xmath90 which are expanded in terms of the energy eigenstates of the susy harmonic oscillator in the form @xmath91 they are going to be expressed in terms of the eigenvalues @xmath92 of the matrix @xmath93 , whose entries are given by the parameters @xmath83 as @xmath94 this induces a classification of the supercoherent states @xmath4 into three different families : degenerate ( @xmath95 ) ; singular ( @xmath96 ) ; generic ( anything else ) .
the evaluation of the supercoherent states of eq .
( [ 40 ] ) for each of these three cases has been performed in @xcite .
since this calculation is similar for our deformed sao , we will make it later on , when we will introduce the deformation addressed in this paper . before doing this ,
let us discuss first the evolution loop of the system and the geometric phase associated to an arbitrary cyclic state .
the dynamics of a quantum system is determined by its evolution operator , a unitary operator @xmath97 which satisfies @xmath98 where the hermitian operator @xmath99 is the system hamiltonian , @xmath100 is the identity operator . in general ,
if at @xmath101 the system is in the state @xmath102 , then at time @xmath103 the system will evolve into @xmath104 .
an evolution loop ( el ) is a circular dynamical processes such that @xmath97 becomes the identity ( up to a phase factor ) at a certain time , @xmath105 where @xmath106 is the loop period and @xmath107 @xcite .
it is important to realize that , for a system executing such a particular evolution , any state @xmath102 turns out to be cyclic , @xmath108 on the other hand , it has been realized that to any cyclic state such that @xmath109 one can associate a geometric phase , which is given by @xmath110 in particular , if the cyclic evolution is induced by a time - independent hamiltonian @xmath1 , the geometric phase becomes @xmath111 let us note that , since the spectrum of the susy harmonic oscillator consists of equidistant energy levels ( compare eq . [ 30 ] ) , then it executes an evolution loop of period @xmath112 .
in fact , it turns out that that @xcite : @xmath113 thus , any state @xmath102 of the susy harmonic oscillator turns out to be cyclic and it has associated a geometric phase which can be calculated using eq .
( [ gptih ] ) with @xmath114 ( mod@xmath115 ) and @xmath112 .
let us discuss next the nonlinear deformations of the supersymmetric annihilation operator we are interested in and their corresponding supercoherent states .
the lack of uniqueness of the supersymmetric annihilator operator is clear from the results discussed in the previous section ( see also @xcite ) ; thus , we introduce next a deformation of this operator , obtain then their eigenstates and compare them with the supercoherent states of @xcite .
we could consider a generic operator of the form @xmath29 @xcite ; however , to simplify the calculations we will choose the simplest cases , for @xmath47 and @xmath62 ( see eqs .
( [ 22 ] ) and ( [ 22.1 ] ) ) .
let us make the substitution @xmath116 in eq .
( [ 39 ] ) so that the deformed annihilation operator @xmath6 becomes @xmath117 ^ 2 & k_4(\hat{n}+\hat{1})\hat{a } \\
\end{array } \right).\ ] ] let @xmath118 be the eigenstate of the operator @xmath6 , such that @xmath119 using this eigenvalue equation , the following recurrence relations are obtained @xmath120 from the first expression with @xmath121 we obtain that @xmath122 _ i.e. _ , @xmath123 and @xmath124 are free parameters . by defining the quantities @xmath125 ^ 3}y^{1-n}s_n$ ] and @xmath126 ^ 3}y^{1-n}t_n$ ] for @xmath127 , the system ( [ 46 ] ) can be expressed as @xmath128 where @xmath93 is the matrix defined in eq .
( [ 40.41 ] ) .
we denote the eigenvalues of the matrix @xmath93 by @xmath92 and , when they are nonzero , define the quantities @xmath129 .
similar to the nondeformed case , due to the dependence of @xmath92 on the parameters @xmath83 , the nonlinear supercoherent states @xmath118 can be grouped in three cases .
[ [ generic . ] ] * generic*. + + + + + + + + + + the solution of eq .
( [ 48 ] ) is @xmath130 where @xmath131 the supercoherent states @xmath118 thus take the form @xmath132 where @xmath133 , @xmath134 and @xmath135 with @xmath136 & -y[k_1\kappa_{-}-(k_1 ^ 2+k_2k_3 ) ] \\
\end{array } \right ) .
\end{aligned}\ ] ] besides the set @xmath137 , we can choose an alternative set of nonlinear supercoherent states given by @xmath138 whereby it is possible to pass from the parameter space @xmath139 to the new one , generated by @xmath140 . [
[ degenerate . ] ] * degenerate*. + + + + + + + + + + + + + this family is obtained if and only if @xmath141 .
explicitly , the deformed supercoherent states become now @xmath142 where @xmath143 with @xmath144 and @xmath145 [ [ singular . ] ] * singular*. + + + + + + + + + + + if the matrix @xmath93 is singular then @xmath146 . hence ,
the corresponding deformed supercoherent state acquires the form @xmath147 with @xmath148 .
consider a linear combination of the states @xmath149 of eq .
( [ 52.4 ] ) with parameters @xmath150 and @xmath151 as follows : @xmath152 where the quantities @xmath153 and @xmath154 are given by @xmath155 to find the uncertainties of the position and momentum operators in the nonlinear supercoherent state of eq .
( [ 52.5 ] ) , we use the following expressions involving two nonlinear coherent states @xmath156 and @xmath157 : @xmath158 , \\ \langle\alpha_1|\hat{p}\:|\alpha_2\rangle_{\mathrm{nl } } & = & \frac{\alpha_2-\alpha^{\ast}_1}{\sqrt{2}\:i}\frac{_0f_2(1,2;\alpha^{\ast}_1\alpha_2)}{\sqrt{_0f_2(1,1;|\alpha_1|^2)}\sqrt{_0f_2(1,1;|\alpha_2|^2 ) } } , \\ \nonumber \langle\alpha_1|\hat{p}^2|\alpha_2\rangle_{\mathrm{nl } } & = & \frac{1}{2\sqrt{_0f_2(1,1;|\alpha_1|^2)}\sqrt{_0f_2(1,1;|\alpha_2|^2 ) } } \\ & & \times \bigg[-\frac{(\alpha_2 ^ 2+\alpha^{\ast2}_1)}{2}\,_0f_2(1,3;\alpha^{\ast}_1\alpha_2)+2\alpha_1^{\ast}\alpha_2\nonumber\\ & & \times\:_0f_2(2,2;\alpha^{\ast}_1\alpha_2)+\:_0f_2(1,1;\alpha^{\ast}_1\alpha_2)\bigg ] .
\end{aligned}\ ] ] note that , when @xmath159 they reduce to those shown in eqs .
( [ 11a])-([11d ] ) .
the uncertainties we are looking for become now : @xmath160 ^ 2\beta(|\varphi_+|)+[\mathbf{im}(\varphi_{+})]^2\sigma(|\varphi_+|)+\frac{1}{2}\right]%[_0f_2(1,2;|\phi_+|^2 ) ] \right . \\
& & \nonumber + \gamma^{-}\left[[\mathbf{re}(\varphi_{-})]^2\beta(|\varphi_-|)+[\mathbf{im}(\varphi_{-})]^2\sigma(|\varphi_-|)+\frac{1}{2}\right ] \\ & & \nonumber + 2\mathbf{re}\left(\frac{\gamma^{+-}}{2\sqrt{_0f_2(1,1;|\varphi_{+}|^2)}\sqrt{_0f_2(1,1;|\varphi_{-}|^2)}}\right.\\ & & \nonumber \times\left.\left.\left[\frac{(\varphi_{-}^2+\varphi^{\ast2}_+)}{2}\;_0f_2(1,3;\varphi^{\ast}_{+}\varphi_{-})+2\varphi^{\ast}_+\varphi_-\;_0f_2(2,2;\varphi^{\ast}_{+}\varphi_{-})+\;_0f_2(1,1;\varphi^{\ast}_{+}\varphi_{-})\right]%[_0f_2(1,2;\phi^{\ast}_{+}\phi_{- } ) ] \right)\right\ } \\ & & \nonumber -\delta_1^{-2}\left\{\sqrt{2}\gamma^{+}\mathbf{re}(\varphi_{+})\frac{_0f_2(1,2;|\varphi_+|^2)}{_0f_2(1,1;|\varphi_+|^2)}+\sqrt{2}\gamma^{-}\mathbf{re}(\varphi_{-})\frac{_0f_2(1,2;|\varphi_-|^2)}{_0f_2(1,1;|\varphi_-|^2)}\right . \\ & & \left.+\sqrt{2}\mathbf{re}\left((\varphi_{-}+\varphi^{\ast}_+)\frac{\gamma^{+-}\;_0f_2(1,2;\varphi^{\ast}_{+}\varphi_{-})}{\sqrt{_0f_2(1,1;|\varphi_{+}|^2)}\sqrt{_0f_2(1,1;|\varphi_{-}|^2)}}\right)\right\}^2,\\ \nonumber \sigma^{2 ( \mathrm{nl})}_p&= & \delta_1^{-1}\left\{\gamma^{+}\left[[\mathbf{re}(\varphi_{+})]^2\sigma(|\varphi_+|)+[\mathbf{im}(\varphi_{+})]^2\beta(|\varphi_+|)+\frac{1}{2}\right]%[_0f_2(1,2;|\phi_+|^2 ) ] \right . \\ &
& \nonumber + \gamma^{-}\left[[\mathbf{re}(\varphi_{-})]^2\sigma(|\varphi_-|)+[\mathbf{im}(\varphi_{-})]^2\beta(|\varphi_-|)+\frac{1}{2}\right]\\ & & \nonumber + 2\mathbf{re}\left(\frac{\gamma^{+-}}{2\sqrt{_0f_2(1,1;|\varphi_{+}|^2)}\sqrt{_0f_2(1,1;|\varphi_{-}|^2)}}\right.\\ & & \nonumber \times\left.\left.\left[-\frac{(\varphi_{-}^2+\varphi^{\ast2}_{+})}{2}\;_0f_2(1,3;\varphi^{\ast}_{+}\varphi_{-})+2\varphi^{\ast}_+\varphi_-\;_0f_2(2,2;\varphi^{\ast}_{+}\varphi_{-})+\;_0f_2(1,1;\varphi^{\ast}_{+}\varphi_{-})\right]%[_0f_2(1,2;\phi^{\ast}_{+}\phi_{- } ) ] \right)\right\ } \\ & & \nonumber -\delta_1^{-2}\left\{\sqrt{2}\gamma^{+}\mathbf{im}(\varphi_{+})\;\frac{_0f_2(1,2;|\varphi_+|^2)}{_0f_2(1,1;|\varphi_+|^2)}+\sqrt{2}\gamma^{-}\mathbf{im}(\varphi_{-})\;\frac{_0f_2(1,2;|\varphi_-|^2)}{_0f_2(1,1;|\varphi_-|^2)}\right .
\\ & & \left.+\sqrt{2}\mathbf{re}\left(-i(\varphi_{-}-\varphi^{\ast}_+)\frac{\gamma^{+-}\;_0f_2(1,2;\varphi^{\ast}_{+}\varphi_{-})}{\sqrt{_0f_2(1,1;|\varphi_{+}|^2)}\sqrt{_0f_2(1,1;|\varphi_{-}|^2)}}\right)\right\}^2 , \end{aligned}\ ] ] where @xmath161 and @xmath162 are given by eq .
( [ 24.3 ] ) , and @xmath163 + \gamma^{-}%[_0f_2(1,2;|\phi_-|^2 ) ] + 2\mathbf{re}\left(\frac{\gamma^{+-}\;_0f_2(1,1;\varphi^{\ast}_{+}\varphi_-)}{\sqrt{_0f_2(1,1;|\varphi_{+}|^2)}\sqrt{_0f_2(1,1;|\varphi_-|^2)}}\right ) .
\end{aligned}\ ] ] let us make the change @xmath164 in eq .
( [ 39 ] ) so that the deformed annihilation operator @xmath165 becomes now : @xmath166 let @xmath7 be the eigenstate of the operator @xmath165 , such that @xmath167 using this eigenvalue equation , the following recurrence relations are obtained @xmath168 from the expressions in eq .
( [ 55 ] ) , we obtain first that @xmath169 whereby @xmath170 and @xmath171 are the free parameters .
if we define now @xmath172 and @xmath173 for @xmath174 , then the system in eq .
( [ 55 ] ) can be expressed as @xmath175 where @xmath93 is once again the matrix defined by eq .
( [ 40.41 ] ) . for the nonzero eigenvalues @xmath92 of the matrix @xmath93
, we define the quantities @xmath176 . as in the previous case ,
the same classification of the nonlinear supercoherent states , according to the values taken by @xmath92 , is obtained .
[ [ generic.-1 ] ] * generic*. + + + + + + + + + + as before , the solution for the matrix equation ( [ 57 ] ) turns out to be @xmath177 where @xmath178 the supercoherent states @xmath7 become now @xmath179 where @xmath180 , @xmath181 and @xmath182 with @xmath183 & -z[k_1\kappa_{-}-(k_1 ^ 2+k_2k_3 ) ] \\ \end{array } \right ) .
\end{aligned}\ ] ] again , from the set @xmath184 we can build a new one formed by the elements @xmath185 _ i.e. _ , we pass from the parameter space @xmath139 to the new one @xmath140 .
[ [ degenerate.-1 ] ] * degenerate*. + + + + + + + + + + + + + the deformed supercoherent states are explicitly given by @xmath186 where @xmath187 with @xmath188 and @xmath189 [ [ singular.-1 ] ] * singular*. + + + + + + + + + + + the corresponding deformed supercoherent state is now @xmath190 with @xmath191 .
consider again the following superposition of states @xmath192 of eq .
( [ 63.1 ] ) with parameters @xmath150 and @xmath151 : @xmath193 where the quantities @xmath153 and @xmath154 are defined before , in eqs .
( [ 45a ] ) and ( [ 45b ] ) . in order to find the uncertainties of the position and momentum operators in the nonlinear supercoherent state of eq .
( [ 68 ] ) , we will use now the following expressions involving @xmath194 and @xmath195 : @xmath196 , \\
\langle\alpha_1|\hat{p}\:|\alpha_2\rangle_{\mathrm{nl } } & = & \frac{\alpha_2-\alpha^{\ast}_1}{\sqrt{2}\:i}\frac{_0f_2(2,2;\alpha^{\ast}_1\alpha_2)}{\sqrt{_0f_2(1,2;|\alpha_1|^2)}\sqrt{_0f_2(1,2;|\alpha_2|^2 ) } } , \\ \nonumber \langle\alpha_1|\hat{p}^2|\alpha_2\rangle_{\mathrm{nl } } & = & \frac{1}{2\sqrt{_0f_2(1,2;|\alpha_1|^2)}\sqrt{_0f_2(1,2;|\alpha_2|^2 ) } } \\ & & \times \left[-\frac{(\alpha_2-\alpha^{\ast}_1)^2}{2}\,_0f_2(2,3;\alpha^{\ast}_1\alpha_2)+3\,_0f_2(1,2;\alpha^{\ast}_1\alpha_2)\right ] .
\end{aligned}\ ] ] when @xmath197 they reduce now to those of eqs .
( [ 17a])-([17d ] ) .
the expressions for the uncertainties are finally : @xmath198 ^ 2\frac{_0f_2(2,3;|\phi_+|^2)}{_0f_2(1,2;|\phi_+|^2)}+\frac{3}{2}\right]%[_0f_2(1,2;|\phi_+|^2 ) ] \right .
\nonumber\\ & & + \gamma^{-}\left[[\mathbf{re}(\phi_{-})]^2\frac{_0f_2(2,3;|\phi_-|^2)}{_0f_2(1,2;|\phi_-|^2)}+\frac{3}{2}\right]\nonumber \\ & & + 2\mathbf{re}\left(\frac{\gamma^{+-}}{2\sqrt{_0f_2(1,2;|\phi_{+}|^2)}\sqrt{_0f_2(1,2;|\phi_{-}|^2)}}\right.\nonumber\\ & & \times\left.\left.\left[\frac{(\phi_{-}+\phi^{\ast}_+)^2}{2}\;_0f_2(2,3;\phi^{\ast}_{+}\phi_{-})+3\,_0f_2(1,2;\phi^{\ast}_{+}\phi_{-})\right]%[_0f_2(1,2;\phi^{\ast}_{+}\phi_{- } ) ] \right)\right\ } \nonumber \\ & & -\delta_2^{-2}\left\{\sqrt{2}\,\gamma^{+}\mathbf{re}(\phi_{+})\frac{_0f_2(2,2;|\phi_+|^2)}{_0f_2(1,2;|\phi_+|^2)}+\sqrt{2}\,\gamma^{-}\mathbf{re}(\phi_{-})\frac{_0f_2(2,2;|\phi_-|^2)}{_0f_2(1,2;|\phi_-|^2)}\right . \nonumber \\ & & \left.+\sqrt{2}\,\mathbf{re}\left((\phi_{-}+\phi^{\ast}_+)\frac{\gamma^{+-}\;_0f_2(2,2;\phi^{\ast}_{+}\phi_{-})}{\sqrt{_0f_2(1,2;|\phi_{+}|^2)}\sqrt{_0f_2(1,2;|\phi_{-}|^2)}}\right)\right\}^2,\\ \sigma^{2 ( \mathrm{nl})}_p&= & \delta_2^{-1}\left\{\gamma^{+}\left[[\mathbf{im}(\phi_{+})]^2\frac{_0f_2(2,3;|\phi_+|^2)}{_0f_2(1,2;|\phi_+|^2)}+\frac{3}{2}\right]%[_0f_2(1,2;|\phi_+|^2 ) ] \right .
\nonumber \\ & & + \gamma^{-}\left[[\mathbf{im}(\phi_{-})]^2\frac{_0f_2(2,3;|\phi_-|^2)}{_0f_2(1,2;|\phi_-|^2)}+\frac{3}{2}\right]\nonumber\\ & & + 2\mathbf{re}\left(\frac{\gamma^{+-}}{2\sqrt{_0f_2(1,2;|\phi_{+}|^2)}\sqrt{_0f_2(1,2;|\phi_{-}|^2)}}\right .
\nonumber\\ & & \times\left.\left.\left[-\frac{(\phi_{-}-\phi^{\ast}_+)^2}{2}\;_0f_2(2,3;\phi^{\ast}_{+}\phi_{-})+3\,_0f_2(1,2;\phi^{\ast}_{+}\phi_{-})\right]%[_0f_2(1,2;\phi^{\ast}_{+}\phi_{- } ) ] \right)\right\ } \nonumber \\ & & -\delta_2^{-2}\left\{\sqrt{2}\,\gamma^{+}\mathbf{im}(\phi_{+})\frac{_0f_2(2,2;|\phi_+|^2)}{_0f_2(1,2;|\phi_+|^2)}+\sqrt{2}\,\gamma^{-}\mathbf{im}(\phi_{-})\frac{_0f_2(2,2;|\phi_-|^2)}{_0f_2(1,2;|\phi_-|^2)}\right .
\nonumber\\ & & \left.+\sqrt{2}\,\mathbf{re}\left(-i(\phi_{-}-\phi^{\ast}_+)\frac{\gamma^{+-}\;_0f_2(2,2;\phi^{\ast}_{+}\phi_{-})}{\sqrt{_0f_2(1,2;|\phi_{+}|^2)}\sqrt{_0f_2(1,2;|\phi_{-}|^2)}}\right)\right\}^2 \end{aligned}\ ] ] where now @xmath199 + \gamma^{-}%[_0f_2(1,2;|\phi_-|^2 ) ] + 2\mathbf{re}\left(\frac{\gamma^{+-}\;_0f_2(1,2;\phi^{\ast}_{+}\phi_-)}{\sqrt{_0f_2(1,2;|\phi_{+}|^2)}\sqrt{_0f_2(1,2;|\phi_-|^2)}}\right ) .
\end{aligned}\ ] ] let the annihilation operators @xmath200 and @xmath201 be given respectively by @xmath202 ^ 2\sin\theta & ( \hat{n}+\hat{1})\hat{a } \\
\end{array } \right ) , \\ \hat{a}_{\theta}''&=&\left ( \begin{array}{cc } \hat{n}\hat{a } & \cos\theta \\ ( \hat{n}\hat{a})^2\sin\theta & \hat{n}\hat{a } \\
\end{array } \right ) , \end{aligned}\ ] ] for @xmath203 , which correspond to two deformations of the sao of @xcite .
the eigenvalues @xmath92 of the matrix @xmath93 of eq . (
[ 40.41 ] ) become now : @xmath204 so that for @xmath205 both @xmath92 are real while for @xmath206 they are complex .
consider first the nonlinear supercoherent states of eq .
( [ 52.5 ] ) with @xmath207 ( _ i.e. _ , those associated to @xmath208 ) : @xmath209 in addition , we will analyze also the nonlinear supercoherent states of eq .
( [ 68 ] ) with the same @xmath150 and @xmath151 ( the ones associated to @xmath201 ) : @xmath210 for comparison , the square of the heisenberg uncertainty relation associated with the linear supercoherentes states of @xcite is shown in fig . [
fig : nl3 ] . for @xmath5 real ( left plot )
this quantity reaches a maximum value equal to @xmath211 at @xmath212 for @xmath92 real ( @xmath205 ) , while it shows a growing behavior as @xmath213 increases for @xmath92 complex ( @xmath206 ) .
moreover , for complex @xmath5 ( right plot ) , the square of the uncertainty relation behaves similarly as for @xmath5 real .
meanwhile , figs .
[ fig : nl4 ] and [ fig : nl5 ] show the same quantities @xmath214 and @xmath215 associated to the nonlinear supercoherent states of eqs .
( [ 73.1 ] ) and ( [ 74 ] ) , respectively . by comparing the results of figs .
[ fig : nl3 ] , [ fig : nl4 ] and [ fig : nl5 ] , it is clear that the uncertainty relation behaves differently , depending if either it is calculated for the linear supercoherentes states @xmath216 of @xcite or for the deformed supercoherent ones of eqs .
( [ 73.1 ] ) and ( [ 74 ] ) .
figure [ fig : nl4 ] indicates that for the nonlinear coherent states @xmath217 of eq . ( [ 73.1 ] ) the uncertainty relation reaches its maximum at the vicinity of @xmath218 , which marks the boundary between the real and complex eigenvalues @xmath219 ( degenerate case ) . on the other hand , fig .
[ fig : nl5 ] shows the opposite behavior , namely , the uncertainty relation decreases around the same @xmath220 value .
also , the uncertainty square for the nonlinear coherent states @xmath217 starts from the minimum value @xmath221 for @xmath222 and then increases slowly as @xmath223 grows , while for the states @xmath224 it departs from the value @xmath225 at @xmath226 , then it decreases for certain interval of @xmath227 in order to increase finally for sufficiently large @xmath227 .
let us note that , although eqs .
( [ 73.1 ] ) and ( [ 74 ] ) at first sight look similar as the expressions for the even and odd coherent states ( see e.g. @xcite ) , these superpositions can not be considered as the appropriate generalization of such states .
in fact , the coherent states @xmath228 ( @xmath229 ) are eigenvectors of the sao @xmath230 ( @xmath231 ) with the same eigenvalue @xmath232 ( @xmath8 ) while the even and odd coherent states are superpositions of coherent states with different eigenvalues . in order to determine the even and odd coherent states for the susy harmonic oscillator we must look for the eigenstates of @xmath233 and @xmath234 , which is a work currently in progress .
as it was shown in section [ sec : level3 ] , the susy harmonic oscillator executes an evolution loop , thus any state turns out to be cyclic and then it has associated a geometric phase . in particular
, this is valid for the nonlinear supercoherent states .
let us make next the explicit calculation of their corresponding geometric phases .
the states @xmath235 are cyclic , with @xmath9 and @xmath114 .
thus , using eq .
( [ gptih ] ) it is obtained the following expression for @xmath236 ( see fig .
[ fig : nl6 ] ) : @xmath237 the states @xmath238 are also cyclic with @xmath9 and @xmath114 . using eq .
( [ gptih ] ) it is obtained now the following geometric phase @xmath239 ( see fig . [
fig : nl7 ] ) : @xmath240\\ \nonumber&&\quad+\gamma^-\left[\frac{|\phi_-|^2}{2}\frac{_0f_2(2,3;| \phi_-|^2)}{_0f_2(1,2;|\phi_-|^2)}+1\right]\\ \nonumber&&\quad + 2\mathbf{re}\left(\frac{\gamma^{+-}}{\sqrt{_0f_2(1,2;|\phi_+|^2)\,_0f_2(1,2;|\phi_-|^2)}}\right.\\ \nonumber & & \quad\left.\times\left[\frac{\phi_+^\ast\phi_-}{2}\,_0f_2(2,3;\phi_+^\ast\phi_-)+\,_0f_2(1,2;\phi_+^\ast\phi_-)\right]\right)\\ & & \quad+|\gamma_{2+}z|^2+|\gamma_{2-}z|^2 + 2|z|^2\mathbf{re}(\gamma_{2+}^\ast\gamma_{2-})\bigg\ } \end{aligned}\ ] ] figures [ fig : nl6 ] and [ fig : nl7 ] show that the behavior of the geometric phases @xmath236 and @xmath239 depends strongly on the parameter @xmath220 .
in fact , both geometric phases grow quickly for @xmath241 while for @xmath242 they increase more slowly but with a less smooth character . on the other hand , in the limit @xmath243 and @xmath244 ,
both @xmath236 and @xmath239 reach the minimum value @xmath245 and @xmath246 , respectively , which is consistent with the fact that the mean value of any hamiltonian is at least equal to the minimum energy eigenvalue , _
i.e. _ , @xmath247 , where @xmath248 denotes the smallest energy involved in the nlscs .
if instead of @xmath13 we set in the harmonic oscillator algebra as annihilation operator either @xmath51 or @xmath64 , important consequences arise concerning the form of their respective eigenvectors ( coherent states ) as well as the associated algebraic structure . while for the usual operator @xmath13 its eigenstates have the form given in eq .
( [ 13 ] ) , the ones corresponding to @xmath35 take a less familiar form , involving now generalized hypergeometric functions , as can be seen in eqs .
( [ 15.03 ] ) and ( [ 23 ] ) .
this difference becomes more apparent when analyzing the uncertainty relation for the nonlinear coherent states : it has a minimum value , equal to @xmath0 , for the nonlinear coherent states of eq .
( [ 15.03 ] ) and it reaches a maximum value equal to @xmath70 for the states of eq .
( [ 23 ] ) ( see figs . [
fig : nl ] and [ fig : nl1 ] ) . on the other hand , the study of the eigenstates of the sao for the susy harmonic oscillator involves naturally generalizations of the standard coherent states @xcite .
as mentioned first in this paper , the analysis of the linear supercoherent states rests on the study of the matrix @xmath93 , given in eq .
( [ 40.41 ] ) , its matrix elements @xmath83 and the behavior of its eigenvalues @xmath92 .
thus , a general expression for the eigenvectors @xmath4 and their heisenberg uncertainty relation were found in @xcite .
here we have used the same treatment to find the nonlinear supercoherent states associated to the deformed sao of the system , which instead of involving the standard coherent states associated to @xmath13 depend now on the nonlinear ones for @xmath35 .
moreover , the heisenberg uncertainty relation for the nonlinear supercoherent states shows a different behavior as compared with the corresponding relation for the linear ones .
this is so since the mean values of the operators @xmath23 , @xmath24 and their squares become quite different for the several coherent states we are dealing with , either linear or nonlinear .
as it was mentioned in subsection [ subsec : level22 ] , the deformations we are studying are also connected with the well known @xmath33-oscillators which , in turn , are extensions of the @xmath249-deformations of the oscillator . for these systems
it turns out that the frequency depends on the amplitude through an arbitrary function @xcite .
this property is not shared by our susy harmonic oscillator since the associated hamiltonian is independent of the @xmath33-deformation , which does not happen for the nonlinear @xmath33-oscillators .
let us remark that , although @xmath250=-\omega\hat{a}$ ] and @xmath251=\omega\hat{a}^\dagger$ ] , the set @xmath252 neither generates a lie algebra nor a polynomial heisenberg algebra @xcite , since @xmath253 $ ] neither is a linear combination of @xmath1 , @xmath2 , @xmath3 nor is a polynomial in @xmath1 .
it remains open to determine to which algebraic structure the generators @xmath1 , @xmath2 , @xmath3 give place . finally , according to previous analysis for systems with equidistant spectrum @xcite , and taking into account that the susy harmonic oscillator has precisely this property , it has been possible to determine both the system evolution loop as well as the geometric phases @xmath10 associated to the nonlinear supercoherent states of eqs .
( [ 52.5 ] ) and ( [ 68 ] ) .
the authors acknowledge the support of conacyt .
edb also acknowledges the conacyt fellowship 290649 . | nonlinear supercoherent states , which are eigenstates of nonlinear deformations of the kornbluth - zypman annihilation operator for the supersymmetric harmonic oscillator , will be studied .
they turn out to be expressed in terms of nonlinear coherent states , associated to the corresponding deformations of the standard annihilation operator .
we will discuss as well the heisenberg uncertainty relation for a special particular case , in order to compare our results with those obtained for the kornbluth - zypman linear supercoherent states .
as the supersymmetric harmonic oscillator executes an evolution loop , such that the evolution operator becomes the identity at a certain time , thus the linear and nonlinear supercoherent states turn out to be cyclic and the corresponding geometric phases will be evaluated . |
the problem of the crossover from superfluidity to superconductivity with charge carrier density or coupling constant changing has a long history @xcite .
the interest to this phenomena has arisen again after the discovery of high - temperature superconductors ( htscs ) in 1986 @xcite .
it was already known upon that time that the superfluidity of composite bosons transforms into the superconductivity of overlapped cooper pairs with chemical doping in the case of the s - wave pairing .
now , in the s - wave pairing case the problem is quite well explored for the 3d systems @xcite and , particularly for the quasi-2d case @xcite . for the 2d case this problem was studied at @xmath6 ( when a long - range superconducting is still possible in a 2d system @xcite ) for the case of local attraction ( see , for example @xcite ) and for the phonon - exchange model @xcite .
most of these problems are reviewed in @xcite .
due to a layered structure and the anisotropic symmetry of the order parameter in htscs , the crossover in the 2d system in the @xmath1-wave pairing channel is of a special interest .
however , because if its complexity , this case is not so well understood at present .
the d - wave pairing for the case of the extended hubbard model with the nearest neighbor attraction case was analyzed in @xcite .
the crossover from superfluidity to superconductivity in the @xmath0-wave and @xmath1-wave pairing channels for a 2d continuum model was studied in paper @xcite , where also thermodynamic properties of the system in the crossover particle density region were considered .
the authors proposed an interaction potential , which is attractive at distances between particles shorter of some value @xmath7 and longer of @xmath8 , and repulsive due to electron correlations at short distances , @xmath9 .
it was found , in particular , that there exists a minimal value of the attractive coupling constant in the @xmath1-wave pairing channel , which gives the crossover from superfluidity to superconductivity at small carrier densities , i.e. the fermion chemical potential changes its sign and becomes positive with carrier density growth .
the correlation length @xmath7 was assumed in @xcite to be a parameter , which does not depend on the carrier density @xmath10 per cell .
however , concerning htscs it would be interesting to consider a more realistic case , when at small carrier densities @xmath3 and the proportionality coefficient is of order of a lattice constant . such a dependence was experimentally observed in htsc cuprates for the length of spin - spin correlations , which are believed to be responsible for the hole attraction in these materials .
in particular , it was found for @xmath4 in underdoped regime , that the magnetic correlation length decreases with carrier density per cell according to the dependence @xmath11@xmath12 @xcite . in what follows
we analyze the possibility of low carrier density crossover for a model analogous to @xcite with particle repulsion at distances @xmath13 and attraction at @xmath14 , where , however , @xmath15 ( @xmath16 is parameter of order of the lattice constant , and its possible value is discussed in the next section ) . obviously , at large carrier densities
such that @xmath17 the superconductivity in this system should disappear .
however , since we are interested in the small carrier concentrations , it is assumed that this relation does not take place .
as it will be shown below such a dependence of @xmath18 leads to a qualitatively different behavior of the system with respect to the case with @xmath19 .
in particular , there exists a minimal value for the coupling constant when the two - particle bound states exist at low carrier densities in the @xmath0-wave pairing channel .
the existence of such a threshold value of the coupling constant is typical for the @xmath1-wave pairing case ( see , for example @xcite ) .
another interesting property is : for any coupling constant in both channels there exists a corresponding carrier density value , below which the system is in the superconducting state .
the hamiltonian of the system which describes the non - retarded fermion interaction can be written in a standard form @xmath20 where @xmath21 is the effective fermion mass , and @xmath22 is the chemical potential ; fermi - operators @xmath23 depend on the space - time coordinate @xmath24 .
the instantaneous interaction potential is chosen in the next form @xmath25 with @xmath26 which corresponds to potential used in @xcite . here
@xmath27 is an inter - particle distance .
positive parameters @xmath28 and @xmath29 correspond to particle repulsion at @xmath9 and particle attraction at @xmath14 .
the charge carrier density dependence of the correlation radius @xmath7 is given by ( [ r0 ] ) with the parameter @xmath30 , @xmath31 is the square lattice constant .
this relation can be easily estimated from the equality @xmath32 , where on the left side the volume of the 2d system is expressed as a volume ( circle of the radius @xmath33 ) occupied by one particle , multiplied by the full number of particles @xmath34 , @xmath35 is an elementary cell number in the system .
the free fermion bandwidth @xmath36 is connected with @xmath37 as @xmath38 .
it should be noted , that the relation ( [ r0 ] ) at @xmath39 is in a good agreement with the experimental data for @xmath4 @xcite , where the plane magnetically ordered lattice parameters are equal to @xmath40 and @xmath41 , and the corresponding parameter @xmath16 is @xmath42 . in order to study the superconducting properties of the model in the channels with different pair angular momentum @xmath43
, it is convenient , similarly to @xcite , to approximate the fourier transform of ( [ magnoninteraction ] ) by a separable potential : @xmath44 where @xmath45 is an effective coupling constant , and functions @xmath46 with coefficients @xmath47 @xmath48 is the momentum modulus @xmath49 , and @xmath50 is its angle in polar coordinates @xmath51 .
parameters @xmath52 and @xmath53 put the momentum range in the proper region .
they are connected with the potential ( [ magnoninteraction ] ) parameters as @xmath54 and @xmath55 .
below we put @xmath56 and @xmath57 . obviously , the expression ( [ potentialk ] ) is not the exact fourier transform of ( [ magnoninteraction ] ) , but it sets the interaction in right momentum range and has the correct asymptotic behavior at small and large momenta : @xmath58 and @xmath59 , correspondingly . since we are mainly interested in the low carrier density region , where the crossover can take place , the correct behavior of the interaction potential at small momenta is the most important . in the case of low carrier concentrations
the small momenta give the main contribution in the integral of the gap equation ( see eq .
( [ gapequation ] ) below ) .
we shall study @xmath0- and @xmath1-wave channels with @xmath60 and @xmath61 separately , so we assume that the parameters @xmath45 for both channels are independent .
it will be assumed also that the pairing takes place at zero pair momentum at @xmath6 .
the minimization of the ground state energy with respect to the superconducting order parameter @xmath62 and the chemical potential gives in the case of the approximation ( [ potentialk ] ) the standard pair of equation for the crossover problem ( see , for example @xcite ) : @xmath63 @xmath64 where @xmath65 is a free particle dispersion law .
the form of the equation ( [ gapequation ] ) allows to search the solution for the superconducting order parameter in the next form @xmath66 where @xmath67 does not depend on momentum @xmath68 and @xmath69 is defined by ( [ w ] ) and ( [ h ] ) . in this case
the solution of the system ( [ gapequation]),([numberequation ] ) gives the dependence of the gap parameter @xmath67 ( @xmath70 for @xmath60 and @xmath71 for @xmath72 ) and the corresponding chemical potential @xmath22 on the particle density @xmath2 and the coupling @xmath73 .
the solution of the system ( [ gapequation]),([numberequation ] ) at @xmath74 gives the crossover line @xmath75 , which separates the parameters regions , where the local ( @xmath76 ) and cooper pairing ( @xmath77 ) take place .
note , that in the @xmath0-wave pairing case the coulomb repulsion parameter @xmath53 does not enter in the equations .
the numerical solution for the crossover line for the @xmath0-wave pairing case is presented in fig.1 , where we have put fermi energy @xmath78 instead of @xmath10 , since in the 2d case they are connected by a linear relation @xmath79 , and the dimensionless coupling constant @xmath80 is introduced .
the solution for the @xmath19 case is also presented .
as it follows from the numerical calculations , there exists a minimal value of coupling @xmath81 , which corresponds to carrier density @xmath82 on the crossover line , necessary to generate the crossover from superfluidity to superconductivity with doping , otherwise the cooper pairing regime takes place at any charge carrier density . in other words
, there is a minimal value of coupling constant which leads to the two - particle bound states in the @xmath0-wave pairing channel at small @xmath83 .
it is important to note , that there is no such a minimal coupling in the @xmath0-wave pairing channel when the correlation length @xmath19 @xcite .
moreover , the crossover with charge carrier density increasing in the @xmath0-wave channel takes place for any known doping independent interaction potential @xcite . another interesting property , which follows from the fig.1b ) ,
is the `` inverse '' crossover from superfluidity to superconductivity with charge carrier density decreasing at small values of @xmath78 ( @xmath84 ) .
this is a consequence of the competition between two opposite processes which occur with @xmath85 decreasing .
the lowering of @xmath78 tends the system to become a superfluid , but at the same time it leads to growing of @xmath18 and makes the pair size larger , i.e. the pairs become bounded weaker . on the other hand ,
the carrier localization on dopants at very small carrier densities also makes a possible density region of superfluidity more narrow @xcite .
however , the region of extremely small particle densities @xmath86 is not very interesting from the point of view of the connection with htsc , since in this region the relation @xmath87 does not hold , and the model is not correct .
as it will be shown in the next subsection , in the @xmath1-wave case this crossover from from superfluidity to superconductivity with doping decreasing takes place at rather large fermi energy values .
the doping dependencies of the gap @xmath70 and the chemical potential @xmath22 at different values of the coupling constant @xmath88 are presented in figs
. 2 , and 3 . the gap is increasing with the doping almost linearly , except the region of extremely low concentrations .
it is interesting to note , that the chemical potential in the superfluidity region ( @xmath76 ) has a minimum at a finite value of @xmath83 and it is equal to zero at @xmath89 , since the effective coupling constant at low doping @xmath90 is zero at @xmath89 . in other words ,
the local pairs are the most strongly coupled at some finite carrier density value .
this is also a consequence of the competition between the fermi surface formation and correlation radius decreasing with charge carrier density growth .
this situation is qualitatively different from the @xmath19 case , where @xmath91 does not depend on particle density and renormalized coupling constant is @xmath83-independent in the @xmath0-channel . in this case
@xmath92 at @xmath89 , where @xmath93 is a two - particle bound state energy . in this case
the coulomb repulsion parameter @xmath94 is present in the equations ( [ gapequation ] ) and ( [ numberequation ] ) .
however , the presence of this parameter leads just to renormalization of the dimensionless coupling constant @xmath95 and the energy gap parameter : @xmath96 , @xmath97 , where @xmath98 is characteristic energy of coulomb repulsion . we shall consider the case , when the coulomb repulsion is much smaller than @xmath36 , i.e. this is the case of a large free - fermion bandwidth
therefore , the coupling constant is assumed to be large : @xmath99 ( see figs .
the crossover line @xmath100 for the @xmath1-wave pairing case is presented in fig.4 .
qualitatively , the behavior of the system with doping and coupling changing in the @xmath1-wave pairing channel is similar to the @xmath0-wave case .
the important difference is that the low carrier density superconducting state exists at rather high values of @xmath83 .
also in this case there is a minimal value of the coupling constant for two - particle bound states @xmath101 at @xmath102 .
it should be noted that the existence of the large threshold value for the coupling constant in both channels can be a possible answer on the question why the crossover has been not observed in cuprates .
the charge carrier density dependence of @xmath103 and @xmath22 at different coupling parameters @xmath104 are presented in figs.5 and 6 .
the superconducting gap @xmath71 at low charge carrier densities is much smaller than in the @xmath0-case , due to stronger @xmath83-dependence of effective coupling constant at low carrier densities ( [ renormg ] ) .
the magnitude of the order parameter @xmath103 starts to grow almost linearly with @xmath83 increasing when the fermi energy is larger of some minimal value ( fig.5 ) .
this behavior is qualitatively similar to the doping dependence of the gap of cuprates in the underdoped regime .
however , because of its simplicity , the model ca nt describe the decreasing of the gap with charge carrier density increasing at large values of @xmath2 htscs . for this other properties of the charge carrier interaction in cuprates
have to be taken into account ( see the last section ) .
the small-@xmath83 superconducting region with @xmath105 as well as region of superfluidity are rather large in the @xmath1-wave pairing channel .
also in this case the chemical potential is equal to zero at @xmath89 , i.e. there are no two - particle bound states at very low charge carrier densities .
it should be noted that the decreasing of the chemical potential with increasing charge carrier density at small @xmath83 both in @xmath0- and @xmath1-wave pairing cases indicates a negative electronic compressibility .
this can be related to increasing of antiferromagnetic correlations at low carrier densities .
theoretical description of the behavior of superconductor with carrier density changing is an interesting and important problem , in particular , because of its possible association with htscs .
the microscopic mechanism of the superconductivity in cuprates is not known so far , and the solution of phenomenological models , which take into account some of the properties of htscs , can help to clarify the nature of their unusual behavior , and maybe even help to understand the microscopic mechanism of the htsc phenomenon . in this paper
the possibility of the crossover from superfluidity to superconductivity with charge carrier density and coupling constant changing in different pairing channels at @xmath6 was studied for a model , which qualitatively takes into account one of the properties of htscs , namely , the doping dependence of correlation length @xmath7 at low carrier densities .
it has been shown , that even this simple model results in interesting and unusual properties , which are rather different from a more standard case with @xmath5 .
in particular , the two - particle bound states in the @xmath0-channel exist only if the coupling constant is larger of the threshold value , similar to the @xmath1-wave pairing case . at any value of coupling constant larger of the threshold one ,
the `` inverse '' crossover from superfluidity to superconductivity takes place with doping decreasing in both @xmath0- and @xmath1-wave channels . of course
, such a simple model can not pretend to describe doping dependence of the gap and chemical potential of htscs .
the momentum dependence of the interaction potential has to be taken into account more carefully , especially in the overdoped regime , where @xmath78 is rather large and the separable potential may be not correct . in general , also the effect of the retardation of interaction can not be neglected .
these and some other questions are planed to be studied in a future work .
v.t . thanks the cfif members , especially prof . p.d .
sacramento and prof .
vieira for kind hospitality .
is partly supported by scopes - project 7ukpj062150.00/1 of the swiss national science fundation .
eagles , phys .
186 ( 1969 ) 456 ; a.j .
leggett , in : a. pekalski and j. przystawa ( eds . ) , modern trends in the theory of condensed matter , springer - verlag , berlin , 1980 , p.13 ; p. nozieres and s. schmitt - rink , j. low temp . phys .
59 ( 1985 ) 195 . | zero temperature crossover from superfluidity to superconductivity with carrier density increasing is studied for a two - dimensional system in the @xmath0-wave and @xmath1-wave pairing channels .
it was assumed that the particle attraction correlation length depends on carrier density @xmath2 as @xmath3 .
such a dependence was found experimentally for the radius of magnetic correlations in @xmath4 .
the short range coulomb repulsion was also taken into account .
it is shown that the behavior of the system with doping is fundamentally different from the case with @xmath5 .
in particular , similarly to the @xmath1-wave case , the crossover in the @xmath0-channel takes place only if the coupling is larger of some minimal value , otherwise the cooper pairing scenario takes place at any small carrier density .
the relevance of the model to the high - temperature superconductors is discussed . : 74.20.-z , 74.62.dh , 74.72.-h _ keywords _ : symmetry , @xmath0-wave , @xmath1-wave , effects of doping |
in four dimensional @xmath0 theory the period map assigns positive real numbers to 4-regular internally 6-connected graphs @xcite .
the periods determine the contributions of primitive logarithmically divergent graphs to the beta function of the underlying quantum field theory .
although periods are originally associated to sub - divergence free four - pointgraphs , it is convenient to complete the graph by adding an extra vertex ( henceforth labeled @xmath1 ) and gluing this vertex to the four external ( half-)edges of the graph . as a remnant of conformal symmetry one can consider the completed graph as an equivalence class of four - point graphs ( obtained by opening up a vertex ) with the same period ( see figure 1 ) .
figure 1 : completed ( @xmath2 ) and uncompleted ( @xmath3 ) zig - zag graphs with five and six loops .
2ex the structure of @xmath0 periods were first studied systematically by broadhurst and kreimer in 1995 @xcite when they found by exact numerical methods that up to seven loops ( the number of independent cycles in the four - point graph ) many periods are multiple zeta values ( mzvs ) , i.e. rational linear combinations of multiple zeta sums @xmath4 multiple zeta values span a @xmath5 vector space @xmath6 which is conjectured to be graded by the weight @xmath7
. there are four equivalent ways to define the period of a graph : position space , momentum space , parametric space , and dual parametric space @xcite . in the context of algebraic geometry
one often uses feynman ( or schwinger ) parameters @xcite , @xcite . in this article , however , it is essential to use position space .
figure 2 : position space feynman rules : edges correspond to quadrics ( or 1 ) .
2ex in a completed feynman graph we label the @xmath8 vertices by @xmath9 @xmath10 @xmath11 @xmath12 @xmath13 @xmath14 , where 0 is the origin of @xmath15 , 1 corresponds to a unit vector @xmath16 , and @xmath17 for @xmath18 .
edges of the feynman graph correspond to propagators .
the propagator is the constant function 1 if the edge has one vertix @xmath1 or if it connects 0 and 1 .
otherwise the propagator of an edge @xmath19 from @xmath20 to @xmath21 is the inverse of the quadric @xmath22 the period of the completed feynman graph @xmath23 is then given by the @xmath24-dimensional integral @xmath25 where the products are over vertices @xmath26 and edges @xmath19 of @xmath23 . because edges that connect to infinity do not contribute to the integral we can remove infinity before we apply the feynman rules .
it is often useful to generalize the definition of a period to non-@xmath0 graphs .
the fourier identity @xcite , for example , maps @xmath0 graphs to non-@xmath0 graphs without changing the period . this map will be used to link the period of the zig - zag graphs to the sequential family of graphs defined in the next subsection .
although completion is also possible for non-@xmath0 graphs if one introduces edges of negative weights ( similar to [ completion ] ) we use non-@xmath0 graphs in an uncompleted form . in this form graphs
have no vertex @xmath1 and the definition of the period ( [ pdef ] ) remains unchanged .
we use capital greek letters for completed and latin letters for uncompleted graphs .
position space feynman rules were used in @xcite and in most calculations of feynman periods .
in fact , so far only eight @xmath0 periods have been calculated : the trivial period 1 ( @xmath27 ) , the wheels with three @xcite and with four @xcite spokes which are @xmath28 and @xmath29 , respectively .
the period of the zig - zag graph ( see figure 1 ) with 5 loops was calculated by kazakov in 1983 @xcite .
the 6 loop zig - zag period was derived by broadhurst in 1985 @xcite and confirmed by ussyukina in 1991 @xcite .
the only calculated non zig - zag periods are @xmath30 in @xcite which is @xmath31 in @xcite , @xmath32 by e. panzer in 2012 , and the bipartite graph @xmath33 in @xcite which is @xmath34 in @xcite . by exact numerical methods
24 more periods are known up to eight loops @xcite , @xcite all of which are mzvs .
the zig - zag periods were conjectured to all orders in @xcite : ( zig - zag conjecture ) .
the period of the graph @xmath35 is given by @xmath36 in spite of its seeming simplicity the zig - zag conjecture remained open for 17 years .
although all known @xmath0 periods are mzvs it is conjectured with strong mathematical evidence that @xmath0 periods in general are non mzv @xcite , @xcite . in this article
we develop a method that allows one to calculate the zig - zag periods and two more families of @xmath0 periods ( and sequential non-@xmath0 periods ) .
modulo products of mzvs an explicit formula for these periods is given .
for a more general class of constructible periods we present a computer algorithm that works up to ten loops . finally , the zig - zag conjecture is proved in @xcite using corollary [ maincor ] .
an important family of graphs can be encoded by words in the three letter alphabet @xmath37 .
figure 3 : the sequential graph @xmath38 .
2ex let @xmath39 be a word in 0,1,2 .
the sequential graph @xmath40 is the graph with two distinguished vertices 0 and 1 and a horizontal chain of vertices that connect either to 0 , to 1 , or to both 0 and 1 ( see figure 3 ) .
reading from left to right the connections are encoded in the letters 0 , 1 , or 2 , respectively .
sequential graphs have well - defined periods @xmath41 in four dimensions if the word @xmath39 begins and ends in 2 ( corollary [ pwelldef ] ) .
certain sequential graphs are related by duality to zig - zag graphs . with the notation @xmath42 for the @xmath43 fold iteration of a word @xmath39
the graphical dual @xmath44 of the uncompleted zig - zag graph with @xmath43 loops is @xmath45 where @xmath46 is the edge 01 .
because the period of an uncompleted graph with a planar embedding equals the period of its planar dual @xcite , @xcite we can express the zig - zag period in terms of the period of a sequential graph , @xmath47 sequential graphs were independently analysed in @xcite .
a graphical function is the evaluation of a graph @xmath48 with three distinguished vertices @xmath49 with position space feynman rules and no integration over @xmath50 , @xmath51 here we assume four dimensions .
arbitrary dimensions greater than two are considered in
a special case is graphical functions in two dimensions which we will define in [ 2d ] . by symmetry @xmath52
depends only on two real parameters : the norm @xmath53 and the angle between @xmath50 and the unit vector @xmath16
. we can hence consider @xmath52 as a function on the complex plane @xmath54 where we identify @xmath16 with 1 and choose any of the two possible orientations .
graphical functions are single - valued functions which in many cases can be proved to be real analytic in @xmath55 .
independence of the orientation of @xmath54 results in the reflection symmetry @xmath56 figure 4 : the graphical functions @xmath57 and @xmath58 .
2ex sequential graphs give rise to sequential functions by appending a horizontal edge connected to @xmath50 ( see figure 4 ) .
the four dimensional graphical function @xmath59 serves as an initial case for constructing sequential functions .
the sequential function @xmath60 is well - defined in four dimensions if the word @xmath39 begins with 2 ( lemma [ convergencelemma ] ) .
zig - zag periods can be expressed as special values of sequential functions at 0 or 1 , @xmath61 for any word @xmath39 in the two letter alphabet @xmath62 we inductively define multiple polylogarithms @xmath63 by @xmath64 and @xmath65 unless @xmath66 in which case we have @xmath67 .
multiple polylogarithms are multi - valued analytic functions on @xmath55 .
the weight of @xmath68 is the length @xmath69 of the word @xmath39 . by taking appropriate linear combinations of products of multiple polylogarithms with their complex conjugates
one can construct single - valued multiple polylogarithms ( svmps ) , see @xcite .
svmps span a shuffle - algebra over @xmath54 which is graded by the total weight @xmath70 for every word @xmath39 in 0 and 1 of length @xmath43 there exists a basis element @xmath71 . by construction the regularized ( setting @xmath72 ) limit of svmps at 0 vanishes . with the notation @xmath73
we have @xmath74 . hence at 1 ( and at @xmath1 ) svmps evaluate to mzvs . for any ring
@xmath75 we set @xmath76 the @xmath5-algebra @xmath77 is a proper subalgebra of @xmath6 , the @xmath5 algebra of mzvs .
assuming transcendentality conjectures one can construct @xmath78 as the largest @xmath5 sub - algebra of @xmath6 with only odd weight generators on which the coaction on @xmath6 coacts @xcite . by holomorphic and antiholomorphic differentiation @xmath79 and @xmath80 svmps
generate a bi - differential algebra @xmath81{{\mathcal p}}.\ ] ] we present results of the theory of svmps ( many of which are due to f. brown @xcite , @xcite ) in [ svmps ] .
in particular there exist integrals with respect to @xmath50 and @xmath82 in @xmath83 , @xmath84 we give an algorithm that allows one to integrate in @xmath83 up to high weights ( @xmath85 30 ) .
a residue theorem in [ residue ] facilitates the integration of functions in @xmath83 over the complex plane . in practice the integration over
the complex plane is more memory and time consuming so that the implementation is limited to smaller weights .
the connection to graphical functions is established by the fact that in many cases graphical functions are expressible in terms of svmps in the sense that @xmath86 with @xmath87 .
let @xmath88 denote the vector space of such @xmath89 .
likewise we define @xmath90 as the set of functions ( [ fgina ] ) where @xmath91 . an example is @xmath92 , ( [ fi ] ) , which is in @xmath88 but not in @xmath93 . by corollary [ maincor ] all sequential functions @xmath94 are in @xmath93 .
there exist relations between graphical functions of different graphs .
the best way to formulate these relations is by completing the graph .
the completion is obtained by adding a vertex @xmath1 in much the same way as for periods .
the completed graph has four labeled vertices @xmath95 .
for some edges @xmath19 in the completed graph we need to introduce propagators of negative integer weight @xmath96 which correspond to quadrics @xmath97 in the numerator of the integral ( [ fdef4completed ] ) .
graphically we indicate negative weight propagators by multiple curly lines .
completion adds edges to @xmath1 and from 0 to 1 in such a way that all unlabeled vertices have total degree four and the vertices @xmath95 have total degree zero ( see figure 5 ) .
completion is always possible and unique ( lemma [ completionlemma ] ) .
figure 5 : the completion of the graphical functions @xmath57 and @xmath58 .
2ex in four dimensions the graphical function of a completed graph @xmath23 is given by @xmath98 clearly , a graphical function does not change under completion , @xmath99 because only completed graphs have a vertex @xmath1 we use the same symbol for their graphical functions .
now , a permutation of the four labels @xmath95 changes the graphical function only by a mbius transformation of the argument that permutes the points @xmath100 in @xmath101 .
more precisely , we denote the group of permutations of four elements by @xmath102 .
it has a normal subgroup @xmath103 with factor group @xmath104 , permuting the three points @xmath100 .
this @xmath104 acts on the argument @xmath50 by the transformations @xmath105 .
concretely , the @xmath102 is generated by the three transpositions @xmath106 where we indicate the action of the @xmath104 by transforming the label @xmath50 .
the action on @xmath50 is such that it fixes the cross - ratio of the four labels .
it maps @xmath88 into @xmath88 and @xmath93 into @xmath93 so that the class of graphical functions that is expressible in terms of svmps in the sense of ( [ fgina ] ) is invariant under the permutation of external labels . figure 6 : appending an edge to the vertex @xmath50 in @xmath48 gives @xmath107 .
2ex a key result in the theory of graphical functions is that in many cases the graphical function is mapped from @xmath88 into @xmath93 by appending an edge to the vertex @xmath50 ( see figure 6 ) . under quite general assumptions given in theorem [ appendthm ]
we obtain @xmath108 the simplicity of this equation is special to four dimensions . in arbitrary dimensions appending an edge
is significantly more complicated , see proposition [ appendprop ] .
we have seen that permuting the labeled vertices and appending an edge to @xmath50 maps @xmath88 into @xmath88 .
a third operation that is trivially of this type is adding an edge that connects two labeled vertices .
in fact , this edge contributes non - trivially to the graphical function only when it connects 0 or 1 to @xmath50 . in this case
the graphical function picks up a factor of @xmath109 or @xmath110 , respectively .
the empty graph with four vertices is the completion of the sequential graph of the empty word .
its graphical function is one , @xmath111 with the help of the above three transformations we can construct graphs from the empty graph and calculate their graphical functions ( see figure 7 ) .
figure 7 : construction of graphical functions which are in @xmath88 .
the bullets @xmath112 stand for one of the four labels @xmath95 and @xmath113 is a permutation of @xmath95 together with a transformation of @xmath50 which is generated by ( [ s4trafos ] ) .
2ex for these constructible graphs the graphical functions can be calculated by computer up to weights @xmath114 which corresponds to 15 internal ( unlabeled ) vertices .
in particular , all sequential graphs are constructible .
the periods of completed primitive feynman graphs can be calculated by using two - dimensional complex integration if the graph decomposes into at most two constructible graphs under the following steps : * label four vertices by @xmath95 . *
delete @xmath95 and decompose the graph into its connected components .
* add @xmath95 to each connected component in the same way the vertices of the component are connected to @xmath95 in the original graph .
* complete each component by adding edges between labeled vertices .
such constructible periods can be calculated by computer up to ten loops .
we argue ( but not fully prove ) in remark [ rsvremark ] that constructible periods are in @xmath115 .
although integration in @xmath83 is well - suited for computer calculation , closed results are hard to obtain .
an all orders result for sequential functions is only available in three cases : for the case of a 2 followed by a sequence of 0s ( or 1s ) , see example [ 0ex ] .
a 2 followed by a string of 0s which contains one 1 , see @xcite , and for the zig - zag graphs in @xcite leading to the proof of the zig - zag conjecture . one practical option to obtain general results
is to calculate the ideal @xmath116 generated by mzvs of weights between two and @xmath43 .
iterating ( [ inteq1 ] ) gives results for all sequential functions @xmath94 modulo @xmath117 . by setting the external variable @xmath50 to 1
this result leads to a formula for the periods of sequential graphs modulo products of mzvs : if we extend the definition of @xmath118 , ( [ zetaw ] ) , to the letter 2 by @xmath119 for words
@xmath120 in 0,1,2 we obtain for the sequential period of a word @xmath121 ( see theorem [ pgthm ] ) @xmath122 where @xmath123 is @xmath39 in reversed order .
the periods of three families of @xmath0 graphs are sequential .
one family is the zig - zag family ( see ( [ pzigzag ] ) and figure 1 ) .
equation ( [ modproducts1 ] ) gives in this case ( proposition [ modprop ] ) @xmath124 due to a result by zagier on mzvs of the above type @xcite ( see also @xcite ) this proves the zig - zag conjecture modulo products .
figure 8 : the completed @xmath125 and @xmath126 families of @xmath0 graphs .
2ex the second and the third family arise from alternating words in 0 and 1 with one internal letter 2 ( on the left of figure 8) .
the @xmath125 type sequence has different letters immediately on the left and the right of the letter 2 whereas the @xmath126 type sequence has equal letters .
the @xmath125 and @xmath126 graphs are the planar duals of the sequential graphs ( on the right of figure 8) .
the number of letters in the left and right sequences is one less than the number of internal arcs on the left and right side of the @xmath125 and @xmath126 graphs .
we have @xmath127 , @xmath128 , @xmath129 ; otherwise the graphs are non - isomorphic .
their periods are given modulo products by mzvs of 2s with 1 or 3 in three slots , see e.g. ( [ aexample ] ) . by corollary [ pw ] the periods of the @xmath125 and
@xmath126 families are in @xmath115 .
they can be calculated and reduced to a standard basis of mzvs up to loop order 12 .
type @xmath125 and @xmath126 periods are examples of @xmath0 periods which assuming standard transcendentality conjectures are proved up to 12 loops to be mzvs which can not be expressed in terms a single riemann zeta sum .
a special case are graphical functions in two dimensions . whereas in higher dimensions
propagators correspond to bosonic particles , in two dimension we consider holomorphic and antiholomorphic propagators which are more closely related to fermionic particles .
a definition of two - dimensional graphical functions is given in
[ 2d ] . in [ concellzeta ]
it is conjectured that the maximum weight piece of the periods of graphical functions in two dimensions reduce modulo products to the sum of two cell zeta values @xcite .
all algorithms of this article are implemented in maple and available under @xcite . in particular constructible periods
can be calculated by a @xmath130 command .
1ex * acknowledgements . *
the article was written while the author was visiting scientist at humboldt university , berlin .
he is highly indebted to francis brown for sharing his knowledge on svmps and for many very valuable discussions .
let @xmath131 be a commutative unitary ring .
consider the two letter alphabet @xmath132 and let @xmath133 be the set of words in @xmath134 together with the empty word 1 .
the shuffle algebra @xmath135 is the free @xmath131-module over @xmath133 together with the shuffle product which is defined recursively by @xmath136 and @xmath137 for all @xmath138 , and @xmath139 .
the shuffle product , extended linearly , makes @xmath140 into a commutative unitary ring .
a lyndon word is a non empty word @xmath141 which is inferior to each of its strict right factors ( for the lexicographical ordering ) , i.e. if @xmath142 , @xmath143 then @xmath144 . by radford s theorem
the @xmath5-algebra @xmath145 is the polynomial algebra generated by lyndon words . for every word @xmath146 let @xmath123 denote the word in reversed order .
we linearly extend @xmath147 to elements in @xmath140 . the length ( i.e. the number of letters ) of a word @xmath39 is @xmath69 .
the shuffle algebra is graded by the length .
the deconcatenation coproduct is defined to be the linear map @xmath148 and the antipode is the linear map defined by @xmath149 . with these definitions
, @xmath140 is a commutative , graded hopf algebra over @xmath131 .
the dual of @xmath135 is the @xmath131-module of non - commutative formal power series @xmath150 equipped with the concatenation product .
we define on @xmath151 a ( completed ) coproduct @xmath152 for which the elements @xmath153 are primitive : @xmath154 for @xmath155 .
the same antipode as in @xmath140 turns @xmath156 into a completed cocommutative but not commutative hopf algebra .
the duality between @xmath157 and @xmath158 is defined as @xmath159 the set of lie monomials in @xmath160 is defined by induction : the letters @xmath161 , @xmath162 are lie monomials and the bracket @xmath163=xy - yx$ ] of two lie monomials @xmath164 and @xmath165 is a lie monomial . a lie polynomial ( respectively a lie series )
is a finite ( respectively infinite ) @xmath131-linear combination of lie monomials .
the set @xmath166 of lie polynomials is a free lie algebra and the set of lie series @xmath167 is its completion with respect to the augmentation ideal ker@xmath168 , where @xmath169 projects onto the empty word .
the bracket form of lyndon words is recursively defined as @xmath170 for all @xmath171 and @xmath172 $ ] if @xmath173 for lyndon words @xmath120 and @xmath26 being as long as possible .
a basis for @xmath166 is given by the bracket forms @xmath174 of lyndon words .
an invertible series @xmath175 ( i.e. , with invertible leading term @xmath176 ) is group - like if @xmath177 .
equivalently , the coefficients @xmath178 of @xmath179 define a homomorphism for the shuffle product : @xmath180 for all @xmath181 , where @xmath182 is extended by linearity on the left - hand side .
the condition for @xmath183 to be group - like is equivalent to the condition that @xmath179 is a lie exponential , i.e. that there exists an @xmath184 such that @xmath185 . by the formula for the antipode
, it follows that for such a series @xmath186 , its inverse is given by @xmath187 in the following we often use the letters 0 and 1 for @xmath161 and @xmath162
. in @xcite chen develops the theory of iterated path integration on general manifolds .
here we need only the elementary one - dimensional case . for a fixed path @xmath188{\longrightarrow}{{\mathbb c}}\backslash\{0,1\}$ ] from @xmath189 to @xmath190 and differential forms @xmath191 and @xmath192 we define @xmath193 ( where the simplex @xmath194 is endowed with the standard orientation and @xmath195 is the pullback of @xmath196 by @xmath197 ) as the iterated path integral of the word @xmath198 along @xmath197 .
iterated path integrals have the following properties : i0 : : @xmath199 ( by definition ) .
i1 : : @xmath200 is independent of the parametrization of @xmath197 .
i2 : : @xmath200 is a homotopy invariant ; it only depends on the homotopy class of @xmath197 .
i3 : : @xmath201 for the constant path @xmath202 and @xmath203 .
i4 : : @xmath204 ( path reversal ) .
i5 : : @xmath205 ( path composition ) .
i6 : : @xmath206 ( shuffle product ) where @xmath207 is the set of @xmath208-shuffles : with @xmath209 and @xmath210 the group of permutations of @xmath211 , @xmath212 i7 : : @xmath213 ( chain rule ) where @xmath214 is a mbius transformation that permutes the singular points @xmath100 and @xmath215 .
concretely , @xmath214 is one out of the six transformations @xmath105 which induce the transformations of letters 0,1 ( extended linearly to words and to iterated integrals ) @xmath216 i8 : : @xmath217 by * i6 * the vector space of iterated path integrals for fixed @xmath197 is a shuffle algebra .
it is hence convenient to form the generating ( chen ) series @xmath218 where we identify the letters @xmath161 and @xmath162 with 0 and 1 in @xmath219 . from the properties of the iterated path integral
@xmath220 inherits the properties s1 : : @xmath220 is independent of the parametrization of @xmath197 .
s2 : : @xmath220 is a homotopy invariant .
s3 : : @xmath221 for the constant path @xmath197 .
s4 : : @xmath222 , where the inverse @xmath223 is @xmath197 with reversed orientation .
s5 : : @xmath224 , if @xmath225 and @xmath226 is the path @xmath227 followed by @xmath228 .
s6 : : @xmath220 is a lie exponential : @xmath229 .
s7 : : @xmath230 for a mbius transformation @xmath214 that permutes @xmath100 .
the action of @xmath214 on @xmath161 , @xmath162 is dual ( transpose ) to the action on 0 and 1 in ( [ fmap ] ) , @xmath162 .
] , @xmath231 s8 : : @xmath232 if @xmath233 and @xmath234 .
if the homotopy class of the path @xmath197 from @xmath235 to @xmath236 is clear from the context we write @xmath237 and consider @xmath179 as a function of the initial and the end - point . if the path @xmath197 approaches the singular values 0 and 1 the following limits are well - defined @xcite , @xmath238 to write the above equations in more convenient form we define regularized limits by nullifying every positive power of @xmath239 in the limit @xmath240 . with this notation
we have @xmath241 and @xmath242 .
regularized iterated integrals for the path @xmath243\to[0,1]$ ] span the shuffle algebra @xmath6 of mzvs , @xmath244 regularization extends @xmath6 to all words .
in particular , @xmath245 .
if @xmath39 begins with 1 and ends with 0 then @xmath118 can be converted into a sum by ( [ zetaw ] ) and ( [ mzv ] ) .
the generating function of regularized mzvs ( denoting @xmath246 as indices ) @xmath247 is drinfeld s associator .
it is group - like and by ( [ sinversion ] ) and ( [ fmapdual ] ) for @xmath248 , @xmath249 the weight @xmath69 of @xmath118 induces a filtration on @xmath6 .
the conjecture that the weight gives a grading on @xmath6 is proved in the motivic analogue .
we define subspaces @xmath250 of weight @xmath251 mzvs .
shuffling the summation indices of a product of two mzvs in the sum representation yields the set of quasi - shuffle identities .
conjecturally regularized shuffle and quasi - shuffle relations generate all relations between mzvs .
the dimension of @xmath252 is conjectured ( and proved in the motivic analogue @xcite ) to have the following generating series @xmath253 by the shuffle identity mzvs span a ring over the integers , @xmath254 an important factor algebra of @xmath6 is obtained by factorizing out the ideal generated by @xmath255 .
this factor algebra coacts on @xmath6 by @xmath256 an explicit formula for @xmath257 is given in @xcite .
the reduction of @xmath257 modulo @xmath255 on both sides of @xmath258 in ( [ coprod ] ) turns @xmath259 into a hopf algebra .
a main conjecture on mzvs ( a theorem for the motivic analogue ) states that this hopf - algebra is non - canonically isomorphic to the shuffle hopf - algebra on words in letters of odd weight greater or equal three @xcite .
conventionally one uses the @xmath214-alphabet @xmath260 , @xmath261 , for the letters .
the isomorphism into the @xmath214-alphabet extends to @xmath6 by adding a generator @xmath262 of weight two which generates a polynomial algebra .
this polynomial algebra is tensored to the right to the hopf - algebra in odd letters so that an mzv maps under the isomorphism to a sum of words in odd letters concatenated to the right by a power of @xmath262 .
the coaction @xmath257 deconcatenates the word in the @xmath214-alphabet and fulfills @xmath263 .
the number of odd letters in a word @xmath39 in the @xmath214-alphabet is the coradical depth of @xmath39 .
the coradical depth gives the @xmath214-alphabet a second grading ( together with the weight ) .
a computer implementation of the basis dependent isomorphism into the @xmath214-alphabet is in @xcite .
multiple polylogarithms for words @xmath39 in the letters 0 and 1 are recursively defined by @xmath264 with initial condition @xmath65 unless @xmath39 is a sequence of zeros in which case we have @xmath67 .
multiple polylogarithms are multi - valued analytic functions on @xmath265 with monodromies around 0 , 1 , and @xmath1 .
they can be expressed as regularized iterated integrals from 0 to @xmath50 , @xmath266 where the dependence on the homotopy class of @xmath197 gives rise to the multi - valuedness of the multiple polylogarithm .
if @xmath50 is in the unit ball @xmath267 and not on the negative real axis we assume that @xmath197 is homotopic to a straight line from 0 to @xmath50 . the generating series of multiple polylogarithms is denoted by @xmath268 where we often suppress the indices @xmath246 .
it is the unique solution to the knizhnik - zamolodchikov equation @xcite @xmath269 which satisfies the asymptotic condition ( see @xcite where the opposite convention is used : differentiation of @xmath63 corresponds to deconcatenation of @xmath39 to the left . ) @xmath270 for all @xmath50 in the neighborhood of the origin , where @xmath271 is a function taking values in @xmath272 which is holomorphic at @xmath273 and satisfies @xmath274 .
the series @xmath275 is a group - like formal power series . in particular , the polylogarithms @xmath63 satisfy the shuffle product formula @xmath276 drinfeld s associator @xmath277 is the regularized limit of @xmath275 at the point @xmath278 .
[ h1 ] there exists a function @xmath279 taking values in the series @xmath280 , which is holomorphic at @xmath278 where it takes the value @xmath281 , such that @xmath282 for @xmath283 we obtain from ( [ fmapdual ] ) for @xmath248 that @xmath284 . by * s5 * we have @xmath285 and from * s4 * , ( [ sinversion ] ) and ( [ zid ] ) this equals @xmath286 . depending on the sheet of @xmath287 at @xmath278
there exists a @xmath288 such that @xmath289 which has the required properties . for @xmath290 ,
let @xmath291 denote analytic continuation around a path winding once around the point @xmath292 in the positive direction .
the operators @xmath291 act on the series @xmath275 and its complex conjugate @xmath293 , commute with multiplication , @xmath79 , and @xmath294 .
[ lemmonodromy ] @xcite .
the monodromy operators @xmath295 , @xmath296 act as follows : @xmath297 the formulae follow from @xmath298 , @xmath299 and the equations @xmath300 , @xmath301 .
a sum representation of multiple polylogarithms in the unit ball can be derived from @xmath302 which expresses @xmath303 in terms of the classical polylogarithm in the case @xmath304 .
we have the following explicit expression of @xmath68 in terms of the @xmath305s .
@xmath306 where @xmath307 . with the above expression for @xmath287 we have @xmath308 . because @xmath67 and @xmath65 if @xmath309 the lemma follows . for @xmath304
we obtain @xmath310 multiple polylogarithms @xmath63 can be combined with their complex conjugates @xmath311 to kill the monodromy at 0 , 1 , and @xmath1 , rendering the function single - valued on @xmath265 . because the monodromy ( [ mon ] ) is homogeneous in the weight ( if one gives @xmath113 the weight one ) a single - valued expression will decompose into components of pure weight , where the weight of a product of multiple polylogarithms with mzvs is the sum of the weights of the polylogarithms plus the weight of the mzvs ( see examples [ wt2ex ] , [ lox0 ] and corollary [ homcor ] ) .
the vector space @xmath312 of single - valued multiple polylogarithms ( svmps ) has a direct decomposition with respect to the weight @xmath70 at weight one single - valued logarithms are @xmath313 and @xmath314 .
they form the two - dimensional vector space @xmath315 .
the differential operator @xmath79 maps svmps into the differential algebra of svmps over @xmath316 $ ] .
likewise antiholomorphic differentiation @xmath294 generates an @xmath317 algebra .
together @xmath318 is the @xmath319 bi - differential algebra generated by svmps .
it is a direct sum of its weighted components , @xmath320 where functions in @xmath321 have weight zero . because @xmath79 ( or @xmath294 ) decreases the weight of a svmp by one
while generating a denominator @xmath50 or @xmath322 it is clear that the dimension of the vector space @xmath323 is at most @xmath324 .
the following theorem states that @xmath312 is a shuffle algebra with @xmath325 .
[ francisthm ] there exists a unique family of single - valued functions @xmath326 , each of which is an explicit linear combination of the functions @xmath327 where @xmath328 , which satisfy the differential equations @xmath329 such that @xmath330 , @xmath331 for all @xmath332 , and @xmath333 if @xmath39 is not of the form @xmath334 .
the functions @xmath335 satisfy the shuffle relations , and are linearly independent over @xmath321 .
every linear combination of the functions @xmath327 , where @xmath328 , which is single - valued , can be written as a unique linear combination of functions @xmath335 . the functions @xmath335 can be constructed explicitly . to this end
we make for the generating series @xmath336 the ansatz @xcite @xmath337 for a yet to be determined function @xmath338 which is substituted in @xmath339 . for any @xmath340 the monodromy of @xmath341 at zero vanishes : from ( [ mon ] )
we have @xmath342 for the monodromy at one we obtain @xmath343 with ( [ zid ] ) a sufficient condition for trivial monodromy at one is @xmath344 which holds if @xmath345 recall that @xmath346 is the ring of integer mzvs .
[ lielem ] there exists an @xmath347 such that ( [ x1cond ] ) holds .
moreover , @xmath348 modulo words with at least two @xmath349 ( depth two ) or modulo words of length greater than or equal to four .
define the function @xmath350 because @xmath351 , equation ( [ zid ] ) , we have @xmath352\langle\langle x\rangle\rangle$ ] . as a product of three lie exponentials @xmath353
is a lie exponential . because @xmath354 we have @xmath355 .
differentiation yields @xmath356 explicitly , @xmath357,{\mathsf{x}}_1]+\zeta(3)([[[{\mathsf{x}}_0,{\mathsf{x}}_1],{\mathsf{x}}_1],{\mathsf{x}}_1]-[{\mathsf{x}}_0,[[{\mathsf{x}}_0,{\mathsf{x}}_1],{\mathsf{x}}_1]])+\ldots.\ ] ] clearly , @xmath358 modulo depth two and @xmath359,{\mathsf{x}}_1]$ ] modulo weight four .
we recursively solve ( [ x1cond ] ) by defining @xmath360 because @xmath361 has lowest weight three , the recursion converges in the weight filtration .
define @xmath340 as the limit of the recursion
. then @xmath362 fulfills ( [ x1cond ] ) . because by induction every @xmath363 we have @xmath347 .
modulo depth two @xmath364 for all @xmath251 .
modulo weight four the recursion stabilizes after @xmath365 for which we have @xmath366 modulo weight four . the first non - trivial contributions to @xmath340 have four generators .
explicitly , @xmath367,{\mathsf{x}}_1],{\mathsf{x}}_1]-[{\mathsf{x}}_0,[[{\mathsf{x}}_0,{\mathsf{x}}_1],{\mathsf{x}}_1]])+\zeta(5)(\dots)+\ldots,\ ] ] where the @xmath368 contribution consists of eight bracket words in six generators ( see @xcite ) . for later use
we define the following in general multi - valued multiple polylogarithms . [ p0def ] for any word @xmath39 in 0 and 1 let @xmath369 the multiple polylogarithms @xmath370 equal @xmath371 modulo @xmath346 .
their generating series is the untwisted version of equation ( [ pgenseries ] ) @xmath372 up to weight three or depth one the svmps @xmath371 equal @xmath373 .
[ wt2ex ] at weight two we obtain @xmath374 by the shuffle identity @xmath375 so that there exists only one genuinely new svmp of weight two .
one may take @xmath376 as this new function and , by the shuffle identity on @xmath68 we obtain @xmath377 where @xmath378 is the bloch - wigner dilogarithm @xcite .
[ lox0 ] at weight four the following svmps gain a @xmath379-term from ( [ x1 ] ) , @xmath380 @xmath381 [ ex0s ] with ( [ r1ex ] ) we obtain @xmath382 from ( [ x1cond ] ) we obtain for the function @xmath383 the identities @xmath384 the holomorphic and antiholomorphic differentials of the generating series @xmath385 are @xmath386 upon projection @xmath387 ( see ( [ sp ] ) ) onto the word @xmath388 the second identity gives @xmath389 to study the number theoretic content of svmps we now reduce to the ground field @xmath5 . in @xmath371due to contributions from @xmath340multiple polylogarithms of lower weights mix with mzvs .
the total weight of a product of mzvs with holomorphic and antiholomorphic multiple polylogarithms is the sum of the individual weights . [ homcor ] @xmath371 is homogeneous of total weight @xmath69 .
giving the generators @xmath161 , @xmath162 a weight @xmath390 then @xmath275 , @xmath293 , @xmath277 have weight 0 while from the proof of lemma [ lielem ] it follows that the total weight of @xmath340 equals @xmath390 .
hence @xmath385 has total weight 0 . by construction the regularized value of @xmath335 at 0 vanishes .
their regularized values at 1 form a ring which by lemma [ lielem ] is a sub - ring of @xmath346 .
[ rsvdef]for any ring @xmath75 define the ring @xmath391 and let @xmath392 be the @xmath5-algebra of regularized values of svmps at 1 .
we will see in the next section that the regularized values @xmath335 at infinity equivalently span @xmath393 .
moreover , @xmath394 by theorem [ stabilitythm ] .
assuming transcendentality conjectures we can characterize @xmath78 @xcite . as a shuffle algebra @xmath78 is generated as an algebra by mzvs of odd weight on which @xmath395 coacts .
the number of generators at weight @xmath396 equals the number of weight @xmath396 lyndon words in odd generators of weight greater or equal three .
there exists a canonical map @xmath397 .
[ rsvex ] up to weight ten @xmath78 is generated by @xmath379 , @xmath368 , @xmath398 , @xmath399 . at weight 11
we have in addition to @xmath400 the mzv @xmath401 corresponding to the lyndon word 335 .
in fact , for @xmath402 , @xmath403 the ring @xmath115 is spanned up to weight eleven by @xmath404 , @xmath368 , @xmath405 , @xmath406 , @xmath407 , @xmath408 , @xmath409 , @xmath410 , @xmath411 , @xmath412 , @xmath413 , and @xmath414 .
a special property of svmps on @xmath265 is that there exists a group of mbius transformations that permute the singular points .
let @xmath104 be the group of mbius transformations of @xmath101 that permute @xmath415 , @xmath416 the generating series @xmath385 of svmps transforms as follows : the following identities hold @xmath417 @xmath418 we first prove the first set of equations in ( [ s3trafos ] ) .
the first formula for @xmath419 is a consequence of ( [ pgenseries ] ) and ( [ fmapdual ] ) . by the argument in the proof of lemma [ h1 ]
@xmath420 leading to the first formula for @xmath421 .
if we apply the transformation @xmath422 to @xmath421 we obtain the first formula for @xmath423 .
the transformation @xmath248 applied to @xmath419 gives with the above formula for @xmath424 the result for @xmath425 .
another transformation @xmath422 yields the formula for @xmath426 by ( [ fmapdual ] ) .
the second set of equations is equivalent to the first because they fulfill the same differential equations with respect to @xmath79 .
an initial value is easily checked for the transformations @xmath427 and @xmath428 .
the other cases follow by iterating @xmath429 and @xmath430 . because @xmath431 is invertible ( it is a lie exponential ) we obtain from the transformation
@xmath422 @xmath432 by the independence of the multiple polylogarithms in @xmath82 we obtain ( [ x1trafo ] ) .
the third set of equations follows like the second set by antiholomorphic differentiation and the second identity in ( [ x1id ] ) .
if we substitute @xmath278 in @xmath433 we obtain @xmath434 .
the ring of svmps generates at @xmath1 the same mzvs as at 1 .
if we set @xmath278 in @xmath426 we get the following corollary : let @xmath435 be linear expressions in @xmath246 such that @xmath436 .
then @xmath437 fulfills the identity @xmath438 if we substitute @xmath278 into @xmath426 in ( [ s3trafos ] ) we obtain the identity @xmath439 by ( [ zid ] ) we have @xmath440 and ( [ xid ] ) is true for @xmath441 , @xmath442 , @xmath443 . by a change of variables this proves ( [ xid ] ) for linearly independent @xmath235 and @xmath236 . if @xmath235 and @xmath236 are linearly dependent then because @xmath277 is a lie - exponential@xmath444 .
note that in general @xmath445 is not 1 .
it follows from the second identity for @xmath426 that the canonical map @xmath397 maps @xmath445 to 1 .
a consequence of theorem [ francisthm ] is that the integrals @xmath446 and @xmath447 of a svmp of weight @xmath43 are unique in @xmath448 . by ( [ diffonaw ] )
the antiholomorphic integrals @xmath449 and @xmath450 uniquely exist in @xmath451 .
these integrals have the property that their regularized limit at zero vanishes and are henceforth written as @xmath452 . for @xmath453 let @xmath454 , @xmath455 be the unique function @xmath456 such that @xmath457
. likewise @xmath458 is uniquely defined by @xmath459 .
for @xmath460 we define @xmath461 and @xmath462 , where @xmath463 and @xmath464 are regularized limits .
the integral @xmath465 is calculated by appending the letter @xmath235 to the basis svmps @xmath371 .
the antiholomorphic integral is inductively defined by integrating ( [ diffonaw ] ) , @xmath466 where @xmath455 .
note that @xmath467 has at least length 4 so that an inductive algorithm rapidly terminates .
[ lox ] for words @xmath39 with length at most two we have @xmath468 for words
@xmath39 of length three with @xmath469 in ( [ weightfour ] ) antiholomorphic integration picks up a @xmath379 term ( for @xmath470 see definition [ p0def ] ) , @xmath471 equation ( [ intonw ] ) uses @xmath340 to give a formula for antiholomorphic integration . likewise the construction of the basis @xmath371 uses @xmath340 . in practice the definition of @xmath340 via generating functions ( [ x1cond ] )
is only accessible to low weights ( @xmath472 ) .
an alternative approach to integration of mzvs provides holomorphic and antiholomorphic integration without using @xmath340 and enables us to construct the basis @xmath371 to high weights ( @xmath473 ) .
we need the following propositions : [ limprop0 ] let @xmath453 be a svmp
. then @xmath474 where @xmath475 is the kronecker delta . by linearity
we may assume without restriction that @xmath476 for some word @xmath39 .
we use induction over the length of @xmath39 . if @xmath477 then @xmath478 and ( [ limeq0 ] ) follows . for the word @xmath479 , @xmath480
, we obtain from ( [ intonw ] ) by induction @xmath481 [ limprop1 ] let @xmath453 be a svmp
. then @xmath482 substituting @xmath248 in ( [ limeq0 ] ) yields for all @xmath483 , @xmath480 @xmath484 because @xmath485 the proposition follows from a substitution @xmath486 , @xmath487 .
the main tool for calculating holomorphic and antiholomorphic integrals is an explicit formula for their commutator .
[ commutelem ] let @xmath453 and @xmath488 .
the integrals @xmath489 , @xmath490 fulfill the following commutation relation : @xmath491 applying @xmath492 to the left hand side of the above equation leads to the ansatz @xmath493 in the regularized limit @xmath494 the left hand side vanishes , hence @xmath495 . applying @xmath79 to both sides of the above equation yields @xmath496 multiplication by @xmath50 and taking the limit @xmath494 gives @xmath497 where @xmath498 . the first term on the right hand side vanishes .
by proposition [ limprop0 ] we get for the second term on the right hand side @xmath499 . from @xmath500
we obtain @xmath501 .
multiplication of ( [ proof1 ] ) by @xmath322 and taking the limit @xmath502 gives @xmath503 with proposition [ limprop1 ] we have @xmath504 this completes the proof .
[ comm0 ] the integrals @xmath505 and @xmath506 commute .
[ commrem ] the above lemma allows us to construct the integrals @xmath507 and @xmath508 by the following steps .
assume @xmath509 of weight @xmath510 .
then @xmath511 with @xmath512 and @xmath513 the integral @xmath514 is given up to a multiple polylogarithm in @xmath82 by integrating the holomorphic multiple polylogarithms in @xmath515 .
using the above equation and ( [ commute ] ) we obtain @xmath516 now , @xmath517 are of smaller weight than @xmath518 .
we hence may assume that @xmath519 is known by induction for @xmath520 .
the above equation fixes @xmath219 by integration of the antiholomorphic multiple polylogarithms up to a holomorphic multiple polylogarithm .
altogether @xmath219 is given up to a constant which is fixed by the condition that the regularized limit of @xmath219 at @xmath521 vanishes .
antiholomorphic integration is given by the analogous algorithm .
[ indef ] let @xmath116 be the ideal in @xmath6 generated by mzvs of weights between two and @xmath43 .
so ( assuming transcendentality conjectures , otherwise we only have inclusions ) @xmath522 , @xmath523 and @xmath524 . recall the definition of @xmath525 in ( [ p0defeq ] ) .
we have for @xmath488 , @xmath526 moreover , from ( [ sinversion ] ) we have for words @xmath39 of positive length @xmath527 hence @xmath528 because @xmath371 is of total weight @xmath69 and @xmath529 we have @xmath530 [ modintlemma ] let @xmath488 and @xmath39 be a word in 0 and 1
. then @xmath531 where ( see examples [ lox0 ] and [ lox ] ) @xmath532 we prove ( [ modints ] ) by induction over the length of @xmath39 using remark [ commrem ] .
the statement reduces to lemma [ lielem ] and ( [ loworderexamples ] ) of @xmath477 .
taking @xmath79 on both sides the first equation in ( [ modints ] ) reduces to an identity by ( [ pp0 ] ) .
hence the result holds up to an antiholomorphic function .
moreover , we have by induction and ( [ pp0 ] ) @xmath533 using lemma [ commutelem ] we obtain @xmath534 applying @xmath294 to the right hand side we obtain from ( [ pp0 ] ) and the second identity in ( [ diffeq0 ] ) @xmath535 for some analytic function @xmath214 .
the constant in front of @xmath536 reduces to ( [ modintsconst ] ) by ( [ p0w1 ] ) .
altogether the first equation in ( [ modints ] ) holds up to a constant .
the constant is zero by the regularized limit @xmath494 . because @xmath537 the second equation in ( [ modints ] ) follows from complex conjugating the first and swapping @xmath235 and @xmath236 .
note that for any word @xmath39 @xmath538 we obtain a refinement of ( [ pp0 ] ) , see example [ lox0 ] , @xmath539 form the series expansions of the multiple polylogarithms @xmath68 at zero we know that every @xmath540 has a laurent series of the form @xmath541 for @xmath542 , @xmath543 and constants @xmath544 .
the laurent series converges for @xmath545 .
likewise at @xmath278 we have the expansion @xmath546 for @xmath547 , @xmath548 , @xmath549 , and @xmath550 . at @xmath551
we have @xmath552 for @xmath553 , @xmath554 , @xmath555 , and @xmath556 . with the above expansions of @xmath540 the holomorphic and the antiholomorphic residues at @xmath557 , @xmath558 , are @xmath559 for certain @xmath540 the two - dimensional integral over the complex plane @xmath560 is well - defined .
in this case the integral is given by residues .
[ residuethm ] assume @xmath540 such that the integral @xmath560 exists .
let @xmath561 be an antiholomorphic primitive of @xmath214 , @xmath562 .
then @xmath563 firstly , notice that @xmath564 is exact on @xmath265 .
let @xmath565 be the @xmath566 oriented sphere around @xmath235 with radius @xmath567 and @xmath568 .
let @xmath569 be the oriented manifold with boundaries @xmath570 , @xmath571 , and @xmath572 .
then @xmath573 is exact on @xmath569 and by stokes theorem we have @xmath574 using ( [ series0 ] ) , ( [ series1 ] ) , ( [ series8 ] ) , and the parametrization @xmath575 we obtain @xmath576 because @xmath560 exists we have for @xmath577 that @xmath578
if @xmath579 or if @xmath580 and @xmath581 . likewise @xmath582 if @xmath583 or if @xmath580 and @xmath581 . in the limit @xmath240 only the @xmath580 , @xmath584 terms survive .
these terms give ( [ residueeq ] ) .
theorem [ residuethm ] remains valid if we interchange the role of @xmath50 and @xmath82 .
moreover , notice that @xmath585 always exists in @xmath83 .
it is unique up to a function @xmath586 . after a partial fraction decomposition @xmath48 is a linear combination of @xmath587 for @xmath588 and @xmath589 for a negative integer @xmath43 .
if @xmath590 these terms do not contribute to the residues in ( [ residueeq ] ) . if @xmath591 the contributions of these terms cancel in ( [ residueeq ] ) .
this confirms that ( [ residueeq ] ) is independent of the choice of @xmath361 .
the right hand side of ( [ residueeq ] ) is well - defined even if the integral on the left hand side diverges .
the two - dimensional integral of powers of the bloch - wigner dilogarithm @xmath592 exist for @xmath593 . because of the reflection symmetry @xmath594 the integrals of odd powers of @xmath378 vanish .
a computer calculation yields with @xmath595 from example [ rsvex ] @xmath596 note that by theorem [ stabilitythm ] and ( [ bwdilog ] ) the right hand side is in @xmath597 .
the right hand side of ( [ residueeq ] ) gives for @xmath598 the value @xmath599 .
svmps over the ring @xmath115 ( see definition [ rsvdef ] ) are stable under canonical operations .
[ adef ] let @xmath600 be the shuffle ring of svmps over @xmath115 and @xmath601 where @xmath602 $ ] and @xmath603 $ ] , its bi - differential @xmath5-algebra .
let @xmath604 , @xmath605 and @xmath606 denote the subspaces of total weight @xmath43 .
we do not assume that the decomposition into total weight subspaces is direct .
[ stabilitythm ] the @xmath607 module @xmath608 is stable under the @xmath104 group of mbius transformations permuting @xmath415 .
the integrals @xmath609 , @xmath610 , @xmath611 , @xmath612 for @xmath558 map @xmath608 into @xmath613 .
moreover , @xmath614 we have @xmath615 where @xmath616 is the @xmath607 module of integer mzvs of weight @xmath43 .
we have @xmath617 with total weight @xmath390 .
finally , for @xmath618 with @xmath619 , @xmath620 and @xmath621 the @xmath5-algebra @xmath622 is stable under @xmath104 transformations , @xmath79 , @xmath294 , @xmath623 , @xmath624 for @xmath558 .
we have @xmath625 whenever the regularized limits exist . for @xmath626 and @xmath558 , @xmath627
the following statements are obvious : the stability of @xmath628 under the transformation @xmath422 by ( [ s3trafos ] ) and the stability of @xmath628 under @xmath505 , @xmath629 .
@xmath630 is trivial , @xmath631 because @xmath115 is a ring and @xmath632 because @xmath633 by ( [ s3trafos ] ) . from lemma [ lielem ]
we have @xmath347 hence @xmath634 .
the integrals @xmath635 , @xmath636 , @xmath455 , differ from @xmath465 by a value in @xmath637 , or @xmath638 , respectively .
the first equation in ( [ diffstab ] ) is also clear .
the group @xmath104 is generated by @xmath422 and @xmath248 .
the stability of @xmath628 under @xmath248 is proved by induction over the total weight .
the statement is trivial for weight zero .
assume @xmath509 .
then from @xmath639 we have @xmath640 and @xmath641 by induction .
next , we show by induction that @xmath394 .
consider @xmath371 for a word @xmath39 of length @xmath43 .
we start with the observation from ( [ diffonaw ] ) that for any word @xmath467 , @xmath642 . by stability of @xmath628 under @xmath248
we obtain @xmath643 on the other hand from ( [ diffonaw ] ) we obtain @xmath644 this proves ( [ x1primeinlie ] ) by induction over the length of @xmath39 .
it was already proved in corollary [ homcor ] that @xmath162 is of total weight @xmath390 ( the weight of @xmath162 ) . from ( [ diffonaw ] )
we obtain the second equation in ( [ diffstab ] ) .
it follows by induction and ( [ intonw ] ) that the antiholomorphic integrals @xmath611 , @xmath612 map from @xmath608 into @xmath613 .
the stability of the residue ( [ resstab ] ) follows for @xmath645 by expanding all multiple polylogarithms @xmath63 and @xmath646 in @xmath214 up to the linear term : by ( [ explicitlw ] ) this expansion does not generate denominators . at @xmath647
the result follows from @xmath645 by @xmath648 . at @xmath649
the result follows from @xmath645 by @xmath650 .
the result for @xmath651 follows by complex conjugation . to prove ( [ 2dintstab ] ) we first prove using integration by parts and induction over the weight that @xmath652 for @xmath653 and @xmath455 maps from @xmath628 into @xmath654
. then we use the residue theorem [ residuethm ] . repeating the argument that lead to ( [ resstab ] ) the residues in @xmath50 of @xmath655 are in @xmath115 . taking the tensor product over @xmath607 with @xmath5 , using integration by parts , and ( [ explicitlw ] ) for ( [ asveq1 ] ) and ( [ asveq2 ] ) yield the statements for @xmath622 .
in this subsection we define graphical functions in @xmath656 dimensions .
although we are mainly interested in four dimensions , graphical functions exist in any dimensions greater than or equal to two .
two dimensions , however , are a special case which is postponed to [ 2d ] . the position space propagator of an edge @xmath19 with vertices @xmath164 and @xmath165 is @xmath657 the power in the denominator originates from fourier transforming a @xmath658 momentum space propagator . we first define uncompleted graphical functions and turn to completion in [ completion ] .
[ fgdef ] the graphical function @xmath659 is defined by applying position space feynman rules to the graph @xmath48 with distinguished vertices @xmath49 , namely @xmath660 where the first product is over all vertices @xmath661 and the second product is over all edges of @xmath48 .
the vertex 0 corresponds to the origin in @xmath662 whereas @xmath663 stands for any unit vector @xmath664 and @xmath665 . by rotational symmetry @xmath659 only depends on @xmath53 and the angle between @xmath50 and @xmath16 .
if we identify @xmath16 with @xmath666 we can consider @xmath659 as a function on the complex plane with the symmetry @xmath667 under complex conjugation . in the following
we always do so unless the argument of @xmath668 is explicitly specified as a @xmath669-dimensional vector .
we can calculate the graphical functions of a complex argument @xmath50 by the integral ( [ fdefnoncomp ] ) with the identification @xmath670 , @xmath671 in four dimensions ( @xmath672 ) we often drop the superscript @xmath673 . in a massless @xmath0 theory
the four - point function can in principle be reconstructed from a completed graphical function ( see [ completion ] ) obtained by labeling the four external vertices with @xmath95 ( see @xcite for a tree - level example ) .
[ exi ] our main examples are sequential graphs ( see figure 4 ) .
the initial case is @xmath674^\lambda}.\ ] ] sequentially appending edges gives for any word @xmath39 that starts with 2 the graphical functions @xmath675 as a @xmath676-dimensional integral . in particular ,
@xmath677 which can be calculated with gegenbauer techniques ( see example [ onevertex ] ) . [ internaldef ] for a subgraph @xmath678 of a graph @xmath48 a vertex @xmath26 is called internal if @xmath679 and all edges adjacent to @xmath26 in @xmath48 are in @xmath678 .
all other vertices are called external . an edge in @xmath678
is internal if both vertices are internal otherwise it is external .
[ finitelem ] the integral in definition [ fdefnoncomp ] is convergent if and only if it is infrared and ultraviolet finite . the integral
@xmath668 is infrared finite if and only if for every subgraph @xmath678 of @xmath48 with @xmath680 internal vertices and @xmath681 edges , @xmath682 the integral @xmath668 is ultraviolet finite if and only if for every subgraph @xmath678 of @xmath48 with @xmath681 and with at most one of its @xmath683 vertices in @xmath684 one has @xmath685 in @xcite it is shown that a feynman integral is finite if the integral is finite by power - counting for every subgraph @xmath678 of @xmath48 .
if @xmath686 convergence is trivial .
otherwise there are two sources for divergences : infrared , when the integral does not converge for large values of the integration variables ; ultraviolet , when the coincidence of integration variables generates a singularity that can not be integrated over . infrared convergence by power - counting is equivalent to the condition that the sum of degrees of the denominators is larger than the dimension of the integral .
this gives ( [ infrared ] ) .
if @xmath678 is minimal with an ultraviolet divergence by power - counting then the integral diverges when all vertices in @xmath678 coincide in the integral of @xmath687 .
because @xmath684 are distinct @xmath678 has at most one labeled vertex .
the locus where all integration variables coincide is @xmath688 dimensional .
power - counting gives ( [ ultraviolet ] ) .
let @xmath39 begin with 2 . to show infrared finiteness we observe that every internal vertex has at least valency three .
counting half - edges gives @xmath693 moreover , every internal vertex is connected to at least one external vertex 0 or 1 , @xmath694 the inequality is strict because either @xmath695 or the leftmost vertex in @xmath696 has at least two external edges .
together we obtain @xmath697 ultraviolet convergence is trivial if neither 0 nor 1 is in @xmath678 because in this case @xmath698 .
otherwise we can assume without restriction that @xmath699 .
let @xmath26 be the valency of 0 in @xmath678 .
every other vertex in @xmath678 has at most valency three , hence by counting half - edges @xmath700 the inequality is strict because the leftmost vertex in @xmath678 has valency at most two .
moreover , @xmath701 because every edge in @xmath678 which is connected to 0 is adjacent to a vertex in @xmath702 .
together we obtain @xmath703 to prove ( g2 ) we first consider the graphical function as a function of @xmath707 .
we add a small @xmath708 to the quadrics in the denominator of the propagators so that the singular locus of the integrand does not intersect the chain of integration .
if we vary the argument @xmath50 along a closed smooth path in @xmath662 the graphical function changes smoothly until it is back to its initial value .
transition to @xmath54 by the decomposition @xmath709 yields ( g2 ) in the limit @xmath240 .
a consequence of completion ( see [ completion ] ) is that we may generalize ( g3 ) .
the graphical function @xmath668 is real analytic in @xmath55 if after completion the graph becomes series parallel with the removal of two labeled vertices .
most conveniently gegenbauer polynomials are defined by the following generating series @xcite , @xcite , @xcite , @xmath715 with ( [ seriesdef ] ) we can expand a position space propagator in gegenbauer polynomials .
let @xmath716 be the angle between @xmath164 and @xmath165 and let @xmath717 .
then @xmath718 the main property of gegenbauer polynomials is their orthogonality with respect to @xmath669-dimensional angular integrations .
if we denote the integral over the @xmath719-dimensional unit - sphere in @xmath669 dimensions by @xmath720 , normalized by @xmath721 , we have @xmath722 [ onevertex ] the sequential function @xmath723 exists for all @xmath724 .
it can be calculated with gegenbauer techniques .
series expansion of @xmath725 and @xmath726 with ( [ ortho ] ) plus an elementary radial integration gives for @xmath727 ( one may specify @xmath728 , @xmath729 , @xmath730 , @xmath731 in ( [ gegenbauerint1 ] ) ) @xmath732 and for @xmath733 @xmath734 the result for @xmath735 can be deduced from @xmath727 by the inversion formula @xmath736 which follows from the defining integral by a coordinate transformation @xmath737 .
now we set @xmath738 and @xmath739 . in the case
@xmath672 we use the identity @xmath740 which we obtain from the generating series .
this relates @xmath741 to the dilogarithm , @xmath742\ln|z|).\ ] ] the combination of logarithms and dilogarithms in the bracket is @xmath743 times the bloch - wigner dilogarithm ( [ bwdilog ] ) .
we obtain @xmath744 by the symmetry of @xmath378 under inversion @xmath745 the above equation holds also for @xmath556 .
we will re - derive equation ( [ 4id ] ) using svmps in example [ f2example ] . in the case
@xmath733 we obtain from the generating series of the gegenbauer polynomials by elementary integration for @xmath738 and @xmath267 , @xmath746 this allows us to express @xmath723 as an integral , @xmath747 for integer @xmath748 the above integral has vanishing residues in @xmath749 . for @xmath750
we obtain @xmath751 the simplicity of the result stems from the fact that in @xmath752 dimensions the three - valent vertex is unique in the sense of @xcite . for @xmath753
we have @xmath754 ^ 2}.\ ] ] for general even dimensions greater than four we obtain a rational expression in @xmath50 and @xmath82 with singularities in @xmath521 and @xmath278 . it is trivially single - valued and extends unchanged to @xmath556 . for odd dimensions one also obtains a logarithm - free result .
it is a rational expression in @xmath755 with singularities at @xmath521 and @xmath278 which extends unchanged to @xmath556 .
the expression is explicitly single - valued .
assume without restriction that @xmath769 .
the gegenbauer polynomials @xmath770 are polynomials in @xmath771 and hence real analytic for @xmath772 .
the norm @xmath53 is real analytic for @xmath772 and so is @xmath773 . due to
( [ conditionab ] ) the sum in ( [ eqle1 ] ) is absolutely convergent for @xmath727 and the sum in ( [ eqgr1 ] ) is absolutely convergent for @xmath735 .
hence the analytic expansions commute with the sums and the claim follows .
we substitute @xmath782 in the integral ( [ fdefnoncomp ] ) defining @xmath783 .
the integration measure changes by a constant . after extracting all factors @xmath784 from the propagators we find that a propagator from @xmath785 to 0 is changed to a propagator from @xmath785 to @xmath786 and the propagator from @xmath787 ( say ) to @xmath50
is changed to a propagator from @xmath787 to @xmath788 . after a substitution @xmath789
the latter becomes a propagator from @xmath787 to @xmath165 .
all other propagators remain intact .
because the propagator from @xmath790 to @xmath786 is in @xmath791 ( example [ expluse1 ] ) we can use induction over the length of the word @xmath39 to prove the result with proposition [ prop2 ] by propositions [ prop1 ] and [ prop3 ] @xmath714 is real analytic for all @xmath50 except 0 and the unit sphere . by propositions [ prop1 ] and [ prop4 ] @xmath792 is real analytic for all @xmath165 except 0 and the unit sphere in @xmath165 .
hence @xmath714 is real analytic except for @xmath793 and the vectors @xmath50 with @xmath794 which is a sphere of radius @xmath784 around @xmath795 .
the intersection of the two loci of possible non - analycities are 0 and @xmath16 .
transition to @xmath54 proves the lemma . one can generalize the proof to corollary [ realanalyticcor ] to all graphs which can be generated by successively using proposition [ prop2 ] .
the condition ( [ alphacondition ] ) on @xmath796 is equivalent to the convergence of the integral defining the graphical function .
completion adds a vertex with label @xmath1 .
we weight the edges of the graph by real exponents .
positive integer weights may be graphically represented by multiple lines . for the weight @xmath390 which lifts the edge - quadric into the numerator we choose to draw a curly line .
integer negative weights may be represented by multiple curly lines ( see figure 5 ) .
we will see that integer weights suffice to complete graphs in three and four dimensions .
a graph @xmath23 is completed in @xmath797 dimensions if it has the labels @xmath95 and weights such that every unlabeled ( internal ) vertex has weighted valency @xmath798 and every labeled ( external ) vertex has valency @xmath799 the graphical function of @xmath23 is @xmath800 if @xmath801 equals @xmath48 up to an edge from 0 to 1 of any weight then @xmath802 ( definition [ fgdef ] ) . in this case @xmath23 is the completion of @xmath48 .
[ completionlemma ] let @xmath48 be a graph with three labeled vertices @xmath49 .
then @xmath48 has a unique completion . in dimensions three and four
the completed graph has integer edge - weights if @xmath48 has integer edge - weights .
let @xmath26 be an internal vertex of @xmath48 with weighted edge - valency @xmath804 .
completion uniquely connects @xmath26 to @xmath1 by an edge of weight @xmath805 . clearly , in three or four dimensions @xmath806 if @xmath807 .
let the vertex @xmath50 in @xmath48 have weighted edge - valency @xmath808 .
then completion connects @xmath50 to @xmath1 by an edge of weight @xmath809 which is integer if @xmath808 is . with these edges we partially complete @xmath48 to @xmath810 in which the vertices @xmath100 have weighted valency @xmath811 , respectively . in @xmath23
we have to add to @xmath810 a weighted triangle @xmath100 with the edge - weights @xmath812 this completes the graph . a change in one edge - weight of a triangle
can not be compensated by a change of the others .
so the completion is unique .
finally , counting weighted half - edges in @xmath810 gives @xmath813 where @xmath814 is the number of internal vertices in @xmath810 and @xmath815 is the weighted sum of edges in @xmath810 . in three and four dimensions we have @xmath816 if @xmath817 which gives the integrality of @xmath818 , @xmath819 , and @xmath820 .
a completed graph @xmath23 has a convergent graphical function ( [ fgamma ] ) if and only if every subgraph @xmath197 of @xmath23 with at most one of its @xmath821 vertices in @xmath822 has @xmath823 where @xmath824 is the weighted edge sum of @xmath197 . for subgraphs @xmath197 with @xmath825 condition ( [ ultravioletcompleted ] )
is equivalent to ( [ ultraviolet ] ) for subgraphs in @xmath826 .
it remains to show that for @xmath197 with @xmath827 condition ( [ ultravioletcompleted ] ) is equivalent to ( [ infrared ] ) .
let @xmath678 be a subgraph of @xmath48 . to check infrared finiteness
we can assume without restriction that every external edge of @xmath678 connects to exactly one internal vertex .
from @xmath678 we construct a graph @xmath197 by connecting all internal vertices to @xmath1 such that their weighted valency becomes @xmath828 and then cutting all other external edges .
now , @xmath197 is a subgraph of @xmath23 which connects to @xmath1 but not to @xmath49 .
conversely deleting @xmath1 in @xmath197 and adding external edges such that all vertices become internal leads back to @xmath678 .
we have to show that ( [ ultravioletcompleted ] ) for @xmath197 is equivalent to ( [ infrared ] ) for @xmath678 .
we have @xmath829 is the weighted sum of edges that connect @xmath197 with @xmath1 plus the number of internal edges in @xmath678 .
moreover , by completion we have @xmath830 where the sum is over internal vertices @xmath26 of @xmath678 and @xmath804 is the valency of @xmath26 in @xmath678 . because @xmath831 we have that ( [ ultravioletcompleted ] ) is equivalent to @xmath832.\ ] ] counting half - edges in @xmath678
gives @xmath833 implying ( [ infrared ] ) .
reversing the arguments shows that ( [ infrared ] ) implies ( [ ultravioletcompleted ] ) .
let @xmath102 be the symmetric group that permutes the four external vertices @xmath95 .
it is generated by the transpositions @xmath834 , @xmath835 , @xmath836 .
the symmetric group @xmath102 has a factor group @xmath104 which in our context is realized as mbius transformations on @xmath265 .
the homomorphism @xmath837 is determined by the images of the generating transpositions .
concretely we have @xmath838=z\mapsto\frac{1}{z},\quad\phi[(0,1)]=z\mapsto 1-z,\quad\phi[(1,z)]=z\mapsto\frac{1}{z}.\ ] ] let @xmath839 be a permutation of the external labels and @xmath840 be @xmath430 followed by a transformation of the label @xmath50 ( see figure 9 ) indicating that the transformed graphical function has the argument @xmath841 . because the @xmath102 is generated by the three transpositions in ( [ transpos ] ) it is sufficient to prove the identities depicted in figure 9
let @xmath847 , @xmath848 denote the internal vertices of @xmath23 and for @xmath849 let @xmath850 be the sum of the weights of all the edges from internal vertices to @xmath26 .
the weighted sum of internal edges is @xmath851 . counting weights of half - edges that connect to internal vertices we obtain @xmath852 for @xmath853 let @xmath854 be the weight of the edge connecting @xmath26 with @xmath39 . in particular , @xmath855 if there exists no edge from @xmath26 to @xmath39 . indexed by the labels @xmath95 the integral in ( [ fgamma ] ) is of the shape ( working in @xmath669 dimensions ) @xmath856 where @xmath857 has no edges that connect external vertices . a variable transformation @xmath858 leaves the integration measure invariant .
propagators between internal vertices remain intact , whereas a propagator from an internal vertex @xmath847 to an external vertex @xmath26 is mapped to a propagator from @xmath847 to @xmath859 .
hence @xmath860 and by swapping the labels 0 and 1 in ( [ ff0 ] ) we have @xmath861 this proves the middle identity in figure 9 . a variable transformation @xmath862 changes the integration measure by @xmath863 ( as can be seen e.g. in angular coordinates ) .
propagators between internal vertices change according to @xmath864 where @xmath865 is the weight of the edge between @xmath847 and @xmath866 .
similarly internal edges that connect to @xmath49 transform as latexmath:[\ ] ] 99 , _ orthogonal polynomials and special functions _ , regional conference series in applied mathematics , vol . 21 ,
siam , philadelphia , pa ( 1975 ) .
, _ on motives associated to graph polynomials _ ,
267 , 181 - 225 ( 2006 ) .
, _ the multiple zeta value data mine _ , comput .
181:582 - 625 ( 2010 ) . ,
open university report out-4102 - 18 ( 1985 ) . , _ evaluation of a class of feynman diagrams for all numbers of loops and dimensions _ , phys .
b164 , 356 - 360 ( 1985 ) . , _
multiple zeta values and other periods in quantum field theory _ , conference talk , bristol , 4 may 2011 . ,
_ knots and numbers in @xmath0 theory to 7 loops and beyond _ , int .
c6 , 519 - 524 ( 1995 ) . , _ single - valued multiple polylogarithms in one variable _ , c.r .
paris , ser .
i 338 , 527 - 532 ( 2004 ) . , _ single - valued hyperlogarithms and unipotent differential equations _ , unpublished notes ( 2004 ) . , _ on the periods of some feynman integrals _ , arxiv:0910.0114v2 [ math.ag ] , ( 2009 ) . , _ on the decomposition of multiple zeta values _ , arxiv:1102.1310v2 [ math.nt ] ( 2011 ) . , _
mixed tate motives over @xmath607 _ , annals of mathematics 175 , no . 2 , 949 - 976 ( 2012 ) . , compositio . math . 146 , no . 3 , 731 - 771 ( 2010 ) . , _
framings for graph hypersurfaces _ , arxiv:1301.3056 [ math.ag ] ( 2013 ) . , _ a k3 in @xmath0 _ ,
duke mathematical journal , vol .
161 , no . 10 ( 2012 ) .
, _ multiple zeta values and single - valued multiple polylogarithms _ , in preparation . , _ modular forms in quantum field theory _ , in preparation . , _ proof of the zig - zag conjecture _ , arxiv:1208.1890 [ math.nt ] ( 2012 ) . , _ three - mass triangle integrals and single - valued polylogarithms _ , arxiv:1209.2722 [ hep - ph ] ( 2012 ) . , _ algebras of iterated path integrals and fundamental groups _ , transactions of the american mathematical society , vol . 156 , 359 - 379 ( 1971 ) . , _
new approach to evaluation of multiloop feynman integrals : the gegenbauer polynomial @xmath164-space technique _ , nucl .
b174 345 - 377 ( 1980 ) .
, _ integration by parts : the algorithm to calculate @xmath1250-functions in 4 loops _ , nucl .
b192 , 159 - 204 ( 1981 ) . ,
_ generalised ladders and single - valued polylogarithms _ , talk given at d. broadhurst s birthday conference , berlin , 27 june 2012 arxiv:1207.3824 [ hep - th ] ( 2012 ) .
, _ galois symmetries of fundamental groupoids and noncommutative geometry _ , duke math .
j. volume 128 , no . 2 , 209 - 284 ( 2005 ) . ,
_ the method of uniqueness , a new powerful technique for multiloop calculations _ , phys .
b133 no . 6 , 406 ( 1983 ) . , _ current algebras and wess - zumino models in two dimensions _ , nucl
. phys .
b247 , 83 - 103 ( 1984 ) . ,
_ mmoires sur la thorie des systmes des quations diffrentielles linaires _ , chelsea , new york ( 1953 ) . , _
another proof of zagier s evaluation formula of the multiple zeta values @xmath1251 @xmath1252 _ , arxiv:1204.2060 [ math.nt ] ( 2012 ) .
, _ calculation of the @xmath0 6-loop non - zeta transcendental _ , arxiv : hep - th/9912149 ( 1999 ) .
, _ quantum periods : a census of @xmath0-transcendentals ( version 1 ) _ , arxiv : 0801.2856v1 [ hep - th ] ( 2008 ) . , _ quantum periods : a census of @xmath0 transcendentals _ , jour .
theory and phys . 4 no . 1 , 1 - 48 ( 2010 ) . , _ the geometry of one - loop amplitudes _ , arxiv:1010:5334 [ hep - th ] ( 2010 ) . ,
zeta_procedures , http://www.mathematik.hu-berlin.de/@xmath1253kreimer/index.php ?
section@xmath1254program&lang@xmath1254en ( 2012 ) . ,
polylog_procedures , http://www.mathematik.hu-berlin.de/@xmath1253kreimer/index .
php?section@xmath1254program&lang@xmath1254en ( 2013 ) . ,
_ calculation of multiloop diagrams in high orders of perturbation theory _ ,
b267 , 382 ( 1991 ) . , _ exact results for three- and four - point ladder diagrams with an arbitrary number of rungs _ , phys .
b305 , 136 - 143 ( 1993 ) .
, _ special functions and the theory of group representations _ , providence , ri , american mathematical society ( 1968 ) . , _ high - energy behavior in quantum field theory _ , phys
. rev .
118 , no . 3 838 - 849
, _ the remarkable dilogarithm _ , in : number theory and related topics .
papers presented at the ramanujan colloquium , bombay 1988 , tifr and oxford university press , pp .
231 - 249 ( 1989 ) and j. math .
22 131 - 145 ( 1988 ) .
, _ evaluation of the multiple zeta values @xmath1255 _ , annals of mathematics , vol .
175 , no . 1 , 977 - 1000 ( 2012 ) . | graphical functions are single - valued complex functions which arise from feynman amplitudes .
we study their properties and use their connection to multiple polylogarithms to calculate feynman periods . for the zig - zag and two more families of @xmath0 periods we give exact results modulo products .
these periods are proved to be expressible as integer linear combinations of single - valued multiple polylogarithms evaluated at one . for the larger family of constructible graphs we give an algorithm that allows one to calculate their periods by computer algebra .
the theory of graphical functions is used in @xcite to prove the zig - zag conjecture . |
the concept to realize and investigate quantum processes with single ions in paul traps was subject to an impressive development over the last years .
as for quantum information with single to a few ions , there has hardly been an experimental challenge that could withstand this development , consider for instance high fidelity quantum gates @xcite , multi - qubit entanglement @xcite , and error correction @xcite .
there has been a recent development in ion trap technology that was primarily initiated by the necessity and possibility to scale up the established techniques to process larger numbers of ions .
this idea how to scale up an ion trap computer is to create multiple trap potentials , each keeping only a small , processable number of ions @xcite .
time dependent potentials can then be used to shuttle the ions between different sections . to generate such spatially separated potentials ,
the respective trap electrodes are segmented along the transport direction and supplied individually with time dependent voltages .
there have also been proposals to incorporate the shuttling process into the quantum gate time by letting the ion cross laser beams or magnetic field arrangements @xcite .
another application of the shuttling ability is the exact positioning of the ion in the mode volume of an optical cavity to perform cavity quantum electrodynamic experiments .
recently , multi - trap configurations with locally controlled , individual trap frequencies were proposed to generate cluster states in linear ion traps to perform one - way quantum computations @xcite .
this paper presents the implementation of a method using a single trapped ion to investigate local electric fields .
the method profits from the outstanding accuracy of spectroscopic frequency measurements and relies on the ability to shuttle the ion in order to reach different probing sites .
first , we describe the experimental system and the way the electric trapping and shuttling potentials are generated . after that
, the measurement scheme using the ion as a local field probe is presented .
then , the quantitative experimental results are discussed and compared to numerical simulations of the trapping fields .
we confine a single , laser cooled @xmath0-ion in a linear paul trap .
the radial confinement is generated by a harmonic radio - frequency pseudo potential resulting from a @xmath1 drive with an amplitude of about @xmath2 .
this results in radial trap frequencies around @xmath3 . for the axial confinement ,
32 pairs of electrodes are available .
each of these 64 segments can be individually biased to an voltage @xmath4 , @xmath5 . in the trap region
the experiments were performed in , the segments are @xmath6 wide and separated by @xmath7 gaps ( details can be found in @xcite ) .
the voltage data preprocessed by a computer and transmitted to a home - built electronics device
. this device houses an array of 16 serial input digital - to - analog converters ( dac , @xmath8 maximum clock frequency ) , each of which supplying four individual voltage lines , so that a data bus conveying 16 bit in parallel suffices to obtain the desired 64 outputs .
the amplitude resolution of the converters is 16 bit with a measured noise level well below the last significant bit .
the serial input design of the converters keeps the amount of data being transmitted in parallel at a minimum .
this makes the design extendable to a large number of channels . in this work ,
we restrict ourselves to the investigation of harmonic potentials .
this is rectified by the fact that a resting , laser cooled ion only experiences the very minimum of the potential .
such potentials are generated by applying control voltages to the trap segments .
thereby , each specific voltage configuration results in an axial trap with a given trap frequency ; the position of the potential minimum this is where the ion resides can be changed to shuttle the ion between different locations on the trap axis .
doing so , two requirements have to be met by the time dependent voltages : ( i ) the amplitudes must range within the limits given by the dac electronics , @xmath9 in our case .
( ii ) each segment s voltage is to change as continuous as possible to avoid unfeasibly high frequencies .
all these requirements are fulfilled by the field simulation and voltage calculation techniques presented in sections [ sec : app_fieldcalc ] and [ sec : app_regul ] , respectively . for the experiments presented here ,
it is advantageous to perform slow , adiabatic potential changes , and thus avoiding excess oscillations of the ion .
for that reason , the update rate of digital to voltage converters was chosen with @xmath10 , and a parallel port connection between computer and dac device was used . in an alternative operation modus of the trap supply device
, the data transmission and updated rate is significantly sped up by using a field programmable gate array . then we reach @xmath11 even when updating all 64 channels simultaneously . for the application presented here , it is sufficient to use a doppler cooled probe ion .
the probe s spatial extent can then be estimated from the ion temperature and the consequential extent of the motional wave function to be @xmath12 .
this value could be reduced to @xmath13 by sub - doppler cooling and a tighter confinement of the ion .
to test both the numerical methods and the developed electronic devices for accuracy , we employ a single ion as a probe for the electrostatic potential .
we exploit the fact that it is possible to exactly measure the trap frequency of the confining axial potential by spectroscopic means .
consider , for example , a voltage configuration @xmath14 to be tested for ; this configuration is meant to result in a trap at position @xmath15 with frequency @xmath16 .
then we can use time dependent potentials to shuttle the ion from a loading position to @xmath15 @xcite , whereby the final voltage configuration must be @xmath14 .
the spectroscopically obtained trap frequency at @xmath15 can then be compared to the theoretically expected one , @xmath17 .
the trap frequency is determined by measuring the difference frequency of the carrier and the first motional red sideband ( rsb ) excitation of the ion in a resolved sideband regime . for that
, we excite the quadrupole transition @xmath18 at @xmath19 . as the carrier resonance frequency
does not depend on the trap frequency , it is sufficient to measure the rsb frequency at different locations @xmath15 .
the measurement scheme is illustrated in figure [ fig : spectroscopy_scheme ] .
it consists of the following steps : \(i ) an initial voltage configuration is chosen to trap and cool the ion at the starting position @xmath20 .
all lasers necessary for cooling , repumping , state preparation and detection are aligned to interact with the ion here . additionally , @xmath20 is the position where fluorescence emitted by the ion can be detected by a photomultiplier tube and a camera .
\(ii ) the ion is shuttled to the probing position @xmath15 .
this is accomplished by applying voltage configurations resulting in a series of harmonic potentials whose minimum positions lead the ion from @xmath20 to @xmath15 . for simplicity ,
these potentials were chosen such that their harmonic frequencies were almost constant and close to @xmath16 .
\(iii ) applying a spectroscopy pulse at the probing position @xmath15 .
now , the voltages are exactly @xmath14 . resting at @xmath15 ,
the ion is exposed to a spectroscopy pulse of fixed duration ( @xmath21 ) .
the pulse is detuned from the ( carrier ) resonance by @xmath22 .
this excites the ion with a probability @xmath23 into the upper state @xmath24 .
\(iv ) shuttle the ion back to @xmath20 , inverting step ( ii ) .
\(v ) having arrived back at the starting position , the state of the ion is read out by illuminating it on the cooling transition .
whenever we detect a fluorescence level above a certain threshold , the ion is found in the ground state @xmath25 , while a low fluorescence level indicates that the ion has been excited to the state @xmath24 .
the excitation probability for a specific detuning , @xmath26 , at the remote position @xmath15 is obtained by averaging over many repetitions of steps ( i ) to ( v ) . by varying @xmath22 , a spectrum of the quadrupole excitation at the remote position
is obtained without moving any lasers or imaging optics but the spectroscopy laser used in step ( iii ) .
this is a significant advantage , because that way even such trap positions can be investigated that can not be directly observed by the imaging devices or reached by all lasers .
the frequency difference between the red sideband and the carrier transition yields the trap frequency .
figure [ fig : spectra ] shows two rsb resonance peaks obtained at different positions within the trap volume .
it is noticeable that each iteration cycle ( i - v ) including two ion transports results due to the binary nature of the projective readout in exactly one bit information about the spectrum .
therefore , thousands of transports , each relying on the calculated potentials , are performed for the determination of one frequency @xmath16 .
the transport , however , can be performed so fast ( @xmath27 ) that its contribution to the overall experiment duration is secondary ; this duration is still dominated by cooling and detection times ( @xmath28 ) .
to implement the measurement scheme , we first calculated voltage sets @xmath29 , where @xmath30 covers the whole extent of the trap in steps of @xmath31 ( for arbitrary positions , the voltages can be interpolated ) .
each set results in a certain , wanted frequency @xmath32 .
that means that for each arbitrary position @xmath30 in the trap , there can be found a set of voltages resulting in a potential with its minimum at @xmath30 and with trap frequency @xmath32 . then , in order to shuttle the ion , we simply subsequently apply the voltage configurations for @xmath33 .
the calculated voltages are tested with high axial resolution , i.e. in small steps of @xmath15 , in two far distant regions of the trap .
doing this , both small local deviations should be detectable and the stability over the whole trap structure can be tested for . to see a variation in @xmath16
when increasing @xmath15 , it is of advantage that small variations of the trap frequency around its means value occur .
this is a reliable way to ensure that the ion in fact probes the remote position .
figure [ fig : remote_results ] shows the expected , simulated trap frequencies and the measured ones .
the data are in excellent agreement with the predicted frequencies . on both ends of the investigated trap structure ,
the predicted course of @xmath32 is confirmed within the spectroscopic accuracy of @xmath34 .
the mean deviation of all measured data points is only @xmath35 .
note that the solid line shown in figure [ fig : remote_results ] is based solely on geometric data from a technical drawing of the trap ; there is no free parameter that was used to match the simulations with the measurement .
the high accordance between the simulated and the real trap frequencies , as can be observed in the presented experiments , indicates the reliability of at least four independent contributions : first , the numerical field simulation is accurate .
this simulation is independent from a specific set of voltages , but relies on a model of the three dimensional trap design .
second , the physical realization of this geometry is very good , i.e. the manufacturing and assembling process of the micro trap is so precise that it is not limiting the field accuracy .
this aspect is coming more an more into the experimental focus , since the miniaturization of the trap layouts still proceeds .
for example , a possible axial misalignment @xmath36 of the two dc electrode wings with respect to each other would mainly result in an axial shift of the oscillations depicted in fig .
[ fig : remote_results]a by approximately @xmath37
. additionally , the amplitude of these oscillations decreases with increasing @xmath36 .
indeed , the residuals ( fig .
[ fig : remote_results]b ) show a small oscillatory behavior .
this oscillation signal is shifted by at most @xmath38 with respect to the trap frequency oscillation .
such an effect could stem from a small misalignment of the two trap layers ( see ref .
@xcite ) or from a small offset of the theoretical and the real , axial zero point of position .
we attribute the observed shift to an offset error , since in this case , the amplitude of the oscillation is not decreased , as it is the case in the measured data from microscope pictures of the trap . ] .
this shift is systematic and could be corrected for .
third , the calculation of the voltages leading to the wanted potentials is highly reliable .
this is also qualitatively confirmed by the successful shuttling process itself . from our measurement
we find that the generation of the voltages and their supply to the trap electrodes is indeed highly accurate .
we also conclude that trapping fields from the applied voltages are not significantly perturbed by background fields from stray charges . as the generation of trapping fields in the multi - segment trap
was found to be accurate over the whole considered volume , it is possible to tailor potentials for special purposes . for quantum computing tasks , a harmonic potential of constant frequency is mostly required .
simulations show that it is possible to obtain fixed frequency potentials with a relative frequency deviation on the order of @xmath39 .
the fast transport of quantum information in harmonic wells also requires a high degree of control over the trap voltages @xcite .
the application of optimal control methods can help optimizing tailored time dependent potentials .
alternatively , one can feed - back information , gained from the measurement presented here , into the generation process to refine the results iteratively @xcite .
however , the scope of possible applications of the described techniques is much wider than quantum computing tasks : ions in time dependent potentials were proposed to be used as a testbed for quantum thermodynamic processes @xcite or for quantum simulation @xcite .
the presented method is just a first proof of principle of using single atoms as ultra - sensitive probes .
the measurement principle is not restricted to the investigation of the electric trapping fields .
only slight modifications allow for the detection of any electric or magnetic field with a nano - meter scaled quantum probe , as an alternative to cold atom sensors @xcite .
then , a single atom field probe might be used as well for probing magnetic micro - structures with a relative accuracy better than 10@xmath40 . also , a study of decoherence and heating effects and their dependence on the ion - electrode separation or ion location is in reach . financial support from the dfg within the sfb / trr-21 , the german - israel science foundation and by the european commission within microtrap , emali and scala is acknowledged .
to create a specific electric potential @xmath41 on the trap axis , it is necessary to find the right set of voltages @xmath42 being applied to the trap electrodes labeled by @xmath43 .
the potential generated by such a set of voltages is the linear superposition of the @xmath44 individual electrodes , whereby the contribution of each electrode @xmath45 is weighted by the applied voltage @xmath4 . after subdividing the axial position into a grid of @xmath46 points @xmath47 ,
we can write the overall potential at any @xmath48 by @xmath49 here we introduced the electrode potential matrix @xmath50 .
it describes the influence of the @xmath45-th electrode to the overall potential at @xmath51 . each row @xmath45 of @xmath52
can be seen as a position - dependent function , describing the potential generated by electrode @xmath45 ( in units of @xmath4 ) , when all other electrodes are set to zero voltage .
this quantity @xmath52 is independent from the specific voltage and is solely given by the trap geometry , i.e. the shape and size of the electrode and its distance from @xmath48 , for instance .
therewith , the potential generation can be logically divided into two parts : the matrix @xmath52 can be calculated independently from voltage constraints and independent from the desired potential .
second , for each desired potential @xmath53 , there has to be found a set of voltages @xmath54 by inverting the matrix equation above . tackling the first problem
, one recognizes that modern segmented trap geometries can be realized in such a geometric complexity that conventional simulation techniques like the finite element method ( fem ) fail .
instead , we obtained the potentials by solving the boundary element problem of the segmented trap design .
details can be found in @xcite .
in the following , we address the problem how to obtain a set of voltages @xmath42 that generates a given potential @xmath55 when applied to the respective electrodes .
the problem is formally solved by matrix inversion as @xmath56 .
several circumstances make this straight forward ansatz unfeasible : first , there is often no exact solution to the problem , because @xmath53 is not an exactly realizable potential ( note that in general , @xmath57 is possible ) . in this case ,
an approximate solution has to be found .
additionally , as a specific electrode does not effectively contribute to the potential at a far distant point , this electrode s voltage is ill - determined .
these cases have to be treated adequately by the algorithm .
we solved the inversion problem with a singular - value decomposition of the matrix @xmath52 to identify its critical , singular values .
the real @xmath58 matrix @xmath52 is decomposed into the product @xmath59 of the unitary matrices @xmath60 ( @xmath61 ) and @xmath62 ( @xmath63 ) and the diagonal @xmath58 matrix @xmath64 with non - negative entries @xmath65 .
this decomposition is part of many standard numerical libraries and can be performed for any input matrix @xmath52 .
the inverse can then be written as @xmath66 this step is numerically trivial , because the unitary matrices are simply transposed and the entries of @xmath67 are given by @xmath68 . at this point ,
the advantage of the decomposition becomes obvious , since small values of @xmath69 indicate an ( almost ) singular , critical value .
one way to overcome these singular values would be to cut off their diverging inverse values .
instead , the tikhonov regularization @xcite method implies a more steady behavior as it makes the displacement @xmath70 .
the latter expression behaves like the original @xmath68 for large values @xmath71 , has its maximum at @xmath72 and tends to zero for small , critical values @xmath73 . from this
can be seen that the choice of the optimization parameter @xmath74 is a compromise between exactness and boundedness of the results . for @xmath75 , the exact solution ( if existent ) is obtained , whereas large values of @xmath74 guarantee small inverse values and thus bounded voltage results .
therefore , we label the regularized quantities with index @xmath74 . the approximate solution @xmath76 is @xmath77 with @xmath78 being the regularized matrix with entries @xmath79 .
before the problem of finding an optimal @xmath74 is addressed , another constraint regarding time dependent voltages , i.e. series of voltage configurations , has to be accounted for .
while moving the ion by one step , i.e. from a position @xmath15 to @xmath80 , the control voltage should vary as less as possible .
this is achieved by extending eqn .
( [ eq : phi1 ] ) , @xmath81 where @xmath82 represents any previous voltage set , providing trapping at @xmath15 . the second term in eqn .
( [ eq : phi2 ] ) contains a diagonal matrix @xmath83 with entries @xmath84 .
@xmath85 tends to zero for @xmath86 , so that uncritical voltages are affected only little by the second term .
for all critical voltages indicated by a value @xmath87 , however , the first term in eqn .
( [ eq : phi2 ] ) vanishes due to the regularization replacement and what remains is the contribution from @xmath88 , since then , @xmath89 .
here , the choice of @xmath74 determines , how strong the algorithm tries to generate similar voltages in a ( time ) series of voltage sets .
the algorithm described above minimizes @xmath90 with respect to the euclidian norm .
that is , the potential @xmath55 is reproduced as good as possible under the constraint that solutions similar to the previous one are preferred .
what remains is to find the proper value of @xmath74 .
hereby , one has to make a tradeoff between the boundedness of the voltages and their continuity . under practical circumstances requiring @xmath91 for some maximal voltage @xmath92 , @xmath74 can be iteratively increased to fulfill this constraint on the one hand , and to obtain as continuous voltage sets as possible , on the other hand .
j. benhelm et al . ,
nature physics * 4 * , 463 ( 2008 ) .
h. hffner et al . ,
nature * 438 * , 643 ( 2005 ) .
d. leibfried et al .
, nature * 438 * , 639 ( 2005 ) .
j. chiaverini et al . ,
nature * 432 * , 602 ( 2004 ) .
d. leibfried et al , phys .
a * 76 * , 032324 ( 2007 ) .
d. kielpinski , c. monroe and d. j. wineland , nature * 417 * , 709 ( 2002 ) .
s. schulz et al . , fortschr . phys . * 54 * , no .
8 - 10 , 648 ( 2006 ) .
l. greengard and v. rokhlin , in : vortex methods , 121 , ( springer , berlin 1988 ) .
k. nabors et al . , sci .
* 16 * , 713 ( 1994 ) .
j. eble et al . , accepted for publication in j. opt .
b ( 2010 ) , arxiv:0912.2527
. k. singer et al .
, arxiv:0912.0196 .
s. schulz et al . , new j. phys .
* 10 * , 045007 ( 2008 ) .
g. huber et al . , new j. phys . * 10 * , 013004 ( 2008 ) . g. huber et al . , phys .
* 101 * , 070403 ( 2008 ) .
r schtzhold et al .
lett . * 99 * , 201301 ( 2007 ) .
m. johanning et al . , j. phys .
b : at . mol .
opt . phys . * 42 * , 154009 ( 2009 ) .
m. murphy et al . , phys . rev .
a * 79 * , 020301(r ) ( 2009 ) .
h. wunderlich et .
al . , phys .
a * 79 * , 052324 ( 2009 ) . c. pozrikidis , a practical guide to boundary element methods with the software library bemlib , chapman and hall / crc , boca raton , fl , usa , ( 2002 ) .
a. n. tikhonov and v. a. arsenin , solution of ill- posed problems , winston and sons , washington , usa ( 1977 ) . | we introduce a measurement scheme that utilizes a single ion as a local field probe . the ion is confined in a segmented paul trap and shuttled around to reach different probing sites . by the use of a single atom probe , it becomes possible characterizing fields with spatial resolution of a few nm within an extensive region of millimeters .
we demonstrate the scheme by accurately investigating the electric fields providing the confinement for the ion . for this
we present all theoretical and practical methods necessary to generate these potentials .
we find sub - percent agreement between measured and calculated electric field values . |
type ia supernovae ( sne ia ) are thought to be the thermonuclear explosion of a white dwarf .
the light curves of sne ia are powered by deposition in the sn ejecta of the @xmath1-ray and positron products of the @xmath2 decay @xcite . the extreme brightness and seemingly uniform light curves of sne
ia make them good candidates for use as standard candle distance indicators . in more recent years , it has been shown that type ia supernovae do not have uniform light curve magnitudes , shape or spectra .
the light curves can , however , be normalized to account for this inhomogeneity , thus allowing these objects to be used at standard candle distance indicators ( e.g. * ? ? ?
* ; * ? ? ?
* ) . between 100 - 200 days
after the explosion the ejecta become transparent to the @xmath1-rays and the light curve is powered by the deposition of the positron kinetic energy into the ejecta .
the escape of a fraction of these positrons from the ejecta has been suggested as a possible source of the galactic 511 kev annihilation radiation @xcite .
there are currently two methods of modeling the late emission of sne ia .
one is radiation transport with complete and instantaneous trapping of the positrons .
@xcite showed , by comparing a model to the late time spectra of sn 1972e , that the ejecta will cool leading to an increased fraction of the emission coming out in the infrared , the so named infrared catastrophe " ( irc ) .
other studies of radiation transport , have reproduced the irc , but they also predict the abrupt fall off of the optical light curves as the emission shifts into the nir and ultimately into the ir , which is not seen in observed light curves .
the other method consists of positron energy deposition modeling without radiation transport . in this type of modeling ,
optical band light curves are used as tracers of the bolometric luminosity and fit to model energy deposition curves .
the results show model curves with varying degrees of positron escape fitting the light curves .
one weakness in this model fitting technique is in using the optical bands as tracers of bolometric .
@xcite constructed bolometric curves using bvri bands and showed those curves roughly fitting the positron escape energy deposition curves .
we preformed aperture photometry on six normal " sne ia at late epochs , sn 2000e , sn 2000ce , sn 2000cx , sn 2001c , sn 2001bg , sn 2001dp .
some of these sne were located in very complicated regions in their host galaxies .
for this reason , we chose to do template image subtraction , on all but sn 2000cx , before preforming the aperture photometry . all data reduction , image subtraction and aperture photometry was performed using the image reduction and analysis facility ( iraf ) software .
the combined light curves can be seen in figures [ radmodel ] & [ posmodel ] , where they are normalized to be zero magnitude at 200days past explosion assuming an 18d rise time to peak light .
the data set for sn 2000e includes photometry from @xcite , and the data set for sn 2000cx includes data from @xcite , @xcite , @xcite , and , where the data from our observations are plotted as the filled symbols .
the decline rates , the slope of the light curve , between 200 - 500 days were calculated for these light curves and the averages are shown in figure [ slopes ] , where the solid line is average for the six sne .
the calculated averages for b , v , r , & i bands were 1.43 ( 0.07 ) , 1.46 ( 0.04 ) , 1.36(0.04 ) , 0.95 ( 0.06 ) , respectively , in magnitudes per day . the shaded bar represents the average decline rate for 16 normal / super - luminous sne ia from @xcite with a @xmath3 error bar . in the r - band , there is a second average , represented by the dot - dashed line , which is the average decline rate leaving out sn 2000ce and sn 2001c .
this was done only to show the agreement with the milne et al
. averages . as shown in figure [ slopes ] ,
the b , v , & r bands have decline rates of @xmath4 mag / day but the i - band has a much shallower slope of 0.95 mag / day .
this is in agreement with the decline rates of sn 2000cx as shown by .
these results suggest that a slower i - band decline rate is a general feature of the late light curves of normal / super - luminous sne ia , and is possibly suggesting a shift in the late emission to longer wavelengths .
a major result of was the constant late time emission seen in the nir curves of sn 2000cx , which supports the idea that the emission is moving into the nir and eventually into the ir resulting in an irc .
our results from the analysis of these sne reinforce the need for more observations of sne ia in the nir in an attempt to reproduce what was seen in sn 2000cx and also in sn 1998bu ]
figure [ radmodel ] shows the combined light curves of the six sne plotted on the radiation transport models of .
the v - band model light curve has been normalized to be zero magnitude at 200d along with the data .
the b , r , & i band model light curves have been adjusted so that the colors of the model are preserved .
the dotted curve is the model including photoionization and the dot - dashed curve is the model without photoionization .
figure [ posmodel ] shows the combined light curves of the six sne plotted on the positron energy deposition curves of @xcite .
the model curves have been normalized to be zero magnitude at 200d .
the solid curve is the energy deposition with the positron kinetic energy trapped and deposited into the ejecta and the dashed curve is the energy deposition curve with a radially combed magnetic field allowing a fraction of the positrons to escape the ejecta without depositing their kinetic energy .
as can be seen in these figures , the shape of the b , v , & r band light curves could be explained by either the color evolution in the radiation transport model or the escape of positrons from the ejecta , while the i - band has a slower decline rate than both models .
this suggests that a model combining radiation transport with positron transport would be preferred to attempt to explain the late light curves of sne ia .
one thing is clear from the light curves ; these sne show very little deviation from each other in a given band , implying that within this class of normally - luminous sne ia there is only one answer for the question of positron escape from sne ia ejecta . , p. , krisciunas , k. , suntzeff , n. b. , gonz ' alez , d. , espinoza , j. , leiton , r. , rest , a. , smith , r. c. , cuadra , j. , tavenner , t. , logan , c. , snider , k. , thomas , m. , west , a. a. , gonz ' alez , g. , gonz ' alez , s. , phillips , m. m. , hastings , n. c. , mcmillan , r. , mar .
optical and infrared photometry of the unusual type ia supernova 2000cx .
pasp 115 , 277294 . , w. , filippenko , a. v. , gates , e. , chornock , r. , gal - yam , a. , ofek , e. o. , leonard , d. c. , modjaz , m. , rich , r. m. , riess , a. g. , treffers , r. r. , oct
the unique type ia supernova 2000cx in ngc 524 .
pasp 113 , 11781204 . , j. , lindahl , j. , kozma , c. , challis , p. , filippenko , a. v. , fransson , c. , garnavich , p. m. , leibundgut , b. , li , w. , lundqvist , p. , milne , p. , spyromilio , j. , kirshner , r. p. , dec .
the late - time light curve of the type ia supernova 2000cx .
a & a 428 , 555568 . , g. , di carlo , e. , massi , f. , dolci , m. , arkharov , a. a. , larionov , v. m. , pastorello , a. , di paola , a. , benetti , s. , cappellaro , e. , turatto , m. , pedichini , f. , dalessio , f. , caratti o garatti , a. , li causi , g. , speziali , r. , danziger , i. j. , tornamb , a. , oct .
optical and near - infrared photometry of the type ia supernova 2000e in ngc 6951 .
apj 595 , 779793 . | the use of type ia supernovae as cosmological tools has reinforced the need to better understand these objects and their light curves .
the light curves of type ia supernovae are powered by the nuclear decay of @xmath0 . the late time light curves can provide insight into the behavior of the decay products and their effect of the shape of the curves . we present the optical light curves of six normal " type ia supernovae , obtained at late times with template image subtraction , and the fits of these light curves to supernova energy deposition models . |
the neutron lifetime @xmath3 is an important parameter in nuclear physics , particle physics , and cosmology .
@xmath3 can be combined with the neutron @xmath4-decay ( @xmath5 ) correlation coefficients to determine the universal weak interaction vector and axial - vector coupling constants whose values allow searches for semi - leptonic scalar and tensor currents beyond the standard model @xcite .
a @xmath3 of reliable precision is also needed for calculations of the neutrino flux expected from solar and reactor sources , including detection efficiencies @xcite , as well as in big bang nucleosynthesis calculations . at present ,
we are confronted by an apparent discrepancy of about three standard deviations between the average @xmath3 from ultracold neutron ( ucn ) storage experiments and the @xmath3 from cold neutron beam experiments .
it is the leading source of uncertainty in predictions of the primordial abundance of @xmath6he @xcite . for reviews of @xmath3 experiments
see @xcite .
one of the most promising methods to achieve higher precision for the storage - type experiments employs ucn storage in magnetic bottles . in suitable non - uniform magnetic field configurations the neutrons in one spin state
should , in principle , experience no losses other than @xmath4-decay , provided that depolarization , i.e. the non - adiabatic transition to the non - storable spin state , is sufficiently suppressed . in most existing and proposed magnetic trap systems the low - field seeking state with spin parallel to the magnetic field is being stored and a spin - flip transition to the anti - parallel state results in a loss by escape from the bottle or by interaction with the bottle walls . until recently ,
ucn depolarization estimates @xcite have been based on majorana s quasi - classical result @xcite for a particle with spin moving with constant velocity vector through an infinitely extended non - uniform magnetic field of specific form . for field parameters as currently used for magnetic ucn storage the probability @xmath7 of a spin flip away from the field direction would be of order @xmath8 , thus immeasurably small .
walstrom @xmath0 @xcite , in 2009 , pointed out that the values of @xmath7 for confined , rather than freely moving , neutrons are much larger . for a particular vertical path in the field of the los alamos gravito - magnetic ucn trap they calculated @xmath9 , which is much larger than the majorana estimate but still negligible in any actual or projected neutron lifetime experiment .
in @xcite we extended this theory to general orbits with both vertical and horizontal velocity components , using the model of an ideal halbach magnetic field @xmath10 where the magnitude @xmath11 only depends on the vertical position .
we found that the lateral motion in the plane where the halbach field rotates is of critical importance .
taking it into account increases the spin - flip loss thus calculated by some 10 orders of magnitude to @xmath12 for a field minimum ( holding field ) of @xmath13 mt .
this translates into a spin - flip loss rate that is a fraction @xmath14 of the @xmath4-decay rate and decreasing rapidly with larger holding field .
the analysis in @xcite is based on the following concepts : for the one - dimensional ( 1d ) field model of @xcite , the potential @xmath15 for the high - field repelled @xmath16 spin state of a neutron with mass @xmath17 and negative magnetic moment @xmath18 nev / t depends only on the vertical @xmath19 coordinate . in this model
the neutrons are exposed to a uniform gravitational field @xmath20 and a non - uniform magnetic field of magnitude @xmath21 .
they perform an oscillatory motion with turning points ( tp ) at the lower and upper horizontal equipotential surfaces ( es ) where @xmath22 and the potential is @xmath23 . here @xmath24 is the neutron energy ; @xmath25 and @xmath26 are the vertical and horizontal velocity components , respectively .
@xmath27 is constant for the 1d field model . as the particle moves from one tp to the next , starting out in a pure @xmath16 spin state , its wave component for the @xmath28 spin state increases .
it may change over many orders of magnitude @xcite , peaking at critical points where the field magnitude @xmath11 is small and the vector @xmath10 rotates rapidly in the reference frame of the moving particle .
the spin flip probability is `` measured '' only at the next tp where , in the copenhagen interpretation , the wave function collapses and ucns in the @xmath16 state return to the trapping region while the @xmath28 projection separates in space and quickly becomes lost .
conceptually , the `` measurement '' could be made by an ideal neutron detector placed just next to the tp , which would intersect the ucns in the `` wrong '' spin state as they exit the storage space .
this measurement resets the ucn wave function to a pure initial @xmath16 state for the next lap where the sequence of wave evolution and collapse at the following tp is repeated .
in the present article we extend this approach to the analysis of depolarization in cylindrical multipole fields such as those described in refs .
@xcite , where the trapping fields are generated by halbach arrays of permanent magnets @xcite or , for @xcite and , earlier @xcite , by superconducting currents . for these cylindrical configurations we use a 2d field model which enables us to obtain a semi - analytic expression for the ensemble - averaged spin - flip loss and which can be analyzed with no need to involve simulations .
the results are consistent with the only experimental spin - flip probabilities with varying holding field available so far @xcite . for the cylindrical @xmath29-pole we approximate the field in cylindrical coordinates @xmath30 ,
@xmath31 , @xmath32 as follows : @xmath33 .
@xmath32 points along the cylinder axis and the holding field @xmath34 is considered constant .
@xmath35 is the trapping field magnitude at the wall and the radius @xmath36 ( typically @xmath37 cm ) is much smaller than the length , which is of order @xmath38 m. this justifies the neglect of gravity for horizontal configurations of this type @xcite since the gravitational energy varies little over the trap radius . to assess the merits of model ( [ 1 ] ) in general
we have performed 3d simulations including gravity both for the vertical and the horizontal cylindrical multipole configurations . as a second application
we extend the previous analysis @xcite of depolarization for a 1d field model of the los alamos ucn@xmath2 trap @xcite to the actual field in this magneto - gravitational trap with its asymmetrically double - curved wall in the shape of a bowl . as in @xcite
we approximate the field for the curved arrays of permanent magnets by that of the corresponding infinite planar array tangent to the bowl surface at the closest point on the bowl surface .
we also use the same expressions for the flat - wall field , dubbed `` smooth '' [ eq .
( 5 ) of @xcite ] , `` one - way ripple '' [ eq . ( 7 ) ] and `` two - way ripple '' [ eq .
the `` one - way ripple '' takes into account the finite magnet size and the `` two - way ripple '' also includes the effect of iron shims between the magnets , a design feature not implemented for the current ucn@xmath2 system ( status of 2016 ) .
the theoretical approach is outlined in sec .
[ sec : iii ] , where we derive a first - order approximation to the spin - flip probability from the spin - dependent schrdinger equation ( se ) , and in sec .
[ sec : iv ] where we average these results over the ensemble of orbits in the field configurations of ucn@xmath2 and of multipole bottles . in sec .
[ sec : v ] we derive a higher - order solution of the spin - dependent se and show that it deviates very little from the first - order approximation .
these various approaches are semi - classical since the field @xmath39 acting on the neutron spin is determined by the classical motion of the particle through the field .
however , the results have been shown @xcite to be consistent also with a fully quantum mechanical analysis starting from the spin and space dependent se .
we show in the appendix that this equivalence also holds for our extension to arbitrary field configurations . as in @xcite
, we use the wentzel - kramers - brillouin ( wkb ) approximation @xcite to solve the se .
this is justified since the spatial variation of field variables is much slower than the variation of the ucn wave function .
the scales are of order cm for gravity and @xmath10 , and of order @xmath40 m or less for the wavelength .
thus the wave function for spin state @xmath16 can be expressed in the wkb form except at a tp @xmath41 , where its amplitude @xmath42 diverges since the wave number @xmath43 vanishes .
( see eq .
( [ a12 ] ) for details . )
neutron lifetime experiments based on magnetic storage require that the spin follows the changes of field direction along the neutron path for a time much longer than the neutron lifetime , implying that the probability @xmath44 for spin @xmath16 , parallel to the local field , is always much larger than the small spin - flipped part @xmath45 . therefore , in order to separate large terms in the se from the small ones it is advantageous to use a reference system which rotates with the field experienced by the moving particle
. thus we use the se for spin 1/2 with quantization axis in the local field direction , as in @xcite : @xmath46 where @xmath47 are the spinors aligned in the direction of or opposite to the local magnetic field @xmath10 , respectively . for
the los alamos `` bowl '' we choose the @xmath48 direction ( defined as the direction of the toroidal holding field @xmath49 measured in the vertical symmetry plane ) as the fixed quantization axis relative to which the local field direction is given by the polar angle @xmath50 and the azimuthal angle @xmath51 . in ( [ 3 ] ) we have used @xmath52 , @xmath53 and @xmath54 . differentiating @xmath55 and @xmath56 we get @xcite @xmath57 with time dependent coefficients @xmath58 @xmath59 and @xmath60 are small quantities to be treated as small perturbations .
using eqs .
( [ 3])-([5 ] ) in ( [ 2 ] ) we obtain for the terms with @xmath61 in first order @xmath62 and for those with @xmath63 @xmath64 where @xmath65 is the local larmor frequency . equations ( [ 6 ] ) and ( [ 7 ] ) correspond to eqs . (
57 ) and ( 58 ) of @xcite
. the solutions of ( [ 6 ] ) and ( [ 7 ] ) are @xcite , in wkb approximation , @xmath66 where @xmath67 is twice the phase angle accumulated since the previous tp . since @xmath68 and @xmath4 have the same phase , @xmath69 , the wave components @xmath68 and @xmath4 propagate between tps as a unit ( they do not run apart ) .
this feature had also been noted for the 1d field model @xcite . from ( [ 9 ] ) , the probability of finding the neutron in the spin - flipped state along the way to the next tp is @xmath70 @xmath71 is the frequency of field rotation about an axis normal to the plane defined by @xmath10 and @xmath72 and can be expressed as @xmath73 .
this form holds since @xmath74 is the squared projection of vector @xmath72 , as drawn from the tip of vector @xmath10 , onto the unit sphere . for any trajectory
we determine the arrival time at , and the position of tps from the condition that the velocity component along the gradient of potential @xmath75 vanishes , @xmath76 , since , at a tp , the orbit is tangential to the es . equations ( [ 9 ] ) and ( [ 10 ] ) make use of an approximation which we will discuss in sec . [
sec : v ] in connection with the higher - order solution given in ( [ 25 ] ) and ( [ 27 ] ) . in short ,
this analysis shows the following features of @xmath77 : starting from zero at a tp , @xmath77 increases to the value given in ( [ 10 ] ) within a short time of order @xmath40s .
( [ 10 ] ) holds over the entire remainder , typically @xmath78 s , of the motion to the next tp where @xmath77 is `` measured '' .
equation ( [ 10 ] ) shows that the result depends only on the local field variables @xmath71 and @xmath79 ; it is independent of the path history . for the sequence of tps encountered along a neutron path of total duration @xmath80
we determine the spin - flip rate between consecutive tps at @xmath81 and @xmath82 by dividing @xmath83 , from ( [ 10 ] ) , by the time interval @xmath84 . taking into account the probability @xmath85 of finding the particle on this path element
the spin - flip rate for the entire path becomes @xmath86 where @xmath87 is the number of tps encountered . finally , the depolarization rate measured in the experiments is the ensemble average over all paths , which is determined by the source characteristics and by spectral cleaning .
we assume an isotropic maxwell spectrum , thus an energy independent phase space density ( psd ) .
the boltzmann factor @xmath88 is close to unity since ucn energies @xmath24 , which are of order @xmath89 ev , are much lower than @xmath90 even for a low trap temperature @xmath91 .
the corresponding velocity dependence of the spectrum is @xmath92 .
relation ( [ 10 ] ) is also obtained in a fully quantum mechanical approach using the space and spin dependent se in wkb approximation @xcite .
this equivalence holds for any field geometry as shown in appendix @xmath93 .
in actual magnetic ucn storage systems it is difficult to make sure that the spectrum is isotropic and fills phase space uniformly up to the trapping limit . in practice
this would require the complete removal of ucns with energies slightly exceeding the limit .
these tend to be in quasi - stable orbits lasting for times of the order of @xmath3 , thus affecting the precision of a measurement of @xmath3 .
deviations from the maxwell spectrum may also be due to the characteristics of the ucn source and of ucn transport to the trap but we will disregard these differences since they are expected to be of minor importance for the specific loss due to spin flip . for the 1d field model of ref .
@xcite the condition of constant psd was taken into account as follows : the spin - flip probability was averaged over the vertical velocity component @xmath94 in the plane @xmath95 where the gravitational downward force , @xmath96 , is balanced by the magnetic upward force , @xmath97 .
@xmath98 is the 1d equivalent of an elliptic fixed point @xmath99 and @xmath95 is the only plane where ucns of any energy @xmath24 can reside , down to @xmath100 and up to the maximum value for trapping .
( here and henceforth we set the potential @xmath101 at @xmath99 and assume that there are no other potential minima in the trapping region ; this is the case for the field configurations presently used or proposed . ) for any other height , the lower energy limit is non - zero . therefore , to include all possible orbits in the averaging process we have to choose @xmath98 as the reference height , and
since the statistical distribution of @xmath94 values is uniform for uniform psd , the mean depolarization rate is given as the average of @xmath102 as a function of @xmath94 . to analyze the depolarization between consecutive tps we start trajectories from a tp , not from the fixed point @xmath99
; so we have to connect the statistics at @xmath98 , which is given by a uniform distribution of @xmath94 , with the distribution @xmath103 of launching height @xmath19 for given @xmath94 .
@xmath103 follows from energy conservation : the potential at the launching point is @xmath104 , thus @xmath105 .
therefore , to represent constant spacing in velocity space ( @xmath106 const . ) for orbits launched at height @xmath19 we have to choose the density @xmath107 as @xmath108 now we adapt this result to arbitrary 3d field models as are relevant for the los alamos bowl @xcite .
the asymmetry introduced by the choice of two different radii of curvature along the rows of halbach field magnets helps to randomize the orbits although fully mixing phase flow can not be achieved . to select a representative sample of orbits for depolarization calculations
we assume that most particles with energy @xmath24 below the trapping limit will , at some time , be found at rest .
( in practice it suffices to require @xmath109 m / s . ) in this case the particle has just reached the es @xmath110 constant .
it will then be accelerated back into the region of lower potential in the direction perpendicular to the es .
taking these points as the initial position for simulated trajectories starting from rest we make sure that the particle remains within the volume bounded by the es with potential @xmath111 ( as long as no spin flip to the @xmath28 state takes place ) .
the statistical distribution of launch points in space is determined by the following extension of ( [ 12 ] ) to 3d geometry : we relate the launching point to the fixed point @xmath99 in the same way as for the 1d field model . equating the initial energy @xmath112 with the kinetic energy at @xmath99 , @xmath113 , we differentiate @xmath114 .
this gives @xmath115 where @xmath116 is the coordinate perpendicular to the es @xmath117 const . at the launching point
furthermore , the potential in the immediate vicinity of @xmath99 is constant , thus the spatial density is stationary .
hence , to satisfy uniformity of psd the density in velocity space must be uniform : @xmath118 const . , and we have to choose the spatial density of launching points according to @xmath119 we take the last form of ( [ 13 ] ) as the ( not normalized ) probability distribution @xmath120 ) representing the number of launch points per unit volume at position ( @xmath121 ) . in the simulations we use
von neumann s acceptance or rejection method to implement this probability distribution . for this purpose
we need the fraction @xmath122 where @xmath123 is the maximum value of @xmath124 for the ensemble of trajectories .
we use this method to select random launching points for simulated orbits in ucn@xmath2 and use the three field models for the halbach array , `` smooth '' , `` one - way ripple '' and `` two - way ripple '' and the toroidal - shaped holding field coil geometry described in @xcite . averaging the spin - flip rate ( [ 11 ] ) over a sample of some @xmath125 orbits ,
each of duration @xmath126 s with @xmath127 tps , for four values of holding field @xmath128 at the bowl bottom , we obtain the depolarization rate @xmath129 shown in table [ table : one ] .
the results are consistent , within a factor of two , with those given in fig . 3 of @xcite for the 1d field model : @xmath130 s@xmath131 for @xmath132 mt and @xmath133 s@xmath131 for @xmath134 mt . fig .
[ fig : one ] shows as squares the 1d calculation of @xcite and as circles the present 3d calculations for the smooth - field model . in the range
@xmath135 mt the data are represented reasonably well by the proportionality @xmath136 as indicated by the dashed line .
.mean depolarization rate @xmath137 @xmath138 $ ] for the los alamos ucn@xmath2 field [ cols="^,^,^,^ " , ] [ table : two ] we approximate the effect of spectral cleaning in ucn storage experiments by specifying the largest energy @xmath139 of stored particles .
the calculations of @xmath129 shown in table [ table : two ] use @xmath140 times the value @xmath141 for the highest field @xmath142 in the trap and are based on the field parameters of the following two designs : \(a ) the hope octupole magnet @xcite has bore radius @xmath143 cm and its axis oriented vertically or horizontally . for the vertical configuration
we assume activation of only the bottom solenoid with a maximum axial field of @xmath144 t while gravity provides the cap . with both end field solenoids
activated in the horizontal configuration we assume fields of @xmath144 t on both ends , separated by a distance of @xmath145 m , without activating the long holding field solenoid .
the radial confinement field is @xmath146 t at the trap wall .
\(b ) two horizontal ioffe type quadrupole magnets have been used at nist @xcite ( versions mark 2 and mark 3 ) . for mark 2 ( mark 3 )
we use maximal fields of , axially : @xmath144 t ( @xmath147 t ) , and radially : @xmath148 t ( @xmath149 t ) , a bore radius of @xmath150 cm ( @xmath150 cm ) and a separation of @xmath151 m ( @xmath152 m ) between the centers of the end field solenoids ; for mark 2 , the latter include `` bucking '' coils @xcite causing the axial field to drop off more quickly to a minimum of @xmath153 t at the trap center . for mark 3 , the minimum field is @xmath154 t. since the depolarization rates depend only weakly on the details of the field distribution , such as field ripples , we use the smoothed fields and approximate the solenoid fields by their values on the solenoid axis , neglecting the variations of field magnitude and direction over the bore cross section .
the results are shown in table [ table : two ] . for the hope - type system
we include , in square brackets , also systems with the same geometries and maximum fields but different multipole order @xmath29 .
we observe the same tendency as for the 2d calculations of fig .
[ fig : two ] : the depolarization rates for @xmath155 and @xmath156 are higher than in the intermediate range ( @xmath157 ) .
this can be explained by higher field gradients , near the axis for low @xmath158 and near the wall for high @xmath158 . as a further general feature ,
the magnitudes and @xmath158 dependence of @xmath137 from the 3d simulations in table [ table : two ] are well approximated by the 2d results from fig .
[ fig : two ] if we use values of holding field close to their minima : @xmath159 t/@xmath153 t for hope ( vertical configuration with @xmath160 t / horizontal with @xmath161 t ) @xcite and @xmath153 t ( @xmath154 t ) for nist mark 2 ( mark 3 ) with @xmath162 t ( @xmath154 t ) @xcite .
finally , @xmath137 approximately scales like @xmath163 .
a similar increase of @xmath129 with decreasing holding field is also seen for ucn@xmath2 as shown by the dashed line in fig .
[ fig : one ] . as an application of scaling in an experiment
, we could deliberately lower the holding field to enhance the depolarization loss to a measurable level to verify that the loss for the actual field is negligible .
in this section we compare the first - order approximation for the spin - dependent se , eqs .
( [ 6 ] ) , ( [ 7 ] ) , with a higher - order approach where we retain all the terms with @xmath164 and @xmath60 [ given in ( [ 5 ] ) ] , which arose from the transformation to the reference system rotating with the field : @xmath165 the coupled first - order odes ( [ 21 ] ) and ( [ 22 ] ) can be solved by direct numerical integration with initial conditions @xmath166 , @xmath167 for a particle starting in the @xmath16 state at @xmath168 .
alternatively , we can use the perturbation approach developed in @xcite for searches for a permanent electric dipole ( edm ) of the neutron , to solve the se for spin @xmath169 up to second order of small perturbations . in the edm case the ucn spin state
is perturbed by magnetic field inhomogeneities and a strong static electric field . in eqs .
( [ 21 ] ) and ( [ 22 ] ) the perturbations are the terms on the rhs , which are much smaller than those on the left . to facilitate comparison with @xcite
we define @xmath170 ( real - valued ) , @xmath171 ( complex ) , @xmath172 and @xmath173 . in practical cases , @xmath174 is at least @xmath175 times smaller than @xmath176 ; thus @xmath177 is very close to @xmath176 .
the transformations @xmath178 , @xmath179 turn eqs .
( [ 21 ] ) , ( [ 22 ] ) into @xmath180 these coupled odes for @xmath181 and @xmath182 have the same form as eqs .
( 7 ) for @xmath183 and @xmath184 in @xcite .
the only difference is the arguments of the exponential functions . in @xcite ,
the se was transformed into the reference frame rotating at constant frequency @xmath185 for constant applied larmor field . in the present case
, @xmath186 can be an arbitrary function of @xmath187 ; thus the phase factor @xmath188 is replaced by @xmath189 .
we combine the two first - order odes ( [ 23 ] ) into the single second - order ode for @xmath190 : @xmath191 which has the same form as eq .
( 8) of @xcite with @xmath185 in the first term of the expression in brackets replaced by @xmath186 .
this additional time dependence does not affect the method of solving ( [ 24 ] ) since the second term is time - dependent in either case .
the initial conditions , @xmath192 and @xmath193 , are the same as for the edm case with initial spin up ( @xmath194 , @xmath195 ) .
following the steps ( 11 ) to ( 18 ) of @xcite we derive @xmath196 where @xmath197 .
the initial value , @xmath193 at @xmath168 , satisfies the required initial condition @xmath167 .
as @xmath187 increases , the integral in ( [ 25 ] ) rapidly increases on a time scale of order @xmath198 .
in practical application this is a very short time since the larmor frequency @xmath176 is large everywhere inside the trap volume , even at the field minimum where @xmath11 is the holding field . for @xmath199 t , @xmath200 @xmath40s . for times
@xmath201 it is advantageous to change the integration variable in ( [ 25 ] ) from @xmath202 to @xmath203 , with @xmath204 , and to integrate by parts : @xmath205^t_{0}-\frac{1}{2}\int_{0}^{t}\,dt^{\prime}\,e^{-i\theta_{1}(t^{\prime})}\frac{d}{dt^{\prime}}\left[\frac{\sigma(t^{\prime})}{\omega_{1}(t^{\prime } ) } \right].\end{aligned}\ ] ] we did not employ the wkb approximation to derive eq .
( [ 26 ] ) but its use enables us to evaluate it analytically : we can neglect the last term in ( [ 26 ] ) since the field variable @xmath206 varies slowly on the wavelength scale and get , with ( [ 5 ] ) , @xmath207 in ( [ 27 ] ) we have set the integration constant from the lower limit @xmath168 of the first term in ( [ 26 ] ) equal to zero , and an equivalent approximation had also been made in deriving the first - order solution ( [ 9 ] ) which differs from ( [ 27 ] ) only by the replacement of @xmath177 by @xmath176 . the detailed justification in @xcite , below eq . ( 28 ) , can be summarized as follows
. equations ( [ 9 ] ) and ( [ 27 ] ) are semi - classical since the se is solved for the time dependent field @xmath39 determined from the classical equations of motion .
a fully quantum mechanical treatment requires the solution of the spin and space dependent se as in @xcite and in appendix a. in this quantum analysis the exact solution for the wave function near a tp involves the airy functions and the tp is blurred into a non - zero region ( typically of order @xmath40 m ) .
so is the starting time at a tp .
thus , in ( [ 26 ] ) the initial value of @xmath208 at @xmath209 , which is @xmath210 , should be averaged over the rapidly varying phase @xmath211 with the result @xmath212 .
this holds except for the initial few micrometers of the path subsequent to a tp .
outside of this region we have , from ( [ 27 ] ) and ( [ 5 ] ) , @xmath213 ) agrees with the first - order solution ( [ 10 ] ) if we replace @xmath176 by @xmath177 .
as we have seen , in practical cases the difference between @xmath176 and @xmath177 is negligible . for ucns released from rest at position @xmath214 m in the los alamos ucn@xmath2 `` smooth field '' @xcite .
the analytic results , eqs .
( [ 10 ] ) and ( [ 28 ] ) ( blue solid ) , and the numerical solutions of ode ( [ 9 ] ) ( red dashed ) and of odes ( [ 21 ] ) , ( [ 22 ] ) ( green dot - dashed ) closely agree .
the first two turning points , shown by arrows , are at @xmath215 s ( when the particles pass the field minimum at a close distance ) and @xmath216 s ( when they are reflected at the high field near the wall ) .
the reset of @xmath4 to zero at these turning points and its subsequent fast increase , within microseconds or less , to the values given by the curves are not shown.,width=321 ] for ucns launched in ucn@xmath2 at time @xmath168 from a tp at @xmath217 m at velocity @xmath218 m / s ( tangential to the local equipotential surface ) .
the numerical integrations of ode ( [ 9 ] ) ( red dashed ) and of odes ( [ 21 ] ) , ( [ 22 ] ) ( green dotted ) show the rapid increase , within a few larmor periods , from @xmath219 at @xmath168 to the asymptotic behavior given by eqs .
( [ 10 ] ) and ( [ 28 ] ) ( blue solid ) .
the ensuing evolution of @xmath220 up to the next tp ( shown by the arrow ) is practically unaffected by the transient behavior at @xmath221.,width=321 ] fig .
[ fig : three ] shows the time - dependence of spin - flip probability @xmath45 for a particle released from rest at an arbitrary position in ucn@xmath2 , here @xmath214 m , calculated in three ways : ( a ) the analytic results from eqs .
( [ 10 ] ) and ( [ 28 ] ) ( solid curve ) , and the numerical solutions ( b ) of differential eq .
( [ 9 ] ) ( dashed ) and ( c ) of odes ( [ 21 ] ) and ( [ 22 ] ) ( dot - dashed ) .
the three curves closely agree except within short time intervals @xmath222 @xmath40s subsequent to passage though tps at non - zero velocity , such as the tps marked by the arrows .
the deviations ( not shown in fig . [
fig : three ] ) are due to the reset to @xmath219 at a tp .
they are shown in detail in fig .
[ fig : four ] for a ucn moving away from a tp at @xmath217 m at initial velocity @xmath223 m / s tangential to the local es . as expected from eq .
( [ 26 ] ) , @xmath220 jumps , within a few larmor periods ( @xmath224 @xmath40s ) , from @xmath225 to the asymptotic curve given by ( [ 10 ] ) ( or ( [ 28 ] ) ) .
the remainder of the wave evolution up to the next tp , shown by the arrow , is practically unaffected by the transient at @xmath221 . as a crucial test of the validity of numerical integration of ( [ 21 ] ) , ( [ 22 ] ) we verified the norm , @xmath226 , to be @xmath38 within @xmath38 ppm .
the demands on the precision of numerical integration of ( [ 9 ] ) , ( [ 21 ] ) and ( [ 22 ] ) become more stringent for paths through regions of higher magnetic field since large larmor frequencies require short time steps .
the spin - flip loss in magnetic storage of ucns in the los alamos ucn@xmath2 permanent magnet trap had been analyzed theoretically in @xcite for neutrons on a specific vertical path , and in @xcite for arbitrary motion . in the latter work we used a 1d model for the trapping field . in the present article
we have extended this analysis to arbitrary orbits in arbitrary fields in 3d space and report calculations of mean spin - flip rates for the ucn@xmath2 system and for multipole fields such as the cylindrical octupole of the hope project @xcite and the ioffe - type quadrupole trapping fields of @xcite .
we have also investigated a simplified 2d field model for the cylindrical multipole fields and shown that it yields analytic results for depolarization probabilities which are consistent with the more elaborate 3d simulations . in all cases relevant to magnetic ucn storage we have established agreement between the semi - classical approach , solving the spin - dependent se for the time - dependent field seen by the particle in a classical orbit , and a fully quantum mechanical analysis based on the space and time dependent se solved in wkb approximation . the relative difference between a first - order treatment ( in sec .
[ sec : iii ] ) and a higher - order analysis ( in sec .
[ sec : v ] ) of depolarization in the semi - classical framework is at most on the order of @xmath227 in practical applications .
we confirm and generalize the earlier conclusions of @xcite relating to `` majorana spin flip at zeros of the magnetic field '' .
magnetic ucn traps avoid locations of vanishing field by applying a holding field @xmath228 perpendicular to the trapping field . for typical values of @xmath228 we calculate spin - flip probabilities which are greater , by many orders of magnitude , than the majorana prediction @xcite which had been derived for an infinitely extended field rather than trapping fields of finite extent . for the magnetic traps investigated we have found an approximate power law behavior of spin - flip loss rate as a function of @xmath228 : @xmath229 ( to be compared with the exponential behavior , @xmath230 with adiabaticity parameter @xmath231 , for the majorana field model @xcite ) .
this implies that @xmath228 can be made large enough to render spin - flip loss negligible as compared to other possible sources of systematic error in neutron lifetime experiments , foremost that due to marginal trapping ( leaving aside the more fundamental question raised in @xcite whether or not neutron lifetime values derived from storage experiments should indeed be identical to those from beam - type experiments ) .
we have shown that this conclusion is not restricted to the simplified 1d field model previously used in @xcite but holds also for the actual 3d fields in suitable magnetic neutron bottles .
we are grateful to p. huffman for very helpful discussions .
the space and spin dependent wave function @xmath232 for a neutron with energy @xmath24 moving in a gravito - magnetic trapping field satisfies the se @xmath233 \psi\ ] ] where @xmath234 and @xmath235 , @xmath236 and @xmath237 are pauli matrices .
superscript @xmath238 indicates that @xmath239 , for the spin - up wave ( relative to the local magnetic field direction ) , and @xmath240 , for the spin - down wave , are functions of the three spatial coordinates .
the derivatives of @xmath61 and @xmath63 with respect to @xmath241 are of the same form as the temporal derivatives ( [ 4 ] ) . in terms of the spin angles @xmath242 and @xmath31 and of @xmath54 as defined below eq .
( [ 3 ] ) we have @xmath243 where the subscript @xmath244 denotes partial differentiation . keeping only the dominant contributions , as in eqs .
( [ 6 ] ) and ( [ 7 ] ) , the laplacian in ( [ a1 ] ) reads @xmath245 thus , in wkb approximation the spatial wave functions satisfy @xmath246 where @xmath247\ ] ] are the squared local wave numbers for the ( @xmath248 ) and ( @xmath249 ) spin state , respectively .
now we consider a ucn with spin ( @xmath248 ) starting at time @xmath168 at a tp and arriving at @xmath250 at the next tp which we label @xmath251 . at @xmath251
the ucn momentarily moves along the local es and we introduce a local cartesian system of coordinates centered at @xmath251 with @xmath252 and @xmath253 in the plane of this es .
@xmath254 points away from the direction into which the ucn is reflected .
coordinate system @xmath252 , @xmath253 , @xmath254 is defined for the narrow space where the particle motion can be considered linear and uniform .
since the @xmath255 and @xmath256 constituents of the wave function move as a unit the wave numbers @xmath257 and @xmath258 are the same for both .
thus we put @xmath259 as in @xcite , @xmath260 can be interpreted as a bloch - wave modulation due to the field rotation . substituting ( [ a8 ] ) in ( [ a5 ] ) and ( [ a6 ] )
we obtain @xmath261 and @xmath262e^{ik_{x^{\prime}}x^{\prime}}\,e^{ik_{y^{\prime}}y^{\prime}}\nonumber\\ & = \sum_{j = x^{\prime}\textrm{,}y^{\prime}\textrm{,}x^{\prime}}\alpha^{(3)}_{j}(x^{\prime},y^{\prime},z^{\prime})(\theta_{j}+i\phi_{j}\sin\theta),\end{aligned}\ ] ] where the wave numbers for the @xmath254 direction are given by @xmath263 with @xmath264 defined in ( [ a7 ] ) . among the components of @xmath265 and @xmath266
, the @xmath254 component @xmath267 plays a special role . even within the narrow space where the primed system of coordinates has been defined
, @xmath43 is not constant .
it becomes zero at the tp @xmath41 and , in this semi - classical picture , is only defined for @xmath268 . in wkb approximation
the solution of ( [ a9 ] ) is @xmath269 with @xmath270 .
( the exact quantum solution in form of an airy function has no singularity at @xmath41 and decays exponentially in the classically forbidden zone @xmath271 . ) to solve ( [ a10 ] ) we substitute the wkb approximation for the partial derivatives on the rhs , @xmath272 , @xmath273 , @xmath274 , and implement the total time derivative in the form @xmath275 with velocity @xmath276 . employing also ( [ a12 ] ) , ( [ a10 ] ) becomes @xmath277 the solution of ( [ a13 ] ) has been outlined in @xcite .
the phase factor for the wave @xmath278 is the same as for @xmath279 : @xmath280 .
therefore , in wkb approximation we have @xmath281 and the solution of ( [ a13 ] ) becomes @xmath282 the depolarization loss measured at tp @xmath251 is given by the probability current for spin - flipped ucns , @xmath283,\ ] ] leaving the storage space at @xmath41 in the positive @xmath254 direction . with ( [ a14 ] ) this current is @xmath284^{2}}=\frac{\hbar}{m}\frac{\omega^{2}}{4\omega^{2}_{l}},\ ] ] evaluated at @xmath41 ( i.e. , for the field variables @xmath285 and @xmath176 at the particle position at time @xmath250 ) . in the last step of ( [ a16 ] ) we have used the larmor frequency @xmath286 . to evaluate the spin - flip loss rate @xmath102 between the consecutive tps we divide the current ( [ a16 ] ) by the number @xmath287 of ( @xmath248 ) spin ucns moving between the tps in a channel with unit cross section centered at the trajectory .
the cross section of this channel is measured parallel to the es at every point along the path .
the @xmath287 particles within this volume contribute to loss current ( [ a16 ] ) , their decay rate @xmath288 equaling @xmath289 . denoting the wave number perpendicular to the ess traversed along the way by @xmath290 and using the wkb form @xmath291 as the particle density we have @xmath292 as for the 1d field model of @xcite , @xmath287 is given directly by the travel time @xmath293 . using ( [ a16 ] ) ,
the depolarization rate becomes @xmath294 evaluated for the field at the endpoint @xmath251 .
this agrees with the semi - classical result @xmath295 with @xmath77 given by eq .
( [ 10 ] ) . generalizing this result to arbitrary time @xmath187
, we choose the ucn position at @xmath187 as the center of reference system @xmath296 , with @xmath254 normal to the local es , and use ( [ a14 ] ) , ( [ a15 ] ) to obtain the identity @xmath297 this shows that the semi - classical and the quantum approaches to depolarization are equivalent , with @xmath77 directly corresponding to @xmath298 times the probability current @xmath299 .
there is an open question of interpretation : in the derivation of ( [ a13 ] ) we used a total time derivative in the form @xmath300 with velocity @xmath301 referring to the particle s motion along its classical path . in this sense ,
the quantum approach of this appendix does involve classical concepts .
use of the wkb method is not the only approximation made .
a similar caveat applies to the possibility of going to higher - order approximations in this quantum approach . in the semi - classical analysis
we were allowed to add , in ( [ 21 ] ) and ( [ 22 ] ) , terms such as @xmath302 which are of second order and had been neglected in the first - order approach of eqs .
( [ 6 ] ) and ( [ 7 ] ) .
however , adding the corresponding second - order contributions in the quantum treatment would require that we also add the second - order quantities neglected in the wkb approximation used to derive the odes ( [ a5 ] ) and ( [ a6 ] ) .
these would be replaced by coupled non - linear pdes of high complexity . at this stage
the semi - classical and quantum approaches clearly diverge .
t. bhattacharya , v. cirigliano , s. d. cohen _ et al .
_ , phys .
d @xmath303 , 054512 ( 2012 ) .
v. cirigliano , s. gardner and b. r. holstein , progress in particle and nuclear physics @xmath304 , 93 ( 2013 ) .
s. gardner and b. plaster , phys .
c @xmath305 , 065504 ( 2013 ) .
g. mention , m. fechner , th .
lasserre _ et al .
_ , phys .
d @xmath306 , 073006 ( 2011 ) . c. zhang , x. qian and p. vogel , phys .
d @xmath305 , 073018 ( 2013 ) .
a. coc , j .-
uzan and e. vangioni , j. cosmology and astroparticle physics @xmath307 , 050 ( 2014 ) .
f. iocco , g. mangano , g. miele _ et al .
_ , physics reports @xmath308 , 1 ( 2009 ) . g. j. mathews , t. kajino and t. shima , phys .
d @xmath304 , 021302 ( 2005 ) .
h. abele , prog .
@xmath309 , 1 ( 2008 ) .
j. s. nico , j. phys .
g : nuclear and particle physics @xmath310 , 104001 ( 2008 ) .
s. paul , nucl .
a @xmath311 , 157 ( 2009 ) .
d. dubbers and m. g. schmidt , rev .
@xmath306 , 1111 ( 2011 ) .
f. e. wietfeldt and g. l. greene , rev .
@xmath306 , 1173 ( 2011 ) . s. j. seestrom , ed . , _ next generation experiments to measure the neutron lifetime _ ,
workshop santa fe , nov . 2012 , world scientific , 2014 .
v. v. vladimirsky , jetp @xmath312 , 740 ( 1961 ) .
n. pokotilovski , jetp lett .
@xmath313 , 131 ( 2002 ) ; @xmath314 , jetp lett .
@xmath315 , 422 ( 2003 ) .
e. majorana , il nuovo cimento @xmath316 , 43 ( 1932 ) .
l. walstrom , j. d. bowman , s. i. penttila _ et al .
methods phys .
. a @xmath317 , 82 ( 2009 ) .
a. steyerl , c. kaufman , g. mller , s. s. malik , a. m. desai , phys .
c @xmath1 , 065501 ( 2012 ) .
a. steyerl , c. kaufman , g. mller , s. s. malik , a. m. desai , in _ 2012 next generation experiments to measure the neutron lifetime _ , ed .
s. j. seestrom , world scientific ( 2014 ) , pp . 75 - 86 .
p. huffman , c. r. brome , j. s. butterworth _ et al .
_ , nature @xmath318 , 62 ( 2000 ) . v. f. ezhov , a. z. andreev , g. ban _ et al .
_ , nucl . instr .
methods phys .
a @xmath311 , 167 ( 2009 ) ; _ measurement of the neutron lifetime with ultracold neutrons stored in a magneto - gravitational trap _ , arxiv.org/pdf/1412.7434 .
k. leung and o. zimmer , nucl .
methods phys .
a @xmath311 , 181 ( 2009 ) .
k. k. h. leung , p. geltenbort , s. ivanov , f. rosenau , o. zimmer , phys .
c @xmath319 , 045502 ( 2016 ) .
m. beck , k. eberhardt , ch .
geppert _ et al .
_ , in _ international workshop : probing fundamental symmetries and interactions with ucn _ , april 2016 , https://indico.mitp.uni-mainz.de/event/59 . s. materne , r. picker , i. altarev _ et al .
_ , nucl .
methods phys .
a @xmath311 , 176 ( 2009 ) .
w. paul , f. anton , l. paul , s. paul , w. mampe , z. physik c @xmath320 , 25 ( 1989 ) .
d. j. salvat , e. r. adamek , d. barlow _
_ , phys .
c @xmath321 , 052501 ( 2014 ) .
a. saunders , m. makela , y. bagdasarova _ et al .
_ , rev .
instrum .
84 , 013304 ( 2013 ) .
m. morse and h. feshbach , _ methods of theoretical physics _ , mcgraw hill , new york , 1953 , chap
h. goldstein , _ classical mechanics _ , addison - wesley , london , 1950 . c. r. brome , j. s. butterworth , s. n. dzhosyuk _ et al .
_ , phys . rev .
c @xmath322 , 055502 ( 2001 ) .
l. yang , c. r. brome , j. s. butterworth _ et al .
_ , rev .
@xmath323 , 031301 ( 2008 ) .
a. steyerl , c. kaufman , g. mller _ et al .
a @xmath321 , 052129 ( 2014 ) . g. l. greene and p. geltenbort , _ the neutron enigma _ , sci .
american , april 2016 . | we analyze the spin flip loss for ultracold neutrons in magnetic bottles of the type used in experiments aiming at a precise measurement of the neutron lifetime , extending the one - dimensional field model used previously by steyerl @xmath0 [ phys .
rev .
c @xmath1 , 065501 ( 2012 ) ] to two dimensions for cylindrical multipole fields .
we also develop a general analysis applicable to three dimensions . here
we apply it to multipole fields and to the bowl - type field configuration used for the los alamos ucn@xmath2 experiment . in all cases
considered the spin flip loss calculated exceeds the majorana estimate by many orders of magnitude but can be suppressed sufficiently by applying a holding field of appropriate magnitude to allow high - precision neutron lifetime measurements , provided other possible sources of systematic error are under control . |
models of planar random geometry appear in physics in the context of two - dimensional quantum gravity and provide an interplay between mathematical physics and probability theory .
causal dynamical triangulation ( cdt ) and its predecessor dynamical triangulation ( dt ) , introduced by ambjrn and loll ( see @xcite ) , constitute attemps to provide a meaning to the formal expressions appearing in the path integral quantization of gravity ( see @xcite , @xcite for an overview ) .
the idea is to approximate the path integral by changing integration with respect to continuous random geometry by integration with respect to dts or cdts . as a result ,
the path integral with respect to continuous random geometries is replaced by a weighted sum over discrete geometries , where the weights are defined by the discrete form of the original action . in this paper
we adopt the framework of cdts coupled with an ising spin system . putting a spin system on the collection of all causal triangulations
is interpreted as a _ coupling gravity with matter_. this type of coupling has been a subject of persistent interest in physics once the matrix model methods were successfully applied to the ising model on random lattice with coordination number @xmath3 ( see @xcite and @xcite for details ) . in this article
we discuss the ising model coupled to causal triangulations in details , study the termodinamic function known as free energy , and formally define the critical curve for the annealed model .
this annealed model was first introduced by @xcite .
this model was also studied by @xcite , where the authors develop a systematic counting of embedded graphs and evaluate thermodynamic functions at high and low temperature expansion ( see @xcite and @xcite for other progress made for the annealed model ) .
the main advantage of the cdts over the dts is that the ensemble of cdts is more regular than the ensemble of the dts , in the sense that does not permite spacial topology changes and leads to a fractal dimension @xmath4 ( see @xcite , @xcite , @xcite , @xcite ) . on the other hand ,
the causality constraints still make it difficult to find an analytical solution of the ising model coupled to cdts , however some progress has been recently made about existence of gibbs measures and phase transitions ( see @xcite , @xcite for details ) . employing transfer matrix techniques and in particular the krein - rutman theory of positivity - preserving operators
, @xcite determines a region in the quadrant of parameters @xmath5 where the finite - volume free energy has a thermodynamic limit . in this article
the authors provide first approximations of the infinite - volume free energy , and conditions for existence of gibbs measures .
furthermore , the authors prove that in the determined region the annealed model presents a subcritical behaviour , i.e. , there exists a unique gibbs distribution for the model , and that with respect to this limit measure the average surface has a form of an infinite narrow tube , which is essentially one dimensional .
computation of the critical curve ( point ) for models in statistical mechanics has always been a challenging problem , and since fk models were introduced by fortuin and kasteleyn in 1969 ( see @xcite ) , they have become an important tool in the study of phase transition for spin models .
the goal of this article is to use the fk representation of the ising model coupled to cdt in order to obtain lower and upper bounds for the critical curve .
we employ the results of @xcite to establish a region where the n - strip gibbs measure has no weak limit as @xmath6 , where the boundary of this region provides a lower bound for the critical curve .
futhermore , we expand the region of subcritical behaviour of the model , and provide a better approximation to the free energy .
the aforementioned fk representation utilizes a family of poisson point processes and the lie - trotter product formula to interpret exponential sums of operators as random operator products .
this representation was originally derived in @xcite ( see @xcite , @xcite for an overview ) .
results presented in this paper , along with the results in @xcite , can be utilized to determine a region in the quadrant of parameters @xmath5 where the critical curve is possibly located . moreover
, the fk representation allows to compute lower and upper bounds for the infinite - volume free energy .
the paper is organized as follows . in sections
[ sect2.1]-[sect2.3 ] we review the main features of lorentzian cdts and describe the ising model coupled to cdt . in addition , we define the free energy and provide some important properties of the model . in section [ sect2.4 ]
we present the main results ( theorems [ theomain1 ] and [ theomain2 ] ) . in section [ fkrepresent ]
we review the fk representation of ising model coupled to cdt
. sections [ sect3.2 ] and [ sect3.3 ] contain the proofs of the theorems formulated in section [ sect2.4 ] .
we start with an introduction to the basic features of cdts .
more details and properties about causal triangulations can be be found in @xcite , @xcite , @xcite , @xcite , @xcite,@xcite , @xcite . as in @xcite
, we will work with rooted causal dynamic triangulations of the cylinder @xmath7 $ ] , @xmath8 , which have @xmath9 bonds ( strips ) @xmath10 $ ] . here
@xmath11 stands for a unit circle ( see figure [ fig1 ] ) .
formally , a triangulation @xmath12 of @xmath13 is called a _ rooted causal dynamic triangulation _ ( ctd ) if the following conditions hold : * each triangular face of @xmath12 belongs to some strip @xmath14 $ ] , @xmath15 , and has all vertices and exactly one edge on the boundary @xmath16 of the strip @xmath17 $ ] ; * the number of edges on @xmath18 should be finite for any @xmath19 : let @xmath20 be the number of edges on @xmath18 , then @xmath21 for all @xmath19 . and have a root face , with the anti - clockwise ordering of its vertices @xmath22 , where @xmath23 and @xmath24 lie in @xmath25 .
the cdts arise naturally when physicists attempt to define a fundamental path integral in quantum gravity .
see @xcite for a review of the relevant literatute ; for a rigorous mathematical background of the model , cf .
additional properties of cdts have been studied in @xcite .
a rooted cdt @xmath12 of @xmath13 is identified with a compatible sequence @xmath26 where @xmath27 is a triangulation of the strip @xmath28 $ ] .
the compatibility means that @xmath29 note that for any edge lying on the slice @xmath30 belongs to exactly two triangles : one up - triangle from @xmath27 and one down - triangle from @xmath31 .
this provides the following relation : the number of triangles in the triangulation @xmath12 , denoted by @xmath32 , is twice the total number of edges on the slices .
more precisely , remind that @xmath33 is the number of edges on slice @xmath34 . then , for any @xmath35 , @xmath36 in the physical approach to statistical models , the computation of the partition function is the first step towards a deep understanding of the model , enabling for instance the computation of the free energy .
follow this approach , the computation of the partition function for the case of pure cdts , was first introduced and computed in @xcite ( see also @xcite for a mathematically rigorous account ) . considering triangulations with periodical spatial boundary conditions , i.e. the strip @xmath37 is compatible with @xmath38 , and let @xmath39 denote the set of causal triangulations on the cylinder @xmath13 with this boundary condition , thus the partition function for rooted cdts in the cylinder @xmath13 with periodical spatial boundary conditions and for the value of the cosmological constant @xmath40 is given by @xmath41 moreover , the periodical spatial boundary condition on the cdts permits to write the partition function @xmath42 in a trace - related form @xmath43 this gives rise to a transfer matrix @xmath44 describing the transition from one spatial strip to the next one .
it is an infinite matrix with positive entries @xmath45 $ ] .
( b ) geometric representation of a cdt with periodic spatial boundary condition.,width=453,height=226 ] employing the @xmath9-strip partition function for pure cdts with periodical boundary condition , defined by the formula ( [ yamb - pf1 ] ) , we define the @xmath9-strip gibbs probability distribution for pure cdts @xmath46 the asymptotic properties of the partition function of the model and existence of the weak limit of the measure @xmath47 for @xmath48 are provided in @xcite .
these results were obtained by utilizing transfer - matrix formalism and tree parametrization of lorentzian triangulations .
in addition , the transfer - matrix formalism and physical considerations suggest that , as @xmath49 , the partition function is controlled by the largest eigenvalue @xmath50 of the transfer matrix ( [ yamb - tmpg ] ) : @xmath51 where @xmath52 ^ 2.\ ] ] that heuristic result was proved in @xcite . the following properties hold and will be utilized to prove the main results . *
property 1 . *
( theorem 1 in @xcite ) .
_ for any @xmath53 the following relation holds : @xmath54 furthermore , the @xmath9-strip gibbs measure @xmath47 converges weakly to a limiting measure @xmath55 . _ * property 2 . *
( proposition 5 , @xcite ) .
_ for any @xmath56 , the @xmath9-strip partition function @xmath42 exists if _ @xmath57 another proof of property 1 can be found in @xcite . in order to prove this property the authors utilize the transfer - matrix formalism and krein - rutman theorem .
inequality ( [ yamb - e14 ] ) in property 2 implies that if @xmath58 , then there exists @xmath59 such that @xmath60 if @xmath61 .
let @xmath62 and @xmath63 denote the set of triangles of the triangulation @xmath12 and the set of triangles of the strip @xmath64 , respectively .
as in @xcite , each triangle from the triangulation @xmath12 is associated with a spin taking values @xmath65 .
an @xmath9-strip configuration of spins is represented by a collection @xmath66 where @xmath67 is a configuration of spins @xmath68 over triangles @xmath69 forming a triangulation @xmath27 .
we consider a usual ferromagnetic ising - type energy where two spins @xmath68 and @xmath70 interact if their corresponding supporting triangles @xmath69 , @xmath71 share a common edge . in this case the triangles
@xmath69 , @xmath71 are referred to as nearest neighbors and denoted by @xmath72 .
thus , the hamiltonian utilized for the annealed model is given by @xmath73 the partition function for the @xmath9-strip ising model coupled to cdt , at the inverse temperature @xmath74 and the cosmological constant @xmath40 , is given by @xmath75 where @xmath76 stands for the number of triangles in the triangulation @xmath27 , @xmath77 denotes the inner energy of the configuration @xmath78 , and @xmath79 is the energy of interaction between the neighboring triangles belonging to the adjacent strips @xmath80 $ ] and @xmath81 $ ] . as in the case of pure cdts ,
application of the transfer - matrix formalism to ising coupled to cdts suggests that the partition function would be rewritten as @xmath82 the formal definition of the operator @xmath83 and a proof of the identity ( [ trmatrix ] ) can be found in @xcite .
following @xcite , we define the operator @xmath83 on the hilbert space @xmath84 where the subscript t - c refers to triangulations and spin - configurations of square - summable functions @xmath85 where the argument @xmath86 runs over single - strip triangulations and supported configurations of spins , with the scalar product @xmath87 and the induced norm @xmath88 . as in the case of the pure cdt ,
we introduce the @xmath9-strip gibbs probability distribution associated with ( [ yamb - pf ] ) as @xmath89 let @xmath90 be the set of _ gibbs measures _ given by the closed convex hull of the set of weak limits : @xmath91 we define the domain of parameters where the weak limit gibbs distribution exists , as follows @xmath92 further , the critical curve @xmath93 for the ising model coupled to cdt is defined as @xmath94 this critical curve is of major importance , since it splits the parameter space into two following regions : a region where the gibbs distributions exist , and a region where the gibbs distribution does not exist , due to large boltzman factors associated with the triangulations which results in an infinite partition function .
this behaviour is observed at zero and infinite temperature in the annealed model . in order to study behaviour of the critical curve we focus on the thermodynamic function which is defined as @xmath95 note that , @xmath96 utilizing ( [ yamb - e13 ] ) , it is easy to show that @xmath97 converges to @xmath98 as @xmath49 .
this result implies that the free energy exists at infinitely high temperature for @xmath99 , @xmath100 it can be noticed that , if @xmath101 , and @xmath102 , then @xmath103 .
thus , utilizing the results ( [ yamb - e13 ] ) and ( [ yamb - e14 ] ) ( see @xcite for details ) , we find the infinite - volume free energy @xmath104 at infinitely high temperature @xmath105 in addition , if the temperature tends to infinity , we obtain that the annealed model is a pure cdt model with a critical parameter @xmath106 .
furthermore , for all @xmath107 the gibbs measure is unique , and as in the case of pure cdts there exist two types of triangulations with respect to this measure . if @xmath99 we obtain subcritical triangulations . utilizing the identity ( [ freeein0 ] ) and the results obtained in @xcite and @xcite
it can be showed that in this case the triangulations have a bounded diameter with probability 1 , which is essentially one dimensional random geometry .
the boundary value @xmath108 yields a two dimensional random geometry with hausdorff dimension @xmath4 ( see @xcite , @xcite and @xcite for details ) .
+ in this paper we show that for any @xmath109 the infinite - volume free energy of the annealed model has the same behaviour as ( [ freeein0 ] ) except for a small interval @xmath110 $ ] where the behaviour of the function is unknown .
on the other hand , according to the numerical results ( see @xcite , @xcite , @xcite ) there exists a unique value @xmath111 such that @xmath112 for @xmath113 , @xmath114 for @xmath115 , and @xmath116 for @xmath117 , however the exact value of @xmath118 has not been computed yet .
the interval @xmath119 is computed explicitly in the next section .
+ note that , if @xmath109 , for any triangulation @xmath120 the energy of any spin configuration @xmath121 on @xmath120 is larger than or equal to the energy of a pure configuration ( all @xmath122 s or all @xmath123 s ) .
i.e. , @xmath124 .
thus , for any @xmath125 the following inequality holds @xmath126 therefore , @xmath127 hence , the inequality @xmath128 provides a sufficient condition for subcritical behaviour ( uniqueness of gibbs measure ) of the ising model coupled to cdts , and provides a region where the infinite - volume free energy exists and is finite , i.e. , does not have singularities .
now , based on ( [ yamb - e13 ] ) and ( [ bd ] ) an upper bound for the critical curve of the annealed model is given by @xmath129 obviously , the upper bound given by ( [ am ] ) is not sharp for all @xmath130 .
for example , for the low values of @xmath130 , which correspond to high temperature , the upper bounds provided by @xcite and @xcite are more accurate . in this paper
we focus on finding lower and upper bounds for the critical curve for all values of @xmath130 .
in the previous subsection we examine basic notions of statistical mechanics of the model and review some properties about the phase space , the free energy and gibbs distributions . in this section
we focus on presenting the main results of this paper . utilizing these results
we provide a first phase diagram for the ising model coupled to cdts in the parameters @xmath130 and @xmath40 , and compute an approximation of the infinite - volume free energy .
the exact phase diagram at infinite temperature was computed in the previous subsection . in order to construct an upper bound for the critical curve
, we define the function @xmath131 as follows @xmath132 where @xmath133 and @xmath134 are determined by @xmath135 let @xmath136 be a strictly increasing function defined as @xmath137 it is easy to see that @xmath138 in addition , it is well known that in the region @xmath139 the free energy @xmath140 is finite . consequently , in this region the annealed model has a unique gibbs measure ( see @xcite for details ) .
let @xmath141 be triangulations of a single strip @xmath142 $ ] and @xmath143 be their corresponding spin configurations .
given @xmath144 , we define a finite - dimensional cylinder @xmath145 as follows @xmath146 [ theomain1 ] if @xmath147 , where @xmath148 then there exist @xmath59 such that if @xmath149 the partition function @xmath150 .
moreover , the gibbs distribution @xmath151 with periodic boundary conditions can not be defined by utilizing the standard formula with a normalising denominator @xmath152 , consequently , there is no limiting probability measure @xmath153 as @xmath6 .
formally theorem [ theomain1 ] states that , for any finite - dimensional cylinder @xmath154 we obtain @xmath155 for @xmath156 .
+ let @xmath157 be positive solutions of the following equations @xmath158 @xmath159 respectively .
theorem [ theomain1 ] , along with the results in @xcite , provides two - side bounds for the critical curve .
[ theomain2 ] the critical curve @xmath93 satisfies the following inequalities . 1 .
if @xmath160 and @xmath161 , then @xmath162 2 . if @xmath160 and @xmath163 , then @xmath164 3 .
if @xmath160 and @xmath165 , then @xmath166 utilizing ( [ beta=0 ] ) and @xmath167 of theorem [ theomain2 ] , we obtain the exact value of the critical curve at infinite temperature , i.e. , @xmath168 . employing the results of @xcite
we conclude that the set of gibbs measures at infinite temperature is a single point for @xmath107 , and is empty for @xmath169 . as a by - product of the proof of theorems [ theomain1 ] and [ theomain2 ]
, we obtain a lower and upper bound for the infinite - volume free energy .
these bounds are provided in the following corollary .
+ defined in ( [ functpsi ] ) ) and dash - dotted line ii is where the limiting gibbs probability measure exists and is unique .
the critical curve lies in the region below the dotted curve i and dash - dotted line ii but above the continuous curve iii and dashed line iv.,width=377 ] let @xmath119 denote the interval @xmath170 $ ] , where @xmath171 @xmath172 [ mainresult3 ] for @xmath173 such that @xmath174 , the free energy @xmath140 for the ising model coupled to cdts is finite and satisfies the following inequalities , @xmath175 here @xmath176 is given by ( [ lambda(g ) ] ) .
finally , we obtain the behaviour of the free energy for the annealed model .
[ mainresult4 ] for @xmath125 fixed , the free energy @xmath177 is an increasing function in @xmath40 , for all @xmath178 and @xmath179 for @xmath180 .
in addition , @xmath181 and @xmath182 .
remember that the exact solution of the free energy at @xmath101 was computed in ( [ freeein0 ] ) .
the result presented in corollary [ mainresult4 ] is illustrated in figure [ freeenergy ] . for @xmath130 fixed . in the gray region
we show a possible behaviour of analytic continuation of the free energy on the right and left side , according to the numerical results given in @xcite , @xcite , @xcite.,width=377 ]
we start this section with a brief review of the existing results for representation of the ising model in terms of a two continuum random - cluster model .
we apply the results obtained in @xcite and @xcite to lorentzian triangulations with periodical boundary condition .
as mentioned above , each triangle from the triangulation @xmath12 is associated with a spin taking values @xmath65 .
this is equivalent to associating spins on each vertex of the dual lorentzian triangulation . starting from this section
we shell use the same notation @xmath12 for the dual lorentzian triangulation with a periodical boundary condition . note that the ising model defined in subsection [ sect2.3 ] was defined on dual triangulation .
let @xmath183 be the partition function of the ising model on a fixed dual lorentzian triangulation @xmath12 , at inverse temperature @xmath125 @xmath184 here @xmath185 represents the energy of the configuration @xmath186 , defined by ( [ isinghami ] ) .
thus , utilizing ( [ 2.20 ] ) the partition function for the @xmath9-strip ising model coupled to cdt , at the inverse temperature @xmath125 for the cosmological constant @xmath40 , can be rewritten as @xmath187 for any dual lorentzian triangulation with periodical boundary condition @xmath12 and any edge @xmath188 of @xmath12 , we select a poisson process @xmath189 of points in @xmath190 with intensity @xmath191 , such that the members of the family of poisson processes @xmath192 are independent of each other , where @xmath193 denotes the set of edges of the dual lorentzian triangulation @xmath12 .
note that @xmath194 for all @xmath195 .
let @xmath196 denote the probability associated with the family of poisson processes on the interval @xmath197 $ ] .
an arrival of @xmath198 at time @xmath69 can be geometrically interpretated as a link between neighbor vertices @xmath199 , @xmath200 at time @xmath69 .
we say that two vertices @xmath201 of @xmath12 ( not necessarily neighbor vertices ) are connected , if and only if there exists a path , in the sense of continuum percolation , connecting @xmath69 and @xmath71 ( see @xcite for an overview ) .
this property will be denoted by @xmath202 . the relation @xmath203 generates a partition of the set of vertices into clusters . given a realization @xmath204 of
the poisson processes @xmath205 on @xmath197 $ ] , we denote by @xmath206 the number of clusters in the realization @xmath204 .
papers @xcite , @xcite and @xcite derived a representation of the ising model in terms of a continuum random - cluster model .
aplication of this representation to an ising model whithout external magnetic field on a fixed dual lorentzian triangulation @xmath12 , yields the following representation for the partition function @xmath183 .
[ prop_aizenman ] let @xmath195 and @xmath125 .
then the following identity holds @xmath207 utilizing the @xmath9-strip gibbs probability distribution @xmath47 , defined in ( [ q ] ) for the pure cdt case , ( [ 2.29 ] ) and proposition [ prop_aizenman ] we obtain the fk representation of the partition function for the annealed model , at the inverse temperature @xmath125 and for the cosmological constant @xmath40 @xmath208 where @xmath209 and @xmath210 is defined by ( [ yamb - pf1 ] ) .
the result ( [ fk_ising ] ) provides asymptotic behaviour for the free energy , for large @xmath130 , as follows .
let @xmath211 be a lorentzian triangulation .
if @xmath130 is large enough then the probability that all the poisson processes @xmath189 , @xmath212 had at least one arrival in the time interval @xmath197 $ ] tends to 1 , which implies that almost all vertices are linked , that is , @xmath213 .
thus , utilizing the representation ( [ fk_ising ] ) we obtain that @xmath214 consequently , the critical inequality @xmath215 gives a necessary ( and probability tight ) criticality condition for the annealed model .
therefore , for any @xmath130 , which is sufficiently large , the free the free energy have the following behaviour , @xmath216 furthermore , this asymptotic property establish that for large @xmath130 @xmath217 where @xmath218 is the euclidean distance between two sets .
a similar heuristic analysis was presented and confirmed by numerical simulations in @xcite .
the purpose of this section is to study the behaviour of the partition function @xmath219 as a function of @xmath130 , for @xmath220 , @xmath99 , @xmath221 .
we further utilize the properties of this function in order to obtain a lower bound ( not sharp ) for the critical curve .
[ prop_positive ] if @xmath222 , then @xmath223 for any @xmath224 . utilizing the definition of the partition function @xmath225 , it is easy to obtain the following expression @xmath226 where @xmath227 denotes the gibbs measure of an ising model on @xmath12 , given by @xmath228 where @xmath185 is defined by ( [ isinghami ] ) .
thus , implementation of the first griffiths inequality ( see @xcite and @xcite for an overview ) yields @xmath229 this completes the proof .
now , in particular , @xmath230 for all @xmath125 .
thus , the following inequality can be derived @xmath231 utilizing property ( [ yamb - e14 ] ) and ( [ triv_lowerbound ] ) , we obtain the trivial lower bound for the critical curve , @xmath232 ( see figure [ fig2 ] ) . in addition , it follows from ( [ triv_lowerbound ] ) that the free energy satisfies the inequality @xmath233 , for @xmath220 and @xmath107 . finally , this lower bound along with the results in @xcite prove that @xmath168 , and that @xmath234 for all @xmath107 .
let @xmath12 be a dual lorentzian triangulation on the cylinder @xmath13 .
given @xmath235 , we define the set of realizations @xmath204 which splits the set of vertices in @xmath236 clusters , @xmath237 thus , we obtain the following representation of ( [ eq3.18 ] ) @xmath238 let @xmath239 and let @xmath240 be the corresponding cluster decomposition of the set @xmath62 .
in fact , each cluster @xmath241 is a subgraph of @xmath12 formed by vertices @xmath242 and edges @xmath243 .
note that @xmath241 is a random variable and that the cluster decomposition @xmath240 can include isolated vertices .
denote by @xmath244 and @xmath245 , the number of vertices ( triangles ) in cluster @xmath241 and the number of edges in @xmath241 , respectively .
note that for any decomposition @xmath240 , @xmath246 and @xmath247 depend on the geometry of the cluster @xmath241 .
note also that @xmath248 .
now , denote by @xmath249 the set of all maximal unordered partitions of @xmath62 , i.e. an element of @xmath249 is an unordered @xmath250-ple @xmath251 of maximal connected subgraphs @xmath252 , with @xmath253 , such that for any @xmath254 , @xmath255 , @xmath256 , @xmath257 and @xmath258 .
a graph @xmath252 is maximal in the sense that : if @xmath259 are nearest neighbor vertices in @xmath252 , then @xmath260 .
given a subgraph @xmath261 belonging to some element of @xmath249 , we define the set of spanning subgraphs of @xmath261 as @xmath262 that is , if @xmath263 , then @xmath264 and @xmath261 have exactly the same vertex set . finally , the probability that two nearest neighbor vertices @xmath259 are linked is given by @xmath265 . then , denoting by @xmath266 , we can rewrite the partition function @xmath267 as follows @xmath268 where @xmath269 here @xmath270 denotes the number of edges in the subgraph @xmath264 .
notice that @xmath271 for all @xmath263 .
a convenient expansion parameter for our analysis is @xmath272 .
we now are interested in deriving expansions of the function @xmath273 , @xmath274 for values @xmath275 and @xmath276 .
note that by the definition of a spanning subgraph , it is easy to see that for any @xmath277 the following inequalities hold @xmath278 if @xmath279 the unique spanning subgraph @xmath264 that do not satisfy the right - hand side of inequality ( [ inequalityclus ] ) is @xmath280 . in this case , the graph @xmath264 has @xmath281 vertices and @xmath282 edges . if @xmath283 the inequality ( [ inequalityclus ] ) is satisfied . in order to obtain lower bounds for the representation ( [ newformpf ] ) , we implement inequality ( [ inequalityclus ] ) for two different cases : @xmath284 and @xmath285 .
_ the case @xmath284 @xmath286 equivalent to @xmath287 @xmath288_. utilizing the inequality ( [ inequalityclus ] ) , we obtain that the following inequality holds for @xmath289 @xmath290 thus , if @xmath291 and @xmath292 ( the case @xmath293 will treate separately later ) , we obtain @xmath294 where @xmath295 , that is , the number of spanning subgraphs contained in @xmath261 .
further , we rewrite the partition function @xmath267 as follows @xmath296 where @xmath297 . + now , observing that @xmath298 and employing the inequality ( [ firstineq ] ) , we obtain the following lower bound @xmath299 notice that , given any decomposition @xmath300 , @xmath301 for all @xmath302 , where equality holds only if @xmath252 is a spanning tree .
further , if the degree of freedom , @xmath236 is large enough , there exists positive proportion of decompositions @xmath303 of @xmath12 with @xmath304 for all @xmath236 .
thus , the lower bound can be obtain as @xmath305 .
for example , for @xmath306 and @xmath307 , all elements of any decomposition are spanning trees .
substituting the lower bound in ( [ 1lowerb ] ) , we get the following inequalities @xmath308 now , the function @xmath309 can be rewritten as @xmath310 if @xmath311 and number of strip @xmath9 is large enough , we can conclude that the behaviour of the function @xmath309 can be described by the term @xmath312 .
in fact , it is easy to see that @xmath313 for any triangulation @xmath12 . utilizing this result
, we derive the following lower bound for the partition function @xmath314 of the ising model on any triangulation @xmath12 and @xmath221 , @xmath315 implementation of identity ( [ fk_ising ] ) , and lower bound ( [ eq3.24 ] ) provides the following lower bound for the partition function of the annealed model , @xmath316 thus , utilizing the asymptotic property given in ( [ yamb - e14 ] ) , we obtain that the partition function @xmath152 exists only if @xmath317 now , let us discuss the case where @xmath6 .
[ prop3.2 ] if @xmath318 such that @xmath319 then there exists @xmath59 such that the partition function @xmath320 whenever @xmath149 .
moreover , the gibbs distribution @xmath151 with periodic boundary conditions can not be defined by the standard formula with @xmath152 being a normalising denominator .
consequently , there is no limiting probability measure @xmath153 as @xmath6 , which implies that for any finite - dimensional cylinder @xmath154 , @xmath155 whenever @xmath156 . _
the case @xmath321 @xmath286equivalent to @xmath322 @xmath288_. similarly to the first case , utilizing ( [ inequalityclus ] ) we obtain the following inequalities @xmath323 which hold for all spanning subgraphs @xmath289
. thus , if @xmath291 and @xmath292 we obtain @xmath324 employing the inequality ( [ secondineq ] ) , we obtain the lower bound for the partition function @xmath325 on dual triangulation @xmath12 , @xmath326 implementation of the same tecnique as in the previous case yields the lower bound for the critical curve , @xmath327 the proof of theorem [ theomain1 ] follows immediately from proposition [ prop3.2 ] and the lower bound given by the griffiths inequality .
the lower bound given by ( [ notimprov ] ) does not improve the lower bound given by the first griffiths inequality . for each @xmath328 , we define the following sets in @xmath329 @xmath330 @xmath331 note that @xmath332 for all @xmath333 , @xmath334 , and that the finite - volume gibbs measure @xmath335 on the set @xmath336 exists .
we also define the @xmath9-strip functions @xmath337 associated with the set @xmath338 , for each @xmath333 , as follows @xmath339 by the properties of the trace class operators ( see @xcite for an overview ) , we deduce that @xmath340 is a monotone decreasing sequence of measurable functions such that @xmath341 , for all @xmath125 , where @xmath342 is defined in ( [ f_1 ] ) .
thus , we prove the existence of the pointwise limit @xmath343 another important fact is that the graph of the function @xmath344 provides an upper bound for the critical curve .
this property is a consequence of the spectral properties of the operator @xmath83 introduced in ( [ trmatrix])(see @xcite for the details and @xcite for an overview ) and the definition of the function @xmath344 .
consequently , implementation the results of @xcite and the condition of subcriticality ( [ am ] ) , we can prove the following proposition .
[ prop3.1 ] there exist @xmath59 such that for @xmath345 , the following property of functions @xmath346 is fulfilled : 1 . if @xmath347 , then @xmath348 where @xmath349 is a positive solution of @xmath350 and @xmath351 is defined in @xmath352 .
2 . if @xmath165 , then @xmath353 combining ( [ f_tc ] ) , ( [ eq5.2 ] ) , ( [ eq5.3 ] ) , and letting @xmath6 , we obtain the desired upper bound for the limit function @xmath354 @xmath355 since that the graph of the function @xmath354 lies above of the critical curve , the right - hand side of ( [ upperbound ] ) provides an upper bound for the critical curve . the upper bound for the critical curve @xmath93 is a consequence of inequality ( [ upperbound ] ) .
the lower bound is a consequence of theorem [ theomain1 ] .
this concludes the proof of the theorem .
in this article we present a step towards improving the subcriticality domain for an ising model coupled to two - dimensional cdt introduced in @xcite . in doing
so we employ fk representation of the ising model on causal triangulations and combinatorial approximation .
in addition , we make a first step towards determining the critical curve of the model .
numerical evidence shows that the model has phase transition ( see @xcite , @xcite , @xcite ) , but this fact has never been proved explicitly . in this article
we presente mathematical evidence of existence of the critical curve by studying the infinite - volume free energy , and computing a region where the critical curve for the annealed model is possibly located ( see figure [ fig2 ] ) . the discussion in section [ sect2.3 ] along with corollary [ mainresult3 ]
suggest that free energy can be expresed as @xmath356 , where the function @xmath357 satisfies @xmath358 .
this result leads to a plausible conjecture that the boundary of the subcritical domain coincides with the locus of points @xmath173 where @xmath359 , however in order to obtain more precise conclusions a further investigation is required .
in addition , theorem [ theomain1 ] and theorem [ theomain2 ] show that with respect to the weak limit gibbs measure @xmath153 , the average diameter of any fixed height @xmath250 is bounded in the region @xmath360 ( by a constant independent of @xmath250 ) with probability 1 , which is essentially a one dimensional random geometry .
thus , we can expect that this subcritical behaviour can be extended to the region @xmath361 , and that any causal triangulation on the critical curve has hausdorff dimension @xmath4 , @xmath362-a.s . finally , if the annealed model has this property , then the region where it presents a phase transition is possibly located on the critical curve .
this direction also requires further research .
i would like to thank prof .
y. suhov and prof .
a. yambartsev for very valuable discussions and his encouragement .
this work was supported by fapesp ( projects 2012/04372 - 7 , 2013/06179 - 2 and 2014/18810 - 1 ) .
further , the author thanks the ime at the university of so paulo for warm hospitality . aizenman m. , klein a. , newman c.m . : _
percolation methods for disordered quantum ising model_. in r. kotecky , editor , phase transition : mathematics , physics , biolgy , ... pages 129137 . world scientific , singapure ( 1992 ) .
hernndez , j.c . , yambartsev , a. , suhov , y. , zohren , s. : _ bounds on the critical line via transfer matrix methods for an ising model coupled to causal dynamical triangulations_. journal of mathematical physics , * v. 54 * , p. 063301
http://arxiv.org/pdf/1301.1483.pdf[1301.1483 ] . | this paper extends results obtained by @xcite for the annealed ising model coupled to two - dimensional causal dynamical triangulations .
we employ the fortuin - kasteleyn ( fk ) representation in order to determine a region in the quadrant of parameters @xmath0 where the critical curve for the annealed model is possibly located .
this is done by outlining a region where the model has a unique infinite - volume gibbs measure , and a region where the finite - volume gibbs measure does not have weak limit ( in fact , does not exist if the volume is large enough ) .
we also improve the region of subcritical behaviour of the model , and provide a better approximation of the free energy . + + * 2000 msc . * 60f05 , 60j60 , 60j80 . + * keywords : * causal dynamical triangulation ( cdt ) , ising model , partition function , gibbs measure , transfer matrix , fk representation * critical region for an ising model coupled to causal dynamical triangulations * _ j. cerda - hernndez_@xmath1 @xmath2 institute of mathematics , statistics and scientic computation , + university of campinas - unicamp , + rua srgio buarque de holanda 651 , cep 13083 - 859 , campinas , sp , brazil .
+ e - mail : [email protected] . |
the dynamics of a classical inflaton field is commonly assumed to be governed by a local equation of motion of the form @xmath0 where the friction term @xmath1 represents the dissipative mechanism through which inflaton energy is converted into particles . in particular , in ` warm inflation ' scenarios @xcite , this term is taken to be large enough for a significant fraction of the energy to be dissipated during inflation .
assuming that the classical inflaton @xmath2 is the expectation value of quantum scalar field @xmath3 , its equation of motion can be derived straightforwardly from a suitable quantum field theory , and it is _ not _ of the form ( [ localeom ] ) .
for example , if @xmath2 couples to a scalar quantum field @xmath4 through an interaction @xmath5 , one obtains a term @xmath6 in the equation of motion , and the expectation value depends on @xmath2 through the effective mass of the @xmath4 particles , @xmath7 .
the field @xmath4 here might be the quantum part of @xmath3 ( that is , @xmath8 , with @xmath9 ) or it might be an independent field . in any case
, the expectation value @xmath10 is a highly _ nonlocal _ quantity : it depends in a complicated way on the entire history of @xmath2 , in contrast to the local equation ( [ localeom ] ) , where @xmath2 , @xmath11 and @xmath12 are all evaluated at a single time @xmath13 .
this situation is quite generic : a yukawa coupling @xmath14 to a spin-@xmath15 field @xmath16 produces a @xmath2-dependent mass for the @xmath16 particles and a term @xmath17 in the equation of motion , and in general the equation of motion contains a sum of terms of this sort , but no friction term .
the question naturally arises , whether the true equation of motion might be adequately approximated , under suitable circumstances , by a local equation similar to ( [ localeom ] ) .
an apparently rather plausible argument suggests that this is indeed possible when @xmath2 changes sufficiently slowly with time , and supplies an estimate of the friction coefficient @xmath18 . to evaluate @xmath19 at a _ fixed _
time @xmath13 , one expresses @xmath20 at some earlier time @xmath21 as @xmath22 the evolution of @xmath23 is governed by a hamiltonian @xmath24 that depends explicitly on time through @xmath20 . inserting the expansion ( [ phiexpansion ] ) , we have @xmath25 where the first term , say @xmath26 is independent of @xmath21 .
assuming that the state of the @xmath4 field is always close to thermal equilibrium , the time - dependent correction @xmath27 can be treated using the standard methods of linear response theory , which yields the approximation @xmath28\rangle^{\mathrm{eq}}\ ] ] where the expectation values are taken in the equilibrium ensemble associated with @xmath29 .
keeping just the first two terms in ( [ phiexpansion ] ) , we find that @xmath27 is proportional to @xmath30 , and in this way we recover an equation of the form ( [ localeom ] ) . using this or equivalent methods , estimates of the friction coefficient were found in the 1980s for a single self - interacting scalar field @xcite , and more recently ( mainly in the context of warm inflation ) for more complicated models @xcite .
plausible though these calculations seem to be , they rely on an unproven assumption : namely , that the true non - equilibrium dynamics can legitimately be replaced by the linear response of a state of exact thermal equilibrium . it is therefore important to ask whether the same results can be recovered from the slow - evolution limit of the non - equilibrium dynamics .
recently @xcite , we have addressed this question by exhibiting a sequence of approximations through which the exact non - equilibrium dynamics can be reduced to a set of _ local _ , coupled evolution equations for @xmath31 and quantities characterizing the state of the particles into which @xmath2 decays . to recover the local equation of motion ( [ localeom ] ) ,
it is necessary that the solution of these evolution equations should have an adiabatic expansion in powers of derivatives of @xmath31 .
somewhat surprisingly , we found , as reviewed in section [ eveqns ] below , that no such solution exists .
formally , this means that the equation of motion for @xmath31 is intrinsically non - local , and does not have a time - derivative expansion .
in particular , the friction coefficient @xmath18 does not exist .
this formal result indicates that the local equation of motion is never strictly valid , but that does not necessarily mean that it can not serve as a reasonably accurate approximation .
intuitively , one might still expect that ( at least under favourable circumstances ) dissipative effects , which certainly _ are _ a feature of the non - equilibrium evolution , should be fairly well represented by the friction term calculated in linear response theory .
it is this issue that we attempt to address here by numerical methods .
more specifically , one can envisage a further approximation to the non - equilibrium evolution equations in which self - energies are replaced with the values they would have in a state of thermal equilibrium .
then ( see section [ eveqns ] below ) the solution of the resulting equations _ does _ have a time - derivative expansion , and one recovers essentially the friction coefficient obtained from linear response theory . in what follows , we try to assess ( within a particular model ) whether the local equation of motion obtained in this way provides a good approximation to the non - equilibrium dynamics .
section [ eveqns ] describes the non - equilibrium evolution equations and the adiabatic approximation to these equations that we solve numerically . in section [ results ] , we present solutions for the evolution from two different initial states : one is a state in which the adiabatic approximation predicts underdamped motion ( section [ ud ] ) , while in the other it predicts overdamped motion ( section [ od ] ) . in both cases , we find that , although although the non - equilibrium dynamics does yield damping , this damping is not well described by the adiabatic approximation .
a brief account of the numerical methods used to obtain these results is given in section [ numeric ] .
finally , section [ discuss ] is a self - contained summary , during the course of which consider what might account for the failure of the adiabatic approximation in the situation we have studied .
as in @xcite we work with a minkowski - space theory . extending our analysis to a robertson - walker spacetime presents no difficulties of principle , but it introduces complications which seem to have no bearing on the issue we address .
consider , then , a quantum field theory defined by the lagrangian density @xmath32 it should be emphasized at once that we will not attempt to study the exact non - equilibrium dynamics of this model . in fact , we will arbitrarily discard various terms in its equations of motion which , though they may well affect the dynamics significantly , are irrelevant to the question on which we wish to focus .
the first such approximation is to replace the quantum field @xmath33 with its expectation value @xmath31 , where we assume a spatially homogeneous state .
then the equation of motion for @xmath31 is @xmath34 while the dynamics of the remaining quantum field @xmath35 is governed by the lagrangian @xmath36 with @xmath37 .
an approximate treatment of this non - equilibrium dynamics yields , for the expectation value in ( [ phieom ] ) ( which is independent of @xmath38 for a homogeneous state ) @xmath39 , \label{chisq}\ ] ] where @xmath40 and the functions @xmath41 and @xmath42 , which characterize the state of the system of @xmath4 particles , are solutions of the evolution equations @xmath43 + \epsilon\frac{\dot{\omega}_k(t)}{\omega_k(t)}{\mathrm{re}}\,\nu_k(t)\label{dndt}\\ \epsilon\partial_t\nu_k(t)&=&-2i\left[\omega_k(t)-i\gamma_k(t)\right]\nu_k(t)-\alpha_k(t ) + \epsilon\frac{\dot{\omega}_k(t)}{2\omega_k(t)}\left[1 + 2n_k(t)\right].\label{dnudt}\end{aligned}\ ] ] in these equations , @xmath44 and @xmath45 are given by @xmath46 \label{alphadef}\\ \gamma_k(t)&=&\frac{c}{4\pi^2}\int d^3k_1d^3k_2d^3k_3 \frac{\delta(\omega_{k_1}+\omega_{k_2}-\omega_{k_3}-\omega_{k } ) \delta(\bm{k}_1+\bm{k}_2-\bm{k}_3-\bm{k})}{\omega_k\omega_{k_1}\omega_{k_2}\omega_{k_3 } } \nonumber\\ & & \times\left[(1+n_{k_1})(1+n_{k_2})n_{k_3}-n_{k_1}n_{k_2}(1+n_{k_3})\right ] \label{gammadef}\end{aligned}\ ] ] with @xmath47 , while @xmath48 , which has the value @xmath49 , is a formal parameter introduced to facilitate an adiabatic expansion .
the approximations that lead to these equations are described in detail in @xcite .
essentially , they arise as a solution to the dyson - schwinger equations for 2-point functions , via a local _ ansatz _ for self - energies , which are evaluated at 2-loop order in perturbation theory . for a free - field theory , with @xmath50 , which also means @xmath51 , the evolution equations are exact ( see , for example @xcite ) and the functions @xmath41 and @xmath42 can be identified in terms of creation and annihilation operators for @xmath4 particles : @xmath52 with interactions included , ( [ dndt ] ) can be interpreted as a boltzmann equation for number densities @xmath41 ; the quantity @xmath53\label{sdef}\ ] ] is the standard scattering integral associated with two - body elastic scattering , while the source @xmath54 represents particle creation and absorption due to the time - dependent mass @xmath55 .
although the evolution equations ( [ dndt ] ) and ( [ dnudt ] ) are local , their solution is not .
one can attempt to obtain a solution in the form of a time - derivative expansion , by writing @xmath56 expanding in powers of @xmath48 and finally setting @xmath49 .
however , it turns out that no such solution exists , for the following reason . to find the leading correction @xmath57
, one has to solve the integral equation @xmath58 where @xmath59 .
it is straightforward to show that conservation of energy in two - body scattering implies the sum rule @xmath60 .
the right - hand side of ( [ deltaneqn ] ) does not respect this sum rule , so the equation is not self - consistent and the desired solution does not exist .
( for elastic scattering , there is a second sum rule associated with particle - number conservation , which is also not respected by ( [ deltaneqn ] ) ; however , corrections to the boltzmann equation ( [ dndt ] ) from higher orders of perturbation theory would introduce inelastic scattering terms , destroying this second sum rule . ) to the extent that our approximations afford an adequate description of non - equilibrium evolution ( an important caveat , upon which we comment further in section [ discuss ] ) this shows that the equation of motion ( [ phieom ] ) does not have a time - derivative expansion and can not be represented by a local equation such as ( [ localeom ] ) .
however , we can recover a time - derivative expansion by making a further approximation .
the functions @xmath44 and @xmath45 are related to quasiparticle self - energies @xmath61 ( whose exact meanings are given in @xcite ) by @xmath62\label{alphapresc}\\ \gamma_k(t)&=&-\frac{i}{4\omega_k(t)}\left[\sigma^{+}_{k}(t ) -\sigma^{-}_{k}(t)\right].\label{gammapresc}\end{aligned}\ ] ] let us replace these self - energies with constant equilibrium values . in what follows
, we will call this the _ adiabatic approximation_. using this approximation , the time - derivative expansion can be carried through without difficulty .
in particular , @xmath44 and @xmath45 no longer depend on @xmath63 , and the troublesome equation ( [ deltaneqn ] ) does not arise . since @xmath64 is proportional to @xmath30 , so are @xmath63 and @xmath65 , and these contributions to @xmath10 give rise to a friction coefficient @xmath66.\label{etaeq}\ ] ] here , @xmath67 stands for the equilibrium bose - einstein distribution @xmath68 , while @xmath69 is found by substituting this equilibrium distribution in ( [ gammadef ] ) .
this expression agrees , for example , with the result given in @xcite in the limit @xmath70 considered by these authors . roughly speaking
, the second term represents a loss of inflaton energy arising from particle creation , while the first represents a transfer of energy to particles already present .
this agreement is not surprising : the adiabatic approximation we have just introduced is roughly equivalent to the assumptions on which linear response theory is based .
intuitively , it would seem that this should be a fair approximation if @xmath31 does not change too rapidly , and if the state of the @xmath4 particles does not stray too far from an equilibrium distribution .
we now attempt to test this intuition by comparing numerically the non - equilibrium evolution resulting from ( [ dndt ] ) and ( [ dnudt ] ) with that implied by the adiabatic approximation . to simplify matters a little , we observe that the frictional effect we want to study arises from the last two terms in the expression ( [ chisq ] ) for @xmath10 , and will discard the first term .
thus , we study the equation of motion @xmath71 with the truncated expectation value @xmath72 , \label{chisqtr}\ ] ] the evolution of @xmath41 and @xmath42 being governed by ( [ dndt ] ) and ( [ dnudt ] )
. the adiabatic approximation yields the estimate @xmath73 where @xmath18 is given by ( [ etaeq ] ) and the contribution from the leading order terms of the adiabatic expansion is @xmath74\label{deltavpr}\ ] ] with @xmath75 in order to evaluate @xmath76 , we need values for the inverse temperature @xmath77 of the assumed equilibrium state . to this end
, we observe that the evolution equations ( [ phieom2 ] ) , ( [ dndt ] ) and ( [ dnudt ] ) conserve exactly the energy given by @xmath78 we will determine the evolving temperature in the adiabatic approximation by requiring this energy to remain constant . to avoid confusion , it may be worthwhile to reiterate that our goal is to test whether the expectation value ( [ chisqtr ] ) is adequately represented by the adiabatic expression ( [ chisqadiab ] ) .
we will do this by comparing solutions to the non - equilibrium evolution equations ( [ phieom2 ] ) , ( [ chisqtr ] ) , ( [ dndt ] ) and ( [ dnudt ] ) with the adiabatic approximation _ to these equations _ , which consists of the equation of motion @xmath79 with @xmath18 and @xmath80 given by ( [ etaeq ] ) and ( [ deltavpr ] ) respectively , and @xmath77 determined by energy conservation .
the fact that we have arbitrarily discarded several terms does not affect this comparison , though of course it does mean that we are not in a good position to study the true dynamics of the original model ( [ model ] ) .
equally , our approximations are based in part on perturbation theory but the neglected higher - order terms do not affect the comparison , and we will not be concerned with choosing parameter values so as to make these terms small .
we have obtained numerical solutions of the evolution equations presented in the previous section , taking the simplest available potential , namely @xmath81 this section presents two sets of solutions . in the first , the motion of @xmath2 is an underdamped oscillation ; in the second , the motion is predicted _ by the adiabatic approximation _ to be overdamped . in each case , we solve , on the one hand the non - equilibrium equation of motion ( [ phieom2 ] ) together with ( [ dndt ] ) and ( [ dnudt ] ) , and on the other hand the adiabatic approximation to this equation ( [ eomadiab ] ) , starting from identical initial conditions . at the end of the section
, we comment briefly on the numerical methods used .
figure [ fig1 ] shows the motion of @xmath2 obtained in the two approximations , starting from a more or less arbitrary initial state . throughout ,
the units are fixed by taking the mass of @xmath4 particles to be @xmath82 . here , we chose the parameter values @xmath83 , @xmath84 and @xmath85 .
the initial state for @xmath2 was @xmath86 , @xmath87 and the initial state of the @xmath4 particles was a state of thermal equilibrium with inverse temperature @xmath88 . with these values ,
the energy of the particles is about 86% of the total energy , so one might expect a significant transfer of further energy to these particles as well as a frictional effect from particle creation .
the two superimposed oscillations in figure [ fig1 ] are not easily distinguished , so we have plotted envelopes for the two curves : the solid envelope indicates the solution to the non - equilibrium evolution equations , while the dashed one corresponds to the adiabatic approximation . at first sight
, there seems to be a similar rate of damping in each case , but there are significant differences .
the friction coefficient of the adiabatic approximation given in ( [ etaeq ] ) is a rather complicated function of @xmath2 , but suppose that it can be written as @xmath89 , where @xmath90 varies relatively slowly with @xmath2 .
this suggests that large - amplitude oscillations should be strongly damped , while at small amplitudes damping would be ineffective .
this behaviour is clearly apparent from figure [ fig1 ] .
however , the solution of the full evolution equations shows a different tendency : damping is smaller at large amplitudes than that predicted by the adiabatic approximation , and it does not disappear at small amplitudes .
thus , in this situation , although the non - equilibrium evolution equations do yield a damped motion , the damping is not well represented by the friction term as estimated from the adiabatic approximation .
this is not necessarily unexpected , since both approximations yield fairly rapid oscillations , and the adiabatic approximation might not be expected to work well .
we therefore turn to a different initial state which , according to the adiabatic approximation , ought to result in overdamped motion and slow evolution .
suppose , as above , that the adiabatic friction coefficient can be represented as @xmath91 and also that the adiabatic correction to the effective potential ( [ deltavpr ] ) is @xmath2 times a slowly varying function .
then the adiabatic equation of motion ( [ eomadiab ] ) becomes @xmath92 where @xmath93 and @xmath94 are , at least over some period of time , approximately constant .
if @xmath95 can be neglected , then this equation has the overdamped solution @xmath96^{1/2},\label{odsolution}\ ] ] with a `` terminal velocity '' @xmath97 under these circumstances , we have @xmath98 , so overdamped motion should occur if @xmath99 , or @xmath100 intuitively , the adiabatic approximation ought to be good if , in addition to this condition for neglecting @xmath95 , the state of the @xmath4 particles thermalizes rapidly . according to the boltzmann - like equation ( [ dndt ] ) , the mode @xmath67 of the particle distribution relaxes towards equilibrium with a characteristic relaxation time @xmath101 . taking the @xmath102 mode as representative , one can guess at a rough criterion for efficient thermalization , namely that the fractional change in the effective particle mass @xmath103 over a time @xmath104 should be small .
that is , @xmath105 or @xmath106 ideally , we would like to study a state in which both conditions ( [ odcondition ] ) and ( [ thermalization ] ) are well satisfied , but this proves to be quite difficult .
in particular , the second condition seems hard to achieve . using ( [ vterm ] ) , it can be rewritten as @xmath107 and the thermal effects that make @xmath93 and @xmath69 large also tend to produce large values of @xmath108 .
the dependence of @xmath18 ( equation ( [ etaeq ] ) ) on the parameters at our disposal is so complicated as to make a systematic survey of the parameter space impractical . by trial and error ,
we have chosen a set of values which yield @xmath109 this means that _ according to the adiabatic approximation _
, the system is strongly overdamped , but thermalization is only moderately efficient .
a careful , but by no means exhaustive , exploration suggests that this situation can not be much improved within the model studied here .
the parameter values leading to this state are @xmath110 , @xmath111 , @xmath112 and @xmath113 .
the terminal velocity ( [ vterm ] ) turns out to be @xmath114 . taking these initial conditions
, we find for @xmath31 the motion depicted in figure [ fig2 ] .
motion of @xmath31 as calculated from the non - equilibrium evolution equations and via the adiabatic approximation .
as determined from the adiabatic approximation , the initial conditions should lead to overdamped motion . ] evidently , the motion obtained from the adiabatic approximation ( the thinner curve ) is indeed overdamped at early times .
in fact the solution agrees well with ( [ odsolution ] ) for times up to about @xmath115 .
but as @xmath2 becomes smaller , the condition ( [ odcondition ] ) for overdamping no longer holds and , as in figure [ fig1 ] , the motion at small amplitudes is essentially undamped .
the solution of the non - equilibrium evolution equations ( the thicker curve in figure [ fig2 ] ) is strikingly different : although there is significant damping , the motion is far from being overdamped .
we observe , though , that the average rate of energy loss is quite similar , as indicated in figure [ fig3 ] , where we plot the energy that exists in the form of particles ( the last term in ( [ energy ] ) ) as a fraction of the conserved total energy . in fact
, this figure shows that the overall conversion of inflaton energy into particles is somewhat more efficient in the non - equilibrium evolution than is suggested by the adiabatic approximation .
energy of particles as a fraction of the conserved total energy . ] of course , the question arises , what accounts for the large qualitative difference between the non - equilibrium evolution and that obtained from the adiabatic approximation ?
a simple answer is that , as shown in section [ eveqns ] , no adiabatic expansion of the non - equilibrium evolution equations actually exists , and our numerical results merely confirm that the adiabatic approximation is wrong .
this may perhaps be the appropriate answer , but it needs some further discussion , because we also showed in section [ eveqns ] that the adiabatic expansion _ does _
exist if we take the extra step of replacing self - energies with their equilibrium values , and one might think that this step is relatively innocuous if the state is not too far from equilibrium .
consider first the initial state , which was chosen to be a state of exact thermal equilibrium . according to the adiabatic approximation ,
the criterion for neglecting @xmath95 is very well satisfied ( equation ( [ initcriteria ] ) ) , and the motion should be strongly overdamped .
however , we see from figure [ fig2 ] that @xmath116 as calculated from the non - equilibrium evolution equations is very large .
the reason for this is not hard to discover . in a state of exact thermal equilibrium ,
the term @xmath117 in the equation of motion ( [ phieom2 ] ) is precisely the quantity @xmath80 that appears in the adiabatic equation ( [ eomadiab ] ) .
thus , while the original equation of motion asserts that @xmath118 is equal to @xmath119 , the adiabatic approximation asserts that @xmath120 , the value of @xmath95 being very small . clearly , _
the adiabatic approximation gives a wrong answer for a state of exact thermal equilibrium_. in fact , the informal arguments given in @xcite and reviewed in @xcite show that the frictional effects arise precisely from small departures from equilibrium .
weighted particle distribution @xmath121 at time @xmath122 ( solid curve ) and the same quantity calculated from the equilibrium distribution for the same total particle energy ( broken curve ) . ]
figure [ fig4 ] illustrates the departure from thermal equilibrium at times near @xmath123 , corresponding to the first negative peak in figure [ fig2 ] , where about one half - cycle of oscillation has been completed .
the quantity plotted is @xmath121 , whose integral gives the total energy in particles .
the solid curve uses the distribution @xmath41 generated by the non - equilibrium evolution , while the broken curve uses the equilibrium distribution @xmath124 with the temperature @xmath125 adjusted to give the same total energy .
( this is quite independent of the equilibrium distribution @xmath126 generated by the adiabatic evolution . ) clearly , these two distributions are different , but the difference does not greatly affect the time evolution .
for example , we show in figure [ fig5 ] the the quasiparticle width @xmath69 calculated from the particle distributions of figure [ fig4 ] . ]
quasiparticle width @xmath69 , which appears in the evolution equations ( [ dndt ] ) and ( [ dnudt ] ) , calculated from @xmath41 and @xmath127 , and the difference is seen to be marginal .
we can use the effective value of @xmath77 calculated as above , together with the non - equilibrium values of @xmath67 and @xmath69 in the expressions ( [ etaeq ] ) and ( [ deltavpr ] ) to find an effective , non - equilibrium value for the ratio @xmath128 which , according to ( [ odcondition ] ) , gives a criterion for overdamping at times near @xmath129 .
the result is that @xmath130 . thus , the criterion for overdamping ( as obtained from the adiabatic approximation ) is well satisfied by the non - equilibrium state at @xmath123 , just as it was by the initial equilibrium state .
the main point of the preceding discussion is this . in order to obtain the adiabatic approximation
, we had to replace non - equilibrium self - energies @xmath131 $ ] with the values @xmath132 $ ] associated with a state of thermal equilibrium . in the initial state at @xmath133
these two self energies are identical , and in the state illustrated in figure [ fig5 ] the difference is quite small . clearly ,
this difference does not account for the drastic failure of the adiabatic approximation in the situation we have examined .
the difference between the two evolutions in figure [ fig2 ] ( and , for that matter , those shown in figure [ fig1 ] ) indicates that the expectation value @xmath134 is not well represented as the sum of the two terms in ( [ chisqadiab ] ) , regardless of whether the coefficients @xmath80 and @xmath18 are calculated using the non - equilibrium state or an effective equilibrium state .
thus , although the rate of thermalization may well be an important factor , its importance does not lie in the relatively small effect that inefficient thermalization would have on the values of these coefficients .
having made this point , we defer further discussion to section [ discuss ] , but we observe here that thermalization in the situation we have studied is rather less efficient than one might infer from the value @xmath135 for the ratio in ( [ thermalization ] ) . in the initial state , the relaxation time @xmath136 has values @xmath137 for all the modes used in the calculation , and this is longer than the whole time for which the evolution was studied . from the point of view of the adiabatic approximation ,
this means that although thermalization might have been expected to be moderately efficient by comparison with the initially slow evolution , it is much less efficient by comparison with the more rapid evolution at later times . in the non - equilibrium evolution , there are quite large departures from equilibrium when @xmath2 changes rapidly .
for example , we show in figure [ fig6 ] the particle distribution ( again weighted by @xmath138 ) at the time @xmath139 where @xmath2 first becomes very small . weighted particle distribution at @xmath140 , compared with the equilibrium distribution for the same energy .
] this figure illustrates the fact that the particles created by the source ( [ jdef ] ) have a non - thermal spectrum weighted towards high energies : it falls off roughly as @xmath141 compared with the exponential fall - off of the thermal distribution .
it is on account of this difference that the integral equation ( [ deltaneqn ] ) fails to respect the energy sum rule , and thus to be self - consistent . comparing figures [ fig4 ] and [ fig6 ]
, it appears that , although the created particles are not thermalized instantaneously , a significant amount of thermalization has taken place during the quarter - cycle between @xmath140 and @xmath122 .
this appearance is somewhat illusory , however , as we explain in section [ discuss ] .
the calculations reported above were performed using modest computing resources ( a pc ) , and a moderate numerical accuracy seems to be adequate for the issues we seek to address .
the most demanding aspect of the calculations is the solution of the evolution equations ( [ dndt ] ) and ( [ dnudt ] ) . to deal with these , we computed the evolution of @xmath142 modes ( that is , @xmath142 values of the momentum @xmath143 ) for the distribution functions @xmath67 and @xmath144 , with an upper cutoff @xmath145 . of the various integrals that are needed ,
the energy integral ( [ energy ] ) attaches the largest weight to high - momentum modes and the chosen cutoff is large enough to make this integral reasonably accurate , as illustrated by the integrands in figures [ fig4 ] and [ fig6 ] .
since the calculations are quite lengthy ( the data needed to produce figure [ fig2 ] were obtained from 3 weeks continuous running ) we did not attempt to make the entire process converge by increasing @xmath146 and @xmath142 systematically .
however , the results presented here , using @xmath147 , which adequately resolves the structure of the distributions generated , differ negligibly from those obtained using @xmath148 , and seem to be insensitive to the cutoff @xmath146 .
it is apparent from ( [ dnudt ] ) that the functions @xmath42 oscillate with frequencies roughly equal to @xmath149 .
the basic time stepper used was a semi - implicit method , which is stable for these oscillations , with a fixed time step @xmath150 about one - tenth of the period of the fastest oscillation .
however , computation of the scattering integrals ( [ alphadef ] ) and ( [ gammadef ] ) is very time - consuming ( even after analytical reduction to triple integrals ) , so @xmath151 and @xmath69 , which change quite slowly relative to the fast oscillations , were updated at longer time intervals @xmath152 , using a predictor - corrector method that typically converged after two or three iterations .
this larger time interval @xmath152 was assigned values between @xmath153 and @xmath154 , according to the rate of change of @xmath69 ( see figure [ fig7 ] ) .
the values of @xmath67 needed in these integrals were obtained by interpolating the @xmath142 tabulated values with a cubic spline .
variation of @xmath2 , @xmath155 and @xmath156 over a typical cycle of oscillation .
data points are separated by the time interval @xmath152 explained in the text . ]
the quality of the data obtained by these methods is illustrated in figure [ fig7 ] , which shows the variation over a typical cycle of @xmath2 and the @xmath102 modes of @xmath69 and @xmath67 .
the sharp peaks in @xmath69 appear in all modes , and are largely due to the factor @xmath157 in ( [ gammapresc ] ) .
the basic operations of evaluating integrals , and so on , were carried out with a tolerance of @xmath158 or better .
the energy ( [ energy ] ) , which formally is exactly conserved , exhibits small variations , but is constant to within about @xmath159 , which we estimate as the overall accuracy of the calculation .
according to quantum field theory , the equation of motion for the expectation value of a scalar field is a complicated , non - local integro - differential equation . in applications to inflationary cosmology , it is typically assumed that this equation of motion can be replaced by a local equation similar to ( [ localeom ] ) .
if it is further assumed that the state of the quantum system is fairly close to thermal equilibrium , then it seems plausible that one should be able to apply the methods of linear response theory , which yield quantitative estimates of the friction coefficient @xmath18 and the effective potential @xmath160 .
here and in @xcite we have tried to address the question whether the local equation of motion obtained in this way adequately approximates the actual non - equilibrium evolution , at least when this evolution is sufficiently slow .
the indications are that it does not , but the treatment we have given can not be considered as definitive , and we wish to set out clearly the issues that are involved .
the first issue is that we have treated the non - equilibrium evolution on the basis of an _ approximate _ set of local equations ( [ phieom ] ) , ( [ chisq ] ) , ( [ dndt ] ) and ( [ dnudt ] ) . as described in detail in @xcite , they arise from a local _ ansatz _ for quasiparticle self - energies , and it is not easy to assess how good an approximation this is . for the purposes of estimating the non - equilibrium behaviour numerically ,
other approximation schemes have recently become available ( see , for example , @xcite ) which are in some respects superior to this one .
however for the purpose of deriving a local effective equation of motion , approximations more or less equivalent to ours would seem to be inevitable . to be precise , the question we address here is whether the fully local equation of motion ( [ eomadiab ] ) as derived from linear response theory yields a good approximation to the non - equilibrium behaviour as described by our evolution equations . if it does not ( as we seem to find ) , then it is also unlikely to provide a good approximation to the true non - equilibrium dynamics .
the phrase `` fully local '' above means the following .
the evolution equations ( [ phieom ] ) , ( [ dndt ] ) and ( [ dnudt ] ) are local , in the sense that they involve @xmath31 , @xmath41 and @xmath42 together with their time derivatives , all evaluated at the same time @xmath13 : they describe a markovian process . in principle
, one can envisage solving ( [ dndt ] ) and ( [ dnudt ] ) for @xmath41 and @xmath42 as functionals of @xmath31 .
these solutions are non - local : they depend on the entire history of @xmath31 .
so , substituting these solutions into the remaining equation ( [ phieom ] ) , we have a non - local equation of motion for @xmath2 .
this non - local equation might be approximated by the `` fully local '' equation ( [ eomadiab ] ) if the quantity @xmath6 has a time - derivative expansion whose first two terms are @xmath80 and @xmath1 . although we can not find the complete functionals @xmath41 , and @xmath42 , we can attempt to find their time - derivative expansions . as explained in section [ eveqns ] , these expansions do not exist .
however , if we make the further approximation of calculating self - energies from an equilibrium particle distribution rather than the non - equilibrium distribution , we _ can _ obtain a time derivative expansion , which essentially reproduces the results of linear response theory .
this is what we have referred to as the adiabatic approximation . in section [ results ] ,
we compared numerically the non - equilibrium evolution with that resulting from the adiabatic approximation , and found very large differences .
we also saw that , for the model studied here , thermalization is not very fast compared with the motion of @xmath2 .
intuitively , it seems that this poor thermalization might account for the failure of the adiabatic approximation .
the second issue we want to discuss is , therefore , that of thermalization . as indicated in section [ od ] ,
the relaxation time @xmath101 is roughly equal to @xmath161 in the initial state from which figure [ fig2 ] was generated .
on the basis of the adiabatic approximation , which indicates strong overdamping , one would perhaps expect the non - equilibrium evolution to exhibit a period of overdamped motion , albeit with significant corrections due to a modest thermalization rate .
what we actually found was oscillatory motion with a period much shorter than @xmath162 . over one quarter - period of this motion ,
the particle distribution changes from that shown in figure [ fig4 ] to that shown in figure [ fig6 ] . from these figures , it would seem that significant thermalization has taken place over a period of time much shorter than @xmath162
. however , figure [ fig7 ] suggests a different explanation . from that figure
, we see that particles with a non - thermal distribution are radiated over a short period of time as @xmath163 decreases towards 0 , and that these particles , rather than being thermalized , are quickly reabsorbed as @xmath163 subsequently increases .
so thermalization is indeed slow on the time scale of the non - equilibrium motion .
we can not altogether rule out the possibility that it is this slow thermalization that accounts for the failure of the adiabatic approximation in the situation we have studied .
however , we do not think that this is the correct explanation .
one reason is that , as we have just remarked , consideration of the adiabatic approximation alone would suggest that thermalization , while modest , is not so slow as to produce the gross discrepancy apparent in figure [ fig2 ] .
a more compelling reason is that , as discussed in section [ od ] , the non - equilibrium motion is not well represented by the adiabatic equation ( [ eomadiab ] ) even at those times when the state is close to thermal equilibrium .
this strongly indicates that the step of replacing , say , @xmath164 $ ] with @xmath165 $ ] is wrong in principle and not just because of the quantitative discrepancy which , as seen in figure [ fig5 ] , may be quite small . in more detail , suppose that the distributions @xmath41 and @xmath42 are indeed close to thermal equilibrium
. then it should be a good approximation to linearize the evolution equations ( [ dndt ] ) and ( [ dnudt ] ) , using @xmath166 and @xmath167 .
in particular , the linearized version of ( [ dndt ] ) is @xmath168 .
\label{linearbe}\ ] ] unlike ( [ deltaneqn ] ) , this equation is self - consistent ; the energy sum rule merely gives an expression for the rate of change of the quantity @xmath169 .
( this equation is not self - contained , however . to solve it , we would need a prescription , such as the conservation of energy , for determining the time - dependent temperature that specifies @xmath170 . as a matter of fact ,
the scattering integral @xmath171 vanishes when @xmath127 has an arbitrary chemical potential . in the adiabatic calculations of section [ results ] ,
we did not allow for the possibility of a non - zero chemical potential because there is no conserved particle number that could be used to determine its value . )
the important point here is that the linearized equation ( [ linearbe ] ) is a _
differential equation_. its solution is again a non - local functional of @xmath2 , which will not reproduce the adiabatic equation ( [ eomadiab ] ) , regardless of how small @xmath57 might be . to do that , we need
not just a linear approximation , but a time - derivative expansion which , as we have emphasized , is not self - consistent . in summary
, our numerical study indicates that that the non - equilibrium evolution of @xmath2 is not well described by a local equation of motion with the friction term estimated by linear response theory .
it is possible that an adiabatic approximation would work better for a model in which thermalization is more efficient , and we plan to explore this in future work .
however , for the reasons just discussed , we do not think that poor thermalization is the principal cause of the large discrepancies we observed . | the equation of motion for the expectation value of a scalar quantum field does not have the local form that is commonly assumed in studies of inflationary cosmology .
we have recently argued that the true , temporally non - local equation of motion does not possess a time - derivative expansion and that the conversion of inflaton energy into particles is not , in principle , described by the friction term estimated from linear response theory .
here , we use numerical methods to investigate whether this obstacle to deriving a local equation of motion is purely formal , or of some quantitative importance . using a simple scalar - field model , we find that , although the non - equilibrium evolution can exhibit significant damping , this damping is not well described by the local equation of motion obtained from linear response theory .
it is possible that linear response theory does not apply to the situation we study only because thermalization turns out to be slow , but we argue that that the large discrepancies we observe indicate a failure of the local approximation at a more fundamental level . |
quasars rank among the most luminous objects in the universe and are believed to be powered by smbhs ( e.g. , @xcite ) .
they constrain the formation and evolution of galaxies and smbhs throughout cosmic time .
the similarity between the cosmic star formation history ( e.g. , @xcite ) and the evolution of quasar abundances ( e.g. , @xcite ) suggests an intriguing link between galaxy formation and black hole growth .
this is strengthened by tight correlations measured locally between the masses of the black holes and the global properties of the spheroid components of their hosts , such as their luminosities and masses ( @xcite ) , light concentration ( @xcite ) , and velocity dispersions ( @xcite ) .
distant , highly luminous quasars are important cosmological probes for studying the first generation of galaxies , the star formation history and metal enrichment in the early universe , the growth of the first supermassive black holes ( smbhs ) , the role of feedback from quasars and black holes in galaxy evolution , and the epoch of reionization . the sloan digital sky survey ( sdss , @xcite ) has contributed significantly to the discovery of high redshift quasars .
currently , there are over 1000 quasars known at @xmath11 and 12 at @xmath12 ( @xcite ) .
as reviewed by @xcite , quasars at @xmath13 are characterized by : ( 1 ) a low space density ( @xmath14 comoving ) ; ( 2 ) high luminosities ( absolute luminosity at rest - frame @xmath15 ) , believed to be powered by smbhs of @xmath1 ; ( 3 ) gunn - peterson absorption troughs @xcite in their spectra , which place these quasars at the end of the epoch of reionization ( e.g. , @xcite ) ; and ( 4 ) a lack of evolution in the spectral energy distribution compared to lower - shift counterparts ( e.g. , @xcite ) , which demonstrates the existence of `` mature '' quasars at early times and comparable metal enrichment in quasars at all cosmic epochs .
the most distant quasar known , sdss j1148 + 5251 ( hereafter j1148 + 5251 ) , was discovered by sdss at z = 6.42 @xcite .
it is extremely bright optically with @xmath16 = -27.8 , and deep imaging surveys in both optical and radio ( @xcite ) show no sign of gravitational lensing or other companions at the same redshift in the vicinity . over the past few years
, this quasar has been extensively studied at many wavelengths . near - infrared observations by @xcite and @xcite
imply a bolometric luminosity of @xmath17 powered by accretion onto a smbh of mass 15 @xmath18 .
radio observations by @xcite and @xcite suggest that the host is a hyper - luminous far - infrared ( fir ) galaxy , with @xmath19 , and these authors estimate a star formation rate of @xmath20 by assuming most of the fir luminosity comes from young stars .
emission from carbon monoxide ( co ) has been detected ( @xcite ) corresponding to a mass of @xmath21 .
dust has been seen by several groups ( e.g. , @xcite ) with an estimated mass of @xmath22 . in particular , _ spitzer _ observations by @xcite and @xcite indicate that the dust is heated by an active galactic nucleus ( agn ) .
furthermore , the detection of iron by @xcite , the carbon fine structure line [ cii ] by @xcite and excess oi absorption by @xcite indicate near - solar metallicity in the quasar host .
these observations raise many fundamental questions for models of quasar and galaxy formation : where do such high - redshift , luminous quasars originate ? how do they form ?
what are the mechanisms and physical conditions for smbh growth ? and
, do these quasar hosts obey the same smbh
host correlations as observed in the local universe ? interpretations of various observations of j1148 + 5251 have painted a rather conflicting picture for the formation site of the quasar halo and the smbh host relationship .
the low abundance of these quasars leads to the view that they are hosted by massive halos ( @xmath23 ) in the rarest density peaks of the dark matter distribution @xcite .
however , it has been argued , based on the lack of companion galaxies in the field , that this quasar may reside in a much lower mass halo in a less overdense region ( @xcite ) .
moreover , @xcite suggest , based on the dynamical mass estimate from co measurements , that j1148 + 5251contains a small stellar spheroid , and that the smbh may have largely formed before the host galaxy .
however , the detection of metal lines ( @xcite ) , along with dust ( @xcite ) , indicates that the interstellar medium ( ism ) of j1148 + 5251 was significantly enriched by heavy elements produced through massive star formation at rates of @xmath24 occurring as early as @xmath25 , and that large stellar bulges form before accreting smbhs undergo luminous quasar activity . in an expanding universe that is dominated by cold dark matter and is accelerated by dark energy , the @xmath2-cold dark matter ( @xmath2cdm )
cosmology , the leading theoretical model for structure formation , assumes that structure grows from weak density fluctuations amplified by gravity , with small objects collapsing first and subsequently merging to form progressively more massive ones , a process known as `` hierarchical assembly '' ( e.g. , see @xcite for a review ) .
the formation of galaxies and quasars is therefore determined by the abundance of dark matter halos ; i.e. , the number density of halos as a function of mass and redshift .
the most widely used analytic model for the halo mass function was first developed by @xcite ( hereafter ps ) , which is based on gaussian density perturbations , linear gravitational growth , and spherical collapse of dark matter . using the ps mass functions , @xcite studied the abundance of rare objects , such as luminous quasars at high redshifts .
these authors predicted a sharp `` cut - off '' of the quasar population at @xmath26 .
however , while the initial , linear growth of density perturbations can be calculated analytically , the gravitational collapse and subsequent hierarchical build - up of structure is a highly nonlinear process that can be followed only through numerical simulations .
it has been shown by previous numerical studies ( e.g. , @xcite ) , and more recently by the state - of - the - art _ millennium simulation _ by @xcite , that the ps formula underestimates the abundance of high - mass halos by up to an order of magnitude .
therefore , whether or not rare quasars such as j1148 + 5251 can form in the @xmath2cdm cosmology remains an open question and an important test of the theory . to date
, only a limited number of analytical or semi - analytic models have addressed the early formation of a @xmath27 smbh at @xmath0 ( @xcite ) .
these approaches use merger trees of dark matter halos generated using the ps theory , and assume a black hole accretion rate at or above the eddington limit .
however , as discussed above , the ps based approach may be inaccurate .
moreover , it is not clear whether sufficient physical conditions for such large accretion rates exist in quasar systems as the hydrodynamics of the large - scale gas flow and feedback from black holes have not been incorporated in earlier modeling .
it is believed that the growth of smbhs is closely linked to galaxy formation ( e.g. , @xcite ) , and that the growth is self - regulated by feedback ( e.g. , @xcite ) .
remarkably , self - regulated models with smbh feedback in the form of thermal energy coupled to the ambient gas have been demonstrated to successfully reproduce many observations of galaxies , including the @xmath28 relation ( @xcite ) , galaxy colors ( @xcite ) , x - ray gas emission @xcite , elliptical kinematics @xcite and the fundamental plane @xcite , quasar properties ( @xcite ) , luminosity functions ( @xcite ) , and populations ( @xcite ) , and the luminosity function of low - level agn @xcite .
furthermore , these simulations of binary mergers identify a plausible , merger - driven formation mechanism for massive black holes and luminous quasars ( e.g. , @xcite ) . here , we present a model that accounts for the smbh growth , quasar activity and host galaxy properties of the most distant quasar known . in our scenario ,
the quasar and its host galaxy form in a massive halo that originates from a rare density peak in the standard @xmath2cdm paradigm , and they grow hierarchically through multiple gas - rich mergers , supporting an average star formation rate of @xmath24 that peaks at @xmath29 . once the progenitors undergo sufficient dynamical friction to coalesce , the multiple smbhs from the progenitor galaxies merge and exponentially increase their mass and feedback energy via accretion . at @xmath30 when the smbh mass exceeds @xmath31 , black hole accretion drives a sufficiently energetic wind to clear obscuring material from the central regions of the system and powers an optically luminous quasar similar to j1148 + 5251 . we have devised a set of novel multi - scale simulations , which include cosmological n - body calculations on large scales and hydrodynamic simulations of galaxy mergers on galactic scales , coupled with the self - regulated growth of smbhs , enabling us to follow galaxy assembly and quasar formation at @xmath0 .
this paper is organized as follows . in 2 , we describe our computational methods and models , which includes a set of large scale cosmological n - body simulations , and hydrodynamical galaxy mergers along the merging history of the quasar halo . in 3 , we present the formation and evolution of the quasar and its host galaxy , including the assembly of the galaxy progenitors , star formation , and smbh growth , as well as the smbh host correlations , and properties of the quasar such as luminosities and lifetimes . we discuss feedback from starburst - driven winds , quasar abundances for cosmological models with different parameters , the implication of black hole mergers , and galaxies in the epoch of reionization in 4 , and summarize in 5 .
rare , high - redshift quasars originate in highly overdense regions in the initial dark matter density distribution and grow through hierarchical mergers , as predicted by the @xmath2cdm theory . simulations of high - redshift quasar formation must model a large cosmological volume to accommodate the low abundance of this population , have a large dynamic range to follow the hierarchical build - up of the quasar hosts , and include the hydrodynamics of the gas flows in galaxy mergers . the cutting - edge _ millennium simulation _ by @xcite follows structure formation in a box with side length of @xmath32 using @xmath33 dark matter particles .
it reproduces the large - scale galaxy distribution as observed @xcite , and identifies an early quasar halo candidate at @xmath34 which ends up in the richest cluster at the present day .
however , even such a large dynamic range still falls short of being able to follow the formation and evolution of the rarest quasars observed at the highest redshifts .
moreover , in order to address the properties of quasars and host galaxies , gasdynamics and physical processes related to star formation and black hole growth must be included . to satisfy these requirements
, we have performed a set of novel multi - scale simulations that enable us to resolve individual mergers on galactic scales and retain the context of large - scale structure formation , as well as the evolution of black holes and stars .
first , we perform a coarse dark matter simulation in a volume of @xmath35 designed to accommodate the low number density of @xmath36 quasars . the largest halo at @xmath37 , within which the descendants of early , luminous quasars are assumed to reside @xcite ,
is then selected for resimulation with higher resolution using a multi - grid zoom - in technique developed by @xcite .
the merging history of the largest halo at @xmath0 , which has reached a mass of @xmath38 through 7 major ( mass ratio @xmath39 5:1 ) mergers between redshifts 14.4 and 6.5 , is extracted .
these major mergers are again re - simulated hydrodynamically using galaxy models scaled appropriately for redshift @xcite and adjusted to account for mass accretion through minor mergers .
the simulations include prescriptions for star formation @xcite , and for smbh growth and feedback ( @xcite ) , as described below .
our multi - scale simulations were performed using the parallel , n - body / smoothed particle hydrodynamics ( sph ) code gadget2 developed by @xcite that is well tested in a wide range of applications from large scale structure formation to star formation . for the computation of gravitational forces , the code uses the `` treepm '' method @xcite that combines a `` tree '' algorithm @xcite for short - range forces and a fourier transform particle - mesh method @xcite for long - range forces .
the pm - method works efficiently in large scale cosmological simulations , while the tree - method provides accurate forces for the large dynamic range of galaxy merger simulations .
gadget2 implements the entropy - conserving formulation of sph @xcite with adaptive particle smoothing , as in @xcite .
radiative cooling and heating processes are calculated assuming collisional ionization equilibrium @xcite .
star formation is modeled in a multi - phase ism , with a rate that follows the schmidt - kennicutt law ( @xcite ) .
feedback from supernovae is captured through a multi - phase model of the ism by an effective equation of state for star - forming gas @xcite .
a prescription for supermassive black hole growth and feedback is also included , where black holes are represented by collisionless `` sink '' particles that interact gravitationally with other components and accrete gas from their surroundings .
the accretion rate is estimated from the local gas density and sound speed using a spherical bondi - hoyle @xcite model that is limited by the eddington rate .
feedback from black hole accretion is modeled as thermal energy injected into surrounding gas , as described in @xcite and @xcite .
the simulations presented in this paper adopt the @xmath2cdm model with cosmological parameters chosen according to the first year wilkinson microwave anisotropy probe data ( wmap1 , @xcite ) , ( @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45)= ( 0.3 , 0.04 , 0.7 , 0.7 , 1 , 0.9 ) . here , @xmath40 is the total matter density in units of the critical density for closure , @xmath46 .
similarly , @xmath41 and @xmath47 denote the densities of baryons and dark energy at the present day .
the hubble constant is parameterized as @xmath48 , while @xmath45 is the root - mean - squared linear mass fluctuation within a sphere of radius @xmath49 extrapolated to @xmath37 .
we have also done the same cosmological simulations with wmap third year results ( wmap3 , @xcite ) , ( @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45)= ( 0.236 , 0.042 , 0.759 , 0.732 , 0.95 , 0.74 ) for comparison .
the quasars at @xmath0 have an extremely low comoving space density , @xmath50 , and are believed to reside in massive dark matter halos with @xmath51 @xcite .
cosmological simulations of quasar formation must therefore model a volume of @xmath52 to account for the rarity of such objects .
however , in order to resolve a @xmath53 halo at @xmath0 in a cosmological simulation with uniform resolution , a dark matter particle mass at least as small as @xmath54 and particle numbers of @xmath55 are required .
tracking the merger history of such halos requires @xmath56 orders of magnitude higher resolution and would be computationally prohibitive with standard techniques .
we achieve the mass resolution requirements for the merger history of a @xmath53 halo at @xmath57 by means of a two - step re - simulation .
first , coarse dark matter cosmological simulations are performed to identify a candidate halo for the quasar host .
a cubic volume @xmath58 on a side is simulated with @xmath59 particles , achieving mass and force resolutions of @xmath60 and @xmath61 ( comoving ) , respectively . to generate the initial conditions
, we use the boltzmann code cmbfast by @xcite to compute a linear theory power spectrum for our chosen cosmology .
a random realization of the power spectrum is constructed in fourier space , sampling modes in a sphere up to the nyquist frequency of the mesh .
the particle distribution is evolved forward in time to @xmath37 from its initial displacement at @xmath62 determined using the zeldovich approximation . at the end of the simulation
, halos are identified using a `` friends - of - friends '' ( fof ) group - finding algorithm @xcite with a fixed comoving linking length equal to 0.2 times the mean dark matter interparticle separation and a minimum of 32 particles per group @xcite .
the mean overdensity of the groups corresponds approximately to the expected density of virialized halos @xcite . from the more than 126000 groups identified in the @xmath63 volume at @xmath37 the largest halo with @xmath64
is selected as a candidate halo for modeling the formation of a quasar at @xmath65 . a multi - grid technique developed by @xcite and @xcite
is used to `` zoom in '' with high resolution on the selected halo region which has an effective side length of @xmath66 .
large - scale tidal forces are captured by binning exterior particles into cells according to their distance from the high - resolution region .
to ensure proper treatment of small - scale structure , the initial displacements of the high resolution particles are calculated assuming a higher initial redshift of @xmath67 and normalized to @xmath45 at @xmath37 . the re - simulation uses @xmath68 particles , with @xmath69 particles inside the high - resolution region . with this technique ,
the mass resolution increases by almost four orders of magnitude to @xmath70 while the spatial resolution reaches @xmath71 .
figure [ figmap ] shows snapshots of both the coarse and high - resolution zoom - in runs that locate the quasar halo candidate . in the coarse run ,
the `` cosmic web '' is clearly seen , although the distribution appears nearly homogeneous on such large scales . in the zoom - in run ,
filamentary structures are prominent .
dark matter collapses along the filaments , and the largest halo forms in the deepest potential wells at the intersections of the filaments .
the high resolution of the zoom - in run enables the identification of more halos with lower masses both at @xmath37 and at high redshifts as early as @xmath72 , which is sufficient to identify the halo progenitors of the candidate quasar at @xmath0 .
it appears that the halo progenitor of the largest one at the present day is also the most massive halo at @xmath0 , when it reaches a mass of @xmath73 , making it a plausible candidate for hosting a rare @xmath74 quasar .
the impact that variations in the cosmological parameters can have on large - scale structure formation can be understood from the theoretical mass function of halos , as derived by @xcite and later developed by @xcite .
the comoving number density @xmath75 of halos of mass between @xmath76 and @xmath77 can be described as , @xmath78 \ , , \label{eq_mf}\ ] ] where @xmath79 is the local mean mass density , @xmath80 is the critical density of collapse at redshift @xmath81 linearly extrapolated to the present day , and @xmath82 is the mass variance , which is a function of the power spectrum @xmath83 with wavenumber k and the window function @xmath84 , @xmath85 .
the abundance of halos depends on the two functions @xmath82 and @xmath86 , each of which involves the cosmological parameters , in particular @xmath45 , @xmath87 and @xmath88 .
these parameters determine the formation epoch of a halo and its mass . the recently - released third year wmap3 results @xcite have lower values of @xmath45 , @xmath89 and @xmath87 , compared to wmap1 @xcite .
the smaller @xmath45 from wmap3 would lower the amplitude of the power spectrum , which in turn reduces @xmath82 .
furthermore , a smaller @xmath87 would reduce @xmath80 and hence delay halo formation .
so , compared to wmap1 , at a given redshift the wmap3 parameters would yield a lower abundance of halos with mass @xmath90 , where @xmath91 is the halo mass corresponding to the characteristic luminosity in the schechter luminosity function for galaxies @xcite ; while for @xmath92 , it predicts a slightly larger halo abundance . to test the sensitivity of our model to the new wmap results
, we have performed the same set of cosmological simulations with parameters from the wmap3 measurements @xcite .
we find that indeed the changes implied by the new parameters primarily affect the formation time and the mass of the candidate quasar halo .
for the same random phases in the initial conditions , the location of the most massive halo at @xmath37 remains the same in both the wmap1 and wmap3 runs , except that its mass is reduced by a factor of @xmath93 for the wmap3 parameters .
similarly , the mass of the largest halo at @xmath94 is altered by roughly the same factor .
other notable changes include : ( 1 ) the formation epoch of the first halo is shifted from @xmath95 in the wmap1 run to @xmath96 in the wmap3 run , and ( 2 ) the merging history of the largest halo at @xmath0 moves to lower redshifts in the wmap3 run , but the number of major mergers remains the same .
figure [ figmf ] shows the halo mass functions from different cosmological simulations .
the ps mass function @xcite , as well as the one corrected to match numerical simulations by @xcite ( st ) , are also shown for comparison .
one important feature in this figure is that the coarse runs agree well with the st mass function , but show a larger comoving density at the high mass end than that predicted by the ps theory .
our results show that the ps formula underestimates the abundance of high - mass halos by nearly an order of magnitude at @xmath97 , and the discrepancy between the ps calculation and numerical simulations becomes larger at higher redshifts , confirming previous findings ( e.g. , @xcite ) .
this may explain why previous models using the ps formula to study the abundance of luminous quasars , which presumably form in massive halos , under - predicted the number of bright quasars at @xmath98 ( e.g. , @xcite ) .
furthermore , these results also suggest that the commonly used analytical merger tree generated using the ps formula may not be suitable to study quasar formation at high redshifts .
there are two clear `` shifts '' of the mass function caused by resolution and cosmological parameters .
those from runs with higher resolution extend to higher redshifts , and at the same redshift , the wmap1 runs produce more massive halos than the wmap3 ones .
as shown in figure [ figmf ] , the coarse runs produce mass functions only up to @xmath99 owing to limited mass resolution , while the zoom - in runs can produce quite reasonable mass functions as early as @xmath100 . because the zoom - in runs were deliberately centered on the highest density peak of the @xmath35 box , they each contain a very massive halo ( @xmath101 at @xmath37 ) by construction .
this explains why the highest mass bin ( which contains only one halo in this case ) is @xmath56 orders of magnitude larger than the theoretical curves ( st , ps ) , which apply only to a _
random _ region that has a much lower density fluctuation . to summarize , at a given redshift , runs with the wmap3 parameters yield slightly less massive halos than ones performed with the wmap1 values . or , to put it differently , objects in the wmap3 cosmology will have masses similar to those for wmap1 , but at slightly later times ( i.e. lower redshifts ) . in
what follows , we are primarily concerned with investigating the plausibility of forming @xmath57 quasars through the self - regulated growth of smbhs in hierarchical mergers , rather than precisely reproducing the properties of an individual quasar at a given redshift , such as j1148 + 5251 .
most of our results are therefore based on runs with the wmap1 parameters , to ease comparison with previous numerical work . if it were firmly established that e.g. @xmath45 is in reality smaller than its wmap1 value , then a more exact match to a particular quasar could presumably be obtained by considering larger simulation volumes and identifying a suitable candidate host that is slightly rarer than the one we have chosen to focus on here .
traced by mergers at different redshifts with mass ratio @xmath102 , which is defined as the mass ratio between the halo and progenitor at a given time .
each of the progenitors joins in this big merging event at a given redshift , interacts with the system , and subsequently merges with others at later times .
the quasar host at @xmath0 is built up by seven successive major mergers of progenitors g1 , g2 , ...
g8 from @xmath96 to @xmath103 , as illustrated by the red lines in this plot .
the first interaction between g1 and g2 takes place at @xmath96 , then g3 and g4 join in the system at @xmath104 , followed by g5 , g6 , g7 and g8 at later times ( see text for more details ) .
the timeline of these events , the mass and other properties of these progenitors are listed in table 1.,width=336 ] to follow the hierarchical mass assembly of the host galaxy over cosmic time , the merger tree of the halo is extracted from the cosmological simulation .
this tree provides key information for computing the physical properties of the progenitor galaxy population . while the merger history of the halo includes a spectrum of progenitor masses , the most massive progenitors contribute the majority of the halo mass over the redshift range considered .
we trace the merger history of the most massive progenitor at each redshift by using particle tags to identify progenitor systems at earlier redshifts in the simulation .
groups that contribute at least 10% of the halo mass at a given time step are considered as the progenitors of the halo and are recorded .
the procedure is repeated until the last progenitor is reached , producing the merging history .
figure [ figtree ] illustrates the merging history of the largest halo at @xmath37 in our cosmological simulation , which has a mass of @xmath105 .
it is also the largest one at @xmath106 with a mass of @xmath107 .
this schematic plot outlines the redshift of merger event , and the mass ratio of the halo to its galaxy progenitors at a given time .
it shows that this halo grows rapidly through hierarchical mergers early on , with seven major mergers ( mass ratio of the merging pairs @xmath108 ) from @xmath96 to @xmath109 that build up a substantial fraction of the halo mass at @xmath0 . in modeling the development of a @xmath57 quasar
, we are primarily interested in `` major '' mergers , where the mass ratio of the merging galaxies is not too far from unity , for several reasons .
first , it is believed that major mergers play the most important role in the formation and evolution of massive galaxies ( e.g. , @xcite ) .
second , and of greater concern to us in this paper , in our picture for quasar fueling , gas in a rotationally supported disk loses angular momentum through gravitational torques excited by tidal forces in a merger , driving the growth of supermassive black holes .
this process operates most effectively in a major merger because the tidal deformation of each galaxy is significant in such an event @xcite .
collisions involving galaxies with a mass ratio as large as @xmath110 can induce gas inflows in disks @xcite , but only for limited orbital configurations .
for these reasons , we focus on mergers from the merger tree having a mass ratio @xmath108 , as outlined by the red color in figure [ figtree ] . in the resimulation of the merger tree as described in [ sec_mt ] , we take into account mass accretion of the halo by adding mass proportionally to each of the eight progenitor galaxies in the major mergers .
this approach preserves the progenitor mass ratios and approximately preserves the dynamics of the major mergers @xcite .
lccccccc [ tab1 ] g1 & 14.4 & 6.3 & 234.1 & 1.0 & 0.15 & &
+ g2 & 14.4 & 5.3 & 220.3 & 1.0 & 0.15 & 0.2 & 7.1 + g3 & 13.0 & 15.0 & 297.8 & 1.0 & 0.51 & 0.2 & 8.5 + g4 & 13.0 & 17.7 & 314.6 & 1.0 & 0.51 & 0.3 & 10.7 + g5 & 10.5 & 49.1 & 401.0 & 1.0 & 6.56 & 0.4 & 11.3 + g6 & 9.4 & 79.6 & 448.6 & 0.9 & 23.3 & 0.5 & 18.2 + g7 & 8.5 & 160.0 & 540.4 & 0.9 & 89.2 & 0.7 & 25.2 + g8 & 8.5 & 207.7 & 589.5 & 0.9 & 89.2 & 1.0 & 34.5 + in order to model the formation and evolution , and properties of the quasar candidate , the merger tree is then re - simulated hydrodynamically with galaxy models that consist of an extended dark matter halo , a rotationally supported , exponential disk of gas and stars and a central supermassive black hole .
we follow the evolution of the system built up by seven major mergers hierarchically from @xmath96 to @xmath0 , as shown in figure [ figtree ] . technically , this is a series of successive merger simulations .
the first simulation includes g1 and g2 interacting at @xmath111 .
it stops at @xmath112 and a new galaxy g3 is added into the system . during this process , all the dynamical properties of the pre - existing system ( e.g. g1 and g2 in this case ) are preserved , while g3 is added based on its properties and orbital parameters derived from cosmological simulations .
then a second merger simulation with g1 , g2 and g3 starts .
such a procedure is repeated until all the progenitors enter the system . in the end ,
the simulation includes all eight galaxies .
eventually all these galaxies and black holes merge together .
the duration of each merger simulation is determined by the merger tree .
the redshift at which each progenitor galaxy enters the merger tree , the properties of each progenitor galaxy , and the numerical parameters of the merger simulations are listed in table 1 .
below we describe the specification of these parameters .
the structure of the galaxy models is motivated from leading theories of dissipational disk galaxy formation in cdm cosmologies that , as shown by @xcite , are successful in reproducing the observed properties of both present - day disk galaxies and damped ly@xmath113 absorbers in quasar spectra at high redshift .
the initial galaxy models are constructed in dynamical equilibrium using a well - tested method ( @xcite ) .
a halo is identified with a virial mass @xmath114 and a virial radius @xmath115 within which the overdensity @xmath116 , where @xmath79 and @xmath117 are the mean and critical density , respectively .
the density profile of the dark matter halo follows a hernquist profile @xcite , scaled to match that found in cosmological simulations @xcite , as described in @xcite : @xmath118 where @xmath119 is a parameter that relates the @xcite profile parameters to the appropriate nfw halo scale - length @xmath120 and concentration @xmath121 ( @xmath122 ) , @xmath123}\,.\ ] ] the exponential disk of stars and gas are then constructed as in @xcite and @xcite .
the properties of the galaxy , including the virial mass @xmath114 , virial radius @xmath115 and halo concentration @xmath121 are scaled appropriately with redshift , as described in @xcite . in particular , for a progenitor with virial velocity @xmath124 at redshift @xmath81 , @xmath114 and @xmath115 are calculated following @xcite , while @xmath121 is adopted from @xcite , as briefly outlined below : @xmath125^{1/2}\ , , \\ { c_{\rm{vir } } } & = & 9 \left[\frac{{m_{\rm{vir}}}}{m_0}\right]^{-0.13 } \left(1+z\right)^{-1 } \,.\end{aligned}\ ] ] where @xmath126 is the gravitational constant , and @xmath127 is the linear collapse mass at the present epoch .
we assume a baryon fraction of @xmath128 for these high - redshift galaxies based on the wmap1 result @xcite .
the gas fraction of each progenitor is extrapolated from the results of semi - analytic models of galaxy formation @xcite , with 100% gas disks at @xmath129 and 90% at @xmath130 .
the multiphase ism is envisioned to consist of cold clouds embedded in a hot , tenuous gas in pressure equilibrium .
stars form out of the cold clouds by gravitational instability @xcite with a rate that is proportional to the surface density of the gas ( @xcite ; @xcite ) . in the adopted ism model for the gas , the equation of state ( eos )
is controlled by a parameter @xmath131 that linearly interpolates between isothermal gas ( @xmath132 ) and a strongly pressurized multiphase ism ( @xmath133 ) .
this eos describes the dynamics of star - forming gas and accounts for the consequences of stellar feedback on galactic scales , and enables us to construct equilibrium disk models even with large gas fractions @xcite .
supernova feedback is modeled through thermal energy input into surrounding gas and can help evaporate the cold clouds to replenish the hot phase . for the simulation presented here a value of @xmath134
is used , but test simulations using @xmath135 produce average star formation and black hole accretion rates that converge to within 15% .
the supermassive black holes are represented by collisionless `` sink '' particles .
they interact with other particles gravitationally , and accrete the gas .
accretion of gas onto the black holes is modeled using a bondi - hoyle - lyttleton parameterization ( @xcite ) , in which the black holes accrete spherically from a stationary , uniform distribution of gas , as described in @xcite and @xcite : @xmath136 where @xmath137 is the black hole mass , @xmath138 and @xmath139 are the density and sound speed of the gas , respectively , @xmath113 is a dimensionless parameter of order unity , and @xmath140 is the velocity of the black hole relative to the gas .
we assume the accretion has an upper limit by the eddington rate , @xmath141 where @xmath142 is the proton mass , @xmath143 is the thomson cross - section , and @xmath144 is the radiative efficiency .
the latter determines the conversion efficiency of mass accretion into energy released as radiated luminosity .
we adopt a fixed value of @xmath145 , which is the mean value for radiatively efficient @xcite accretion onto a schwarzschild black hole . in the simulations ,
the accretion rate is then the minimum of these two rates , @xmath146 .
the feedback from the black holes is associated with the mass accretion .
we assume that a small fraction ( @xmath147 ) of the radiated energy couples to the surrounding gas isotropically as feedback in form of thermal energy . this fraction is a free parameter , determined by matching the observed @xmath28 relation @xcite . for more discussions on this prescription ,
see @xcite .
this feedback scheme self - regulates the growth of the black hole , and has been demonstrated to successfully reproduce many observed properties of elliptical galaxies , as mentioned earlier . to grow a black hole up to @xmath148 in less than 800 million years , a wide range in seed masses , from @xmath149 to @xmath150 ,
have been suggested ( e.g. , @xcite ) .
the formation of the black hole seeds remains an open question , and several scenarios have been proposed .
in particular , @xcite show that rapid collapse of massive popiii stars due to pair instability could produce black hole of @xmath151 ; @xcite suggest that hot and dense gas clump may collapse monolithicly to form a massive black hole of @xmath152 in metal - free galaxies with a virial temperature of @xmath153 k ; while @xcite propose that @xmath154 black holes could form by direct collapse of self - gravitating gas due to global instabilities in protogalactic halos , they then grow to @xmath155 with super - eddington accretion .
we adopt the picture where black hole seeds are the remnants of the first stars ( @xcite ) .
the remnant black hole mass is currently uncertain and widely debated .
recent theory of popiii star formation predicts a mass range of @xmath156 , but there are two regimes where a smbh could form , either @xmath157 or @xmath158 ( @xcite , see also @xcite for a recent discussion ) .
we have tested the seed mass in the range of 100300 @xmath159 and find that the exponential growth of the black holes during the merger makes our results insensitive to the choice in that range .
we therefore assume that the black hole seed starts with an initial mass of @xmath160 after the collapse of the first star at @xmath62 .
these seed black holes then grow in the centers of @xmath152 halo which contains a large amount of high density primordial gas , as current theories predict that only one star forms per such mini - halo .
the dense gas in the central region provides abundant fuel for bh accretion . to account for their evolution before the major mergers
take place , the black holes are assumed to grow at the eddington rate until their host galaxies enter the simulated merger tree .
such an approximation is supported by the fact that the eddington ratio in the simulations depends on the galaxy interaction and strength of the feedback from the black holes . in our simulations ,
most black holes grow at nearly the eddington rate in the early stages of a galaxy interaction when the feedback is weak .
however , when the interaction and the feedback become stronger , the eddington ratios fluctuate by orders of magnitude .
so a constant accretion rate at the eddington limit is no longer appropriate , as we show below . under this assumption ,
the first progenitor galaxies ( g1 and g2 ) of the quasar host have black hole seeds of order @xmath161 by the time it enters the merger tree at @xmath162 .
however , we should emphasize that this assumption serves only as an upper limit of the early growth of the black holes .
our results in the next sections imply that even if all the black hole seeds had a uniform mass of @xmath163 when they enter the merger tree , it is still possible to build a massive one to @xmath148 at @xmath3 through gas - rich mergers . in our model ,
mergers are invoked in the formation of the most massive black holes of @xmath164 because that requires large supplies of gas .
early on , however , this may not be necessary to grow the black hole seeds from @xmath165 to the @xmath163 we start from , because the accretion rate is small so other gas fueling could be sufficient .
as demonstrated in @xcite , faint agns could be fueled by stochastic accretion of cold gas that does not involve mergers .
a similar process could go on in the black hole seeds left by the popiii stars at very high redshifts .
we should point out that in our simulations , it is necessary for galaxy progenitors in the merger tree to have reasonable massive black hole seeds ( @xmath163 ) initially in order to build a @xmath148 black hole at @xmath3 .
however , our results are insensitive to specific formation recipes of the seeds .
the formation of seed black holes at high redshifts is a challenging problem , and some of the proposed scenarios mentioned above may indeed be necessary to make our seeds . however , currently there is no observation available to test these models . in the picture
we adopt in which the seeds come from the first stars , the early accretion may be complicated by the feedback from the stars .
we note that recent studies by @xcite and @xcite show that hii regions form around the first stars , and that the halo gas would be photo - ionized , photo - heated , and evacuated by the radiation feedback from the stars .
@xcite suggest that such feedback would deplete the gas in the central region , and would delay the black hole accretion by up to @xmath166 yrs .
however , this destruction effect depends sensitively on the lifetime of these massive stars , and more importantly on the environment which determines both the gas density profile and gas replenish from inflow of the expelled gas or neighboring halos . in the simulations presented in @xcite ,
the box size is only @xmath167 , too small to contain the large scale gravitational potential and the large wavelength density modes that drive gas infall , so the initial gas density is low and the destruction timescale is long in this case . however , the quasar halo in our simulation resides in the highest density peak in a volume of @xmath168 , where the halo potential and gas density , as well we the accretion rate are much higher @xcite .
for a @xmath160 black hole , the accretion rate at eddington limit is only @xmath169 , which corresponds to the bondi accretion of molecular gas with a typical temperature of @xmath170 k at density @xmath171 , as implied from equation ( [ eq_bondi ] ) .
such a density requirement is satisfied with the initial conditions of our model .
therefore , the gas re - incorporation timescale in our case may be substantially shorter than that estimated in @xcite .
we will investigate in a future project the growth and evolution of the early black holes after the death of the first stars in such a cosmological environment , using hydro - radiation simulations that include both radiative transfer and black hole accretion with ultra - high resolutions .
the merger tree contains eight galaxies engaging in seven major mergers at different times . for each merger event ,
the initial orbits of the incoming progenitors are set to be parabolic , consistent with the majority of the major mergers in our cosmological simulation and with previous findings @xcite .
the orientation of each merging galaxy is selected randomly .
the initial separation between each merging pair is set to @xmath172 , where @xmath173 is the virial radius of the incoming system , while the pericentric distance is chosen as @xmath174 , where @xmath175 is the radial disk scale length of the incoming system .
we have tested different choices of @xmath176 and orientations , and found that the impact of these parameters is minor because the orbital properties of the progenitors change rapidly through interactions with the multiple galaxies in the system . throughout the merger simulation ,
the mass and force resolutions are fixed for each particle type , and the total initial particle number of @xmath177 results in particle masses of @xmath178 for the halo and @xmath179 for both the gas and stars .
the gravitational softening lengths are @xmath180 for halo particles and @xmath181 for both gas and stars . in the simulations
, it is impossible to resolve individual stars , and the accretion radii of some small black holes are under - resolved .
however , with the sub - resolution implementation in our models , we can calculate time - averaged rates of star formation and black hole accretion from the large - scale properties of the gas , which are well resolved in our simulations .
resolution studies of a single merger @xcite with particle numbers from @xmath182 to @xmath183 show that resolution affects some fine structures of the gas and the instantaneous growth rates of star and black holes , but the time - averaged properties of the system converge to within 20% . as a function of distance @xmath184 ( @xmath115 is the virial radius ) to the halo center for various models in our merger simulations .
this plot includes the isolated halo progenitors g1 g8 in table 1 , as well as the first merger remnant at @xmath185 and the last one at @xmath186 , as labeled in the legend .
the shaded region indicates the range of the escape velocities of the mergers in our simulations , with the values in the central regions being @xmath187 .
, width=336 ] in a galaxy merger with black holes , the black holes may merge into one , or may be ejected by gravitational recoil in the final stage .
their fate depends on the halo escape velocity @xmath188 .
if the recoil velocity is larger than @xmath188 , then the black hole will be kicked out of the halo .
we follow @xcite to calculate this important parameter @xmath189 .
it is defined by @xmath190 where @xmath191 is the gravitational potential at a given radius @xmath192 . because the halo is spherical ,
the potentials of different spherical shells add linearly , so @xmath191 is contributed by two parts , i.e. , shells within @xmath192 ( @xmath193 ) and outside ( @xmath194 ) : @xmath195\,.\ ] ] where @xmath196 is again the @xcite density profile of dark matter halo as in equation [ eq_hernquist ] .
figure [ figvesc ] shows the escape velocities of the halo progenitors g1 g8 in table 1 , as well as two merger remnants at @xmath197 and @xmath186 , respectively .
the escape velocity depends on the halo mass , redshift , and distance from the halo center .
the @xmath188 remains constant in the central region , begins to decline around @xmath198 . at the center , @xmath199 , while at the virial radius @xmath115 , the escape velocity is comparable to the virial velocity ( by a factor of @xmath200 ) .
the isolated halo progenitors g1 g8 have a @xmath188 range of @xmath201 @xmath202 . the first merger halo at @xmath185 , which has a mass of @xmath203 as the merger of g1 and g2 , has a central escape velocity of @xmath204 , while the final merger halo at @xmath186 , which has a mass of @xmath205 , has @xmath206 .
the shaded region indicates the range of the halo escape velocities of the mergers in our simulations .
in particular , the escape speed in the halo central region has a range of @xmath207 .
this range is important for analysis of black hole ejection from gravitational recoil in the black hole binaries in [ subsec_bhgrowth ] and [ subsec_bhbinary ] .
the vigorous merging history of the quasar host is illustrated through selected snapshots of the gas and stellar distributions in figure [ figpic_gas ] and figure [ figpic_star ] , respectively .
the progenitors at high redshifts are very compact and gas rich . as the host galaxy of the quasar builds up hierarchically , strong gravitational interactions between the merging galaxies lead to tidal tails , strong shocks and efficient gas inflow that triggers large - scale starbursts , a phenomenon that has been demonstrated by many numerical simulations ( e.g. , @xcite ) , as reviewed by @xcite
. the highly concentrated gas fuels rapid accretion onto the smbhs ( @xcite ) . between @xmath208 149
, the merging systems are physically small and the interactions occur on the scale of tens of kiloparsecs .
by @xmath208 97 , when the last major mergers take place , the scale and strength of interactions have increased dramatically .
galaxies are largely disrupted in close encounters , tidal tails of gas and stars extend over hundreds of kiloparsecs , and intense bursts of star formation are triggered .
the black holes continue to grow rapidly during this period but are heavily obscured by a significant amount of circumnuclear gas . during galaxy mergers ,
the black holes follow their hosts to the center of the system and can interact closely with each other .
it has been shown that black hole binaries decay rapidly in a gaseous environment and can merge within @xmath209 yrs ( @xcite ) . because the galaxies in our simulations are very gas rich and the gas is highly concentrated during the mergers
, we therefore assume that the black holes merge efficiently owing to strong dynamical friction with the gas @xcite .
we will return to more discussions of this process in [ subsec_bhgrowth ] and [ subsec_bhbinary ] .
at redshift @xmath210 the progenitor galaxies coalesce , inducing high central gas densities that bring the smbh accretion and feedback to a climax .
the smbh feedback then drives a powerful galactic wind that clears the obscuring material from the center of the system .
the largest smbh becomes visible as an optically - bright quasar @xcite during this phase , after which quasar feedback quenches star formation and self - regulates smbh accretion .
consequently , both star formation and quasar activity die down , leaving a remnant which reddens rapidly , as illustrated schematically in figure [ figpic_star ] .
the evolution of the star formation rates ( sfrs ) of each individual galaxy , and total sfr of the whole system are shown in figure [ figsfr ] ( _ top panel _ )
. the system forms stars rapidly as these compact and gas rich progenitors undergo strong interactions .
the total sfr ranges from @xmath211 to @xmath212 between redshifts @xmath213 98 when the galaxies begin their final major mergers , while the sfrs of individual galaxies fall below a few@xmath214 , within the starburst intensity limit of @xmath215 proposed by @xcite and @xcite . at @xmath216
the star formation rate decreases gradually owing to a depletion of the gas supply and progressively stronger feedback from the smbhs . at the time of final coalescence ( @xmath210 ) the star formation rate is @xmath217 , an order of magnitude lower than for estimates of j1148 + 5251(@xcite ) .
we note , however , that the estimates by these authors are based on the assumption that the fir luminosity is dominated by young stars , and they can not rule out the possibility that agn may contribute significantly to the luminosity . in a forthcoming paper @xcite ,
we have calculated the infrared properties of the quasar system using a 3-d monte carlo radiative transfer code that incorporates adaptive grids and treats dust emission self - consistently .
we find that the far - infrared luminosity of our quasar is not dominated by young stars but instead has a substantial quasar contribution of over @xmath218 .
this finding is supported by observations of j1148 + 5251 in near - ir ( e.g. , @xcite , and more recently @xcite ) , which show a remarkably flat spectral energy distribution and suggest an agn origin for the flux excess .
furthermore , adopting a total gas mass of @xmath219 ( @xcite ) in j1148 + 5251 , a simple application of the schmidt - kennicutt star formation law ( @xcite ) gives a star formation rate of @xmath220 , close to what we find here . within only about 600 myrs from @xmath162 to @xmath65 ,
the system accumulates a stellar mass of @xmath221 as shown in figure [ figsfr ] ( _ bottom panel _ ) .
the specific star formation rate ( ssfr ) , or `` b - parameter '' ( e.g. , @xcite ) , is defined as @xmath222 .
it is a measure of the fraction of the total stellar mass currently forming at a specific time .
the ssfr is typically larger in high - redshift galaxies than in ones at low redshifts owing to vigorous star formation . during the past several years
, there has been rapid progress in observing galaxies at @xmath12 using the _ hubble space telescope _ ( hst ) , and the _ spitzer space telescope _ ( spitzer ) coupled with ground - based observatories ( e.g. , ) , and hundreds of these distant objects have been detected .
these frontier observations suggest that the universe experienced rapid star formation during the redshift interval @xmath223 , and the development of large stellar systems in the mass range of @xmath224 @xmath225 .
in particular , several groups ( @xcite ) find ssfrs in the range of @xmath226@xmath227 in their observations , consistent with our simulations . ) of each galaxy ( black curve ) .
the blue hatched region indicates the range of 25%75% of the gas metallicity .
, width=336 ] , from both gas ( _ top panel _ ) and stars ( _ bottom panel _ ) , respectively .
the images are projected metallicity adaptively smoothed over 32 particles ( analog to the sph kernel in a 2-dimensional plane ) .
the black dot indicates the center of the quasar.,width=336 ] rapid star formation in the quasar progenitors produces an abundant mass of heavy elements to enrich the ism .
observations of j1148 + 5251 show solar metallicity in the system ( @xcite ) .
figure [ figmett ] shows the metallicity in our simulated quasar system at different times .
note that the dips and jumps in the curves owe to incoming new galaxies which bring in metal - poor pristine gas and newly - formed stars .
the quasar host reaches solar metallicity as early as @xmath228 , and maintains similar levels to later times .
the spatial distribution of metallicity from both gas and stars at the peak quasar phase at @xmath229 is shown in figure [ figmetr ]
. the metals are widely spread owing to outflow from the quasar feedback and gas infall toward the merger center .
the metallicity in the central region of the merger remnant is slightly above the solar value . in some outer regions , because the gas and stars are still falling back to the system center , the infalling material triggers small - scale bursts of star formation .
so the metallicity in these blobs appears to be super - solar , as shown in figure [ figmetr ] .
calculations of carbon monoxide emission using non - local thermodynamic equilibrium radiative transfer codes @xcite by @xcite show co luminosities , excitation patterns , and morphologies within the central @xmath230 of the quasar host center that are consistent with observations of j1148 + 5251 ( @xcite ) .
these results were derived using galactic co abundances , and thus support our conclusions that significant metal enrichment takes place early in the quasar host , as a result of strong star formation in the progenitors .
( _ middle panel _ ) , and black hole mass ( _ bottom panel _ ) .
note that the black curves represent totals , while colored curves show individual black holes , as indicated in the legend.,width=336 ] in the simulations , the quasar host at @xmath0 is built up by eight progenitors , each containing a black hole in the center .
figure [ figar ] shows the evolution of the black hole accretion rate , the eddington ratio , and the integrated masses of the whole system and individual black holes .
the total black hole accretion rate grows steadily during the hierarchical assembly of the host galaxy and peaks at @xmath231 around @xmath210 during the final coalescence .
the eddington ratio , @xmath232 , of each individual black hole varies with time , depending on the galaxy interaction and feedback from the black holes .
the black holes maintain accretion at the eddington limit for only a fraction ( @xmath233 ) of the time . at the peak of quasar activity
, the eddington ratio of the most massive black hole is near unity , while that of the other black hole is only 0.1 .
however , collectively , the whole system appears to accrete at @xmath234 at @xmath235 , as implied in figure [ figar ] .
studies of black hole accretion ( e.g. , @xcite ) show that the eddington ratio has a wide range of 0.011.0 , and it varies with both luminosity and redshift .
luminous systems tend to have higher @xmath232 than less - luminous counterparts , and at @xmath11 , most quasars shine at nearly eddington luminosity .
our results suggest that _ individual _ black holes do not always necessarily accrete at the eddington rate .
however , since high - redshift , luminous quasars may form through mergers of several galaxy progenitors containing black holes as in our case , therefore the growth of the quasar represents a _ collective _ contribution from each individual black hole .
the total black hole mass increases from @xmath236 at @xmath237 to about @xmath238 at @xmath229 , close to that estimated for j1148 + 5251 by @xcite and @xcite . in the simulations
, we do not have sufficient resolution nor the relativistic physics to consider the ejection of black holes by gravitational recoil .
the black holes are assumed to merge efficiently once their separation is below the spatial resolution . in the final stage of black hole mergers ,
the emission of gravitational wave carries linear momentum , which could cause the black holes to recoil ( e.g. , @xcite ) .
if the recoil velocity is larger than the halo escape velocity , then the black holes will be kicked out from their halo ( e.g. , ) .
previous studies by @xcite suggest that constant or super - eddington accretion is required to produce @xmath148 black holes at @xmath57 if ejection of black holes is included .
in particular , @xcite suggests a black hole will be ejected if the kick velocity @xmath239 for the coalescing smbh binary is larger than twice the halo velocity dispersion @xmath240 , @xmath241 , as the dynamical friction timescale for the kicked black hole to return to the halo center is longer than the hubble time @xcite . by applying this ejection criterion to a press - schechter merger tree of a @xmath242 halo within which j1148 + 5251 is assumed to reside , @xcite finds that the smbh of the quasar gains most of its mass rapidly from seed holes during @xmath243 due to black hole ejection , and the smbh likely accretes with super - eddington rate in order to build a mass as that of j1148 + 5251
. however , the halo escape velocities or velocity dispersions in our model are much larger than the currently best estimates of the kick velocity .
the quasar halo in our simulations has active merging history from redshifts @xmath244 to @xmath186 , and the halo progenitors have masses much higher than those considered in the previous studies @xcite .
the quasar halo builds its mass from @xmath245 at @xmath246 ( the sum of progenitors g1 and g2 in table 1 ) to @xmath205 at @xmath65 .
as shown in figure [ figvesc ] ( the shaded region ) , the central escape velocity of the mergers in our simulations is in the range @xmath247 .
currently , the maximum kick velocity for unequal - mass , non - rotating bh binary is in the range of @xmath248
@xmath249 from both analytic post - newtonian approximation ( e.g. , @xcite ) and the ground - breaking full relativistic numerical simulations ( e.g. , @xcite ) . for equal - mass ,
spinning bh binary , @xcite estimate a range of @xmath170 @xmath250 using bh perturbation theory , and @xcite derive a formula from relativistic simulations , @xmath251 , where @xmath252 is the bh spin .
this gives a maximum kick of @xmath253 for maximal spin .
although it is also reported that the recoil velocity can be as large as thousands @xmath254 @xcite for bh binary in the orbital plane with opposite - directed spin . however , as pointed out by @xcite , such a configuration is rather uncommon , especially in gas - rich galaxy mergers , because torques from accreting gas suffice to align the orbit and spins of both black holes with the large - scale gas flow .
the resulting maximum kick velocity from such a configuration is @xmath255 .
overall , the kick velocity from the latest calculations of black hole binary is in the range of @xmath170
@xmath256 , falling safely below the escape velocities of the quasar halos in our simulations , so black hole ejection may be insignificant in our case .
moreover , we find that our model can produce a @xmath148 smbh even if ejection is allowed . from figure [ figar ] ( bottom panel ) , the @xmath148 smbh is dominated by bh5 , most of its mass comes from gas accretion .
even if the less massive black holes , for example bh7 or bh8 was ejected , the most massive one bh5 is still able to reach @xmath1 in the end .
furthermore , even if all the seeds started with @xmath257 in the merger tree , the result would be about the same .
we therefore conclude that the results from our modeling are robust .
suppermassive black holes of @xmath1 can grow rapidly through gas accretion and mergers hierarchically in the early universe , constant or super - eddington accretion is not necessary , unless the recoil velocity of the coalescing black hole binary is extremely high such that most of the black hole seeds in our simulations are ejected ( e.g. , @xmath258 ) . at the present day , as parameterized by @xcite.,width=336 ] tight correlations between supermassive black holes and hosts have been observed in local galaxies ( e.g. , @xcite ) , but the inference of these relationships at higher redshifts remains an open question . because the eight galaxies in our system interact vigorously with each other , the stellar components are widely spread and mixed , it is impossible to separate individual galaxy - smbh pairs , so we only consider the correlations in _ total _ quantity of the whole system .
a comparison of the total stellar mass and total black hole mass is shown in figure [ figmas ] . at early time , both the stars and black holes grow rapidly through galaxy mergers . shortly after the peak quasar phase
, strong feedback suppresses both the accretion and star formation , the masses of the black holes and stars become saturated gradually , and in the end satisfy @xmath259 , similar to the correlation measured in nearby galaxies ( @xcite ) .
our results are consistent with findings by @xcite and @xcite , and demonstrate that the observed @xmath10 scaling relation is a result of the coeval growth of the smbh and its host galaxy , and that it holds across different cosmic times .
we note , however , that the velocity dispersion of the stars in the remnant center is about @xmath260 ( after the system relaxes ) owing to the deep potential well of the merger system , so the @xmath28 relation falls below the correlation observed locally ( @xcite ) .
single mergers of progenitor galaxies constructed in a redshift range of z=06 by @xcite appear to follow the observed @xmath28 correlation with a weak redshift dependence of the normalization , which results from an increasing velocity dispersion of the progenitors at higher redshift .
the multiple mergers we derive from cosmological simulations take place at much higher redshifts and hence the progenitors have larger velocity dispersions , implying a larger deviation from the local @xmath28 relation than in the work of @xcite .
however , because we do not follow subsequent mergers and accretion into the host halo below @xmath57 , the implications of this result for the evolution of the @xmath28 relation are unclear .
observations of active galaxies have yielded ambiguous results about the smbh spheroid relationship .
for example , @xcite report a lower zero - point of the @xmath28 of local active galaxies than that of the inactive sample @xcite ; at @xmath261 , @xcite found the same @xmath28 relation in the redshift range @xmath26213 , while others ( e.g. , @xcite ) show correlations with various offsets .
in particular , @xcite estimate a dynamical mass of @xmath263 using the co linewidth measured in j1148 + 5251 , and suggest that the bulge is under - massive by at least one order of magnitude compared to the local @xmath10 relation .
however , the co calculation by @xcite finds that the co linewidth of the quasar in our simulation is larger than the mean 280 km s@xmath264 measured by @xcite and @xcite by almost an order of magnitude , and that the derived dynamical mass is @xmath265 , putting the simulated quasar on the @xmath10 correlation .
@xcite further suggest that the observed emission line may be sitting on top of a much broader line , which may be tested by future observations with large bandwidths .
the different relations reported from the observations may reflect a divergence of the methods used to estimate the black hole mass and stellar properties , or may represent different evolutionary stages of the systems @xcite .
more observations and better measurements of black hole mass and properties of host bulges will be crucial to study the smbh host relations in high - redshift quasar systems @xcite and to test our hypothesis .
.,width=336 ] both the bolometric and attenuated luminosities of the quasar and the host galaxy in the simulations can be readily calculated following the methodology of @xcite .
the bolometric luminosity @xmath266 of stars is calculated using the stellar population synthesis model of @xcite , while that of a black hole is calculated as @xmath267 , where @xmath268 is the radiative efficiency , @xmath269 is the black hole accretion rate , and @xmath270 is the speed of light . in this calculation ,
the black holes are assumed to be non - rotating .
if the black holes are spinning , their radiative efficiencies and luminosities would be higher due to the shrink of the innermost stable circular orbit , by up to a factor of 4 for maximal rotation .
the b - band luminosity of each source is corrected for attenuation by absorption from the ism along the line - of - sight .
we first calculate the line - of - sight column - density of the gas from each source to a distant observer . for each black hole
we generate 1000 radial sight lines originating at the black hole particle location and uniformly spaced in the @xmath271 solid angle @xmath272 , while for the stars , an accurate estimate of the luminosity is possible with only one sight line per source owing to the extended distribution . along each ray
, the gas column density is calculated using a radial spacing of @xmath273 , where @xmath274 and @xmath275 is the local sph smoothing length .
the distribution of line - of - sight properties converges for @xmath276 rays and at a distance of @xmath277 . in the calculation , only the diffuse - phase density is considered because of its large volume filling factor @xmath278 , allowing for a determination of the lower limit on the column density along a particular line of sight . adopting
the mean observed intrinsic quasar continuum spectral energy distribution @xcite gives a b - band luminosity which is well approximated by the following equation given by @xcite , @xmath279 , where @xmath280 , and @xmath281 .
we then use the milky way gas - to - dust ratio scaled by metallicity , @xmath282 to determine the extinction along a given line of sight for this band . in the above calculation
, we do not include a full treatment of radiative transfer , and therefore do not model scattering or re - processing of radiation by dust in the infrared .
however , for the b - band luminosity , results using a 3-d monte carlo radiative transfer code are close to those calculated using the methods we present here @xcite . figure [ figlum ] shows both the bolometric and attenuated b - band luminosities of the quasar , compared with observations of j1148 + 5251 .
the system is intrinsically bright with a total luminosity @xmath283 , and the host appears as an ultraluminous infrared galaxy ( ulirg ) with @xmath284 for most of the time . at high redshifts , @xmath285
, starlight dominates the total luminosity . however , black holes take over at a later time .
the quasar light - curve increases dramatically , peaking at @xmath30 , when it is powered by the most massive black hole accreting at near the eddington rate .
the estimated @xmath266 of j1148 + 5251 differs from that of the simulated quasar by less than the uncertainty in the luminosity estimate .
the rest - frame b - band absolute magnitude reaches @xmath286 , almost one magnitude fainter than that of j1148 + 5251 derived from 1450@xmath287 data @xcite .
however , we should emphasize that in this paper , our main goal is to investigate the plausibility of forming luminous @xmath13 quasars through hierarchical mergers , rather than precisely reproducing the properties of an individual quasar such as j1148 + 5251 , so the disagreement shown in figure [ figlum ] should not be taken too literally .
moreover , the _ exact _ luminosity can change by a factor of several from relatively trivial or random details in the simulations .
if the black hole spin is taken into account , then the simulated luminosities would increase by a factor of up to 4 , which would match the observation of j1148 + 5251 better .
feedback - driven outflows create un - obscured lines - of - sight , allowing the growing central smbh to be visible as an optically - bright quasar between redshifts @xmath288 and @xmath74 . at the peak of the quasar activity , more than 50% of the 1000 sight lines have @xmath289 .
the absorbed light is re - emitted at infrared wavelengths by dust .
we find that the luminosity in the far infrared @xcite is close to @xmath290 estimated for j1148 + 5251 by @xcite .
moreover , we find that up to 80% of the fir light comes from the black hole , while stars contribute only @xmath291 .
this may explain why the star formation rate at @xmath292 during the peak quasar phase is an order of magnitude lower than the estimate from fir observations @xcite , which will be contaminated by the agn .
another prominent feature of figure [ figlum ] is a clear phase transition from starburst to quasar .
it has long been suggested that ulirgs are powered by starbursts in galaxy mergers ( for reviews , see @xcite ) , and that bright quasars are the descendants of ulirgs ( @xcite ) .
this conjecture has been supported by observations of quasar hosts ( e.g. , @xcite ) , and theoretical modeling @xcite . in @xcite , we calculate the spectral energy distributions ( seds ) of the quasar system and its galaxy progenitors . we find that the seds of the system at @xmath285 are characterized by those of starburst galaxies @xcite , while at the peak quasar phase , the seds resemble those observed in @xmath0 quasars @xcite .
we also find that the system evolves from cold to warm ulirg as it transforms from starburst to quasar phase .
our results provide further theoretical evidence for the ulirg
quasar connection in quasar systems in the early universe .
the quasar lifetimes depend on the observed luminosity threshold , as proposed by @xcite . in our simulation , at the peak luminosity of @xmath293 , the quasar lifetime is roughly @xmath294 yrs , as shown in figure [ figtq ] . again
, if black hole spin is included in the calculation , the luminosity of the quasar would increase by a factor of several , and the quasar lifetime would be longer .
however , when increasing the radiative efficiency , the salpeter time ( @xmath295-folding time for eddington - limited black hole growth , @xcite ) is increased by an identical factor , meaning it would also require a longer time to reach the same mass .
if high - redshift quasars are rapidly rotating , then , our calculations demand that either the seeds be _ much _ more massive at @xmath12 , or that they accrete in a super - eddington manner . in other words
, if the observed sloan quasars at @xmath0 shine with eddington luminosity but are rotating rapidly , then our model suggests that their masses would be considerably smaller than estimated .
we note that recent _ spitzer _ observations by @xcite show that 2 out of 13 quasars at @xmath57 have a remarkably low nir - to - optical flux ratio compared to other quasars at different redshifts , and these authors suggest that the two quasars may have different dust properties . according to our model ,
however , these two outliers may be young quasars that have just experienced their first major starburst but have not yet reached peak quasar activity , so the light from star formation may be dominant , or comparable to that from the accreting smbh still buried in dense gas .
this may explain the low nir flux , as well as the b - band luminosity and the narrow @xmath296 emission line , which are primarily produced by the starburst .
we will address this question further in a future paper with detailed modeling and ir calculations @xcite .
our multi - scale simulations that include large - scale cosmological n - body calculations and hydrodynamic simulations of galaxy mergers , as well as a self - regulated model for black hole growth , have successfully produced a luminous quasar at @xmath3 with a black hole mass of @xmath297 and a number of properties similar to those of j1148 + 5251 , the most distant quasar detected at @xmath298 @xcite .
our approach differs from previous semi - analytic studies by @xcite and @xcite in the following ways : 1 .
we use a realistic merger tree derived directly from multi - scale , high - resolution cosmological simulations .
the previous studies used merger trees of dark matter halos generated with the extended press - schechter theory @xcite , which may underestimate the abundance of high - mass halos by up to one order of magnitude , as shown in [ subsec_mf ] . also , the merger trees in those studies started from much higher redshifts than what we consider here . in our model ,
the quasar halo is the largest one in a volume of @xmath299 .
it has a mass of @xmath300 at @xmath301 built up through seven major mergers from @xmath244 to @xmath302 .
we follow the evolution of the system and treat the gas dynamics , star formation , and black hole growth properly .
this approach is critical to investigation of the properties of both black holes and host galaxies , and their evolution ( e.g. , @xcite ) , but it was not included in those previous studies on formation of @xmath0 quasars .
we employ a self - regulated model for the growth of supermassive black holes , in which the accretion is regulated by the black hole feedback , and the rate is under the eddington limit . in the previous studies ,
the black hole growth was unregulated , but instead a constant- or super - eddington accretion rate was used .
4 . in our simulations
, we do not consider black hole ejection caused by gravitational recoil owing to insufficient resolution and lack of relativistic physics .
however , the halo escape velocities in our simulations are in the range of @xmath303
@xmath304 , much larger than the kick velocity @xmath170 @xmath253 ( e.g. , @xcite , see [ subsec_vesc ] and [ subsec_bhgrowth ] for more details ) .
therefore , black hole ejection may be negligible in our case .
previous studies had much smaller halo progenitors at higher redshifts than ours , so the black hole seeds would be more likely subject to ejection from their halos .
this leads to the conclusion in these studies that constant or super - eddington accretion is needed owing to significant black hole ejection .
our results are robust within the best estimates currently available for the recoil velocity of the black hole binary .
the black hole seeds in the galaxy progenitors in our simulations are massive ( e.g. , @xmath163 at @xmath305 ) .
the sub - resolution recipe in our model does not allow us to resolve the actual formation and accretion of such black holes below this mass scale .
the formation of these seeds is an unsolved problem , but our results do not depend on the specific prescription of the formation process .
we adopt a picture in which the seed holes come from the remnants of the first stars ( which have a mass @xmath160 at @xmath306 ) and grow under eddington limit until they enter the merger tree we simulated .
if the growth is delayed by radiation feedback from the popiii stars ( e.g. , @xcite ) , then super - eddington accretion , or other proposed scenarios ( e.g. , @xcite ) may be necessary to form massive seeds of @xmath307 in the protogalaxies .
overall , we conclude that the results from our simulations , which are more realistic and more detailed than the models previously done , are robust .
suppermassive black holes of @xmath1 can form rapidly through gas - rich hierarchical mergers under eddington limit , even within a short period of time .
we find that constant or super - eddington accretion is not necessary unless the above assumptions in our modeling break , i.e. , there are no massive black hole seeds of @xmath308 available at @xmath309 , or the recoil velocity of the coalescing black hole binary is extremely high ( e.g. , @xmath310 ) . under these extreme circumstances ,
some `` exotic '' processes such as super - eddington accretion may be necessary to grow a @xmath1 smbh within a few hundred million years .
however , we should note , as pointed out by @xcite , that most gas - rich galaxy mergers have a configuration such that the orbit and spins of both black holes are aligned with the large - scale gas flow owing to torques from accreting gas .
such a configuration has a maximum kick velocity @xmath255 , which is well below the escape velocity of a @xmath311 dwarf galaxy , as well as those of the halos in our modeling . during the galaxy mergers ,
the black holes follow their host halos to the system center and can form binaries ( or multiple systems ) .
the coalescence of a black hole binary includes three distinct phases : inspiral , merger , and ringdown ( e.g. , @xcite ) .
whether black hole binaries can coalesce on short timescales is a matter of debate . in a stellar environment
, it has been argued that a binary hardens very slowly owing to an eventual depletion of stars that cause the binary to lose angular momentum ( e.g. , @xcite ) . in a gaseous environment , however , numerical simulations by @xcite and @xcite show that the binaries decay rapidly owing to strong dynamical friction with the gas , and they likely merge within @xmath312 years . because our galaxies are very gas rich and have large central concentrations of gas during the mergers , we assume the black hole particles coalesce once their separation decreases below our spatial resolution ( @xmath313 ) and their relative speed falls below the local gas sound speed @xcite . in the simulations
, we do not have sufficient resolution nor the relativistic physics to consider the ejection of black holes by gravitational recoil during the merger phase .
however , as discussed in
[ subsec_vesc ] and [ subsec_bhgrowth ] , the halo escape velocities in our simulations are much larger than the maximum kick velocity for black hole binary estimated from the latest relativistic calculations .
so black hole ejection is likely unimportant in our modeling . to accurately address gravitational recoil in the galaxy merger simulations , we need to include general relativity , resolve the dynamics of black hole binaries with extremely high resolution , and calculate the halo potential in a cosmological context ( in which halo potential distribution may be different from that of a single object ) .
however , such a comprehensive treatment is impossible at the moment .
we therefore assume the black holes merge quickly once they reach the stage of gravitational radiation .
these coalescing supermassive black holes will be strong sources of gravitational radiation detectable by the laser interferometer space antenna ( _ lisa _ , @xcite ) , as suggested by many authors ( e.g. , @xcite ) . by tracing the merging history of the smbhs ,
_ lisa _ could shed light on the distribution , structures and evolution of the associated dark matter halos .
because luminous , high - redshift quasars are likely sites of vigorous hierarchical mergers , they may be the best targets for _ lisa _ to explore the early universe . ) , and the specifications of the wind model is : wind efficiency @xmath314 , wind energy coefficient @xmath315 , wind free travel length @xmath316 , and a wind velocity @xmath317.,width=336 ] vigorous star formation would induce a galactic wind and mass outflow , a phenomenon that has been observed to prevail in both local star - forming galaxies as indicated by blueshifted optical absorption lines ( e.g. , @xcite ) , and lyman break galaxies at @xmath318 as indicated by blueshifted interstellar absorption lines and redshifted ly@xmath113 emission lines ( e.g. , @xcite ) , as well as ly@xmath113 emitters at @xmath319 @xcite .
these galactic winds are generally thought to play a significant role in galaxy evolution ( e.g. , see @xcite and @xcite for recent reviews ) .
the strong starburst preceding the major quasar phase in our simulations may drive strong galactic winds and affect the black hole growth . to investigate the impact of the feedback from a starburst - driven wind on the growth of the black hole , we have done the same merger simulation with lower resolution ( @xmath320 ) and with a canonical wind model from @xcite : the wind efficiency @xmath314 , which measures the coefficient of the star formation that determines the mass outflow ; the energy fraction from supernovae injected into the wind @xmath315 , wind free travel length @xmath316 , and a wind velocity @xmath321 . as demonstrated by @xcite ,
this wind model is able to reproduce the starbursts as observed in lyman break galaxies , and therefore is suitable to our study .
we find that the impact of the starburst - driven wind on the quasar evolution is minor , as shown in figure [ figwind ] .
both the histories of star formation and black hole growth remain roughly the same as in the simulation without a starburst wind , only the amplitude is lowered by a factor of @xmath200 .
similarly , the final masses of the black hole and the stars are reduced by roughly the same factor , but the quasar host is still on the @xmath10 correlation .
the peak quasar phase is delayed to @xmath0 .
overall , the starburst wind affects the gas dynamics locally , but owing to the deep potential of the system , its impact on the process of quasar formation is minor .
our results support the finding by @xcite that feedback from starburst - driven winds alone is ineffective at regulating the growth of the central black hole , so feedback from the black hole plays the dominant role in the formation and evolution of quasars . from the zoom - in simulations with parameters from both wmap1 ( solid line ) and wmap3 ( dashed line ) .
the volume of the high - resolution zoom - in region is @xmath322.,width=336 ] because we have so far simulated only one quasar in a volume of @xmath323 , we are not yet able to constrain the expected abundance of quasars @xmath57 .
as mentioned in 2.3 , at a given redshift , cosmological simulations with parameters from wmap1
produce more massive halos than runs with wmap3 owing to a larger value of @xmath45 .
figure [ fignum ] shows the number of halos at @xmath57 from the zoom - in runs with parameters from both wmap1 and wmap3 .
there are about three dozen halos with mass @xmath324 in the wmap1 run , while in the wmap3 run there are only a handful of such halos .
however , since in our picture the quasar activity depends not only on the halo mass , but also on the merging history , an accurate estimate of the quasar abundance and luminosity function would require hydrodynamical simulations of all the quasar candidates in a large box , which are currently unavailable . nevertheless , all conditions being equal , the change from the wmap3 parameters would produce fewer luminous quasars at @xmath0 .
this suggests that in a wmap3 cosmology , the quasar observed with the largest redshift , j1148 + 5251 , might have formed in a slightly higher overdensity peak than that we have presented here . in that event , if the wmap3 determination of @xmath45 were correct , we would need to identify a rarer density fluctuation to match j1148 + 5251 at its observed redshift .
however , this does not change our conclusion that the most distant and luminous quasars can form from hierarchical galaxy mergers in the @xmath2cdm cosmology .
imaging surveys of j1148 + 5251 show that there is no other luminous quasar from the same epoch in the field ( @xcite ) . in our simulations , around the peak of quasar activity at @xmath292 , there are no other halos of mass @xmath325 within a few mpc of this quasar .
however , the numerous major mergers this halo experienced prior to the peak quasar activity demonstrates that this region was once highly clustered with massive halos , but they merged to become a bigger one by @xmath326 . as seen from figure [ figtree ] , this quasar halo will undergo a handful of major mergers at a later time from @xmath327 41 , and eventually end up as a cd - like galaxy at the center of a rich cluster . since we do not follow hydrodynamically the evolution of the quasar at @xmath328 , the physical conditions of these mergers remain undetermined .
it is not clear whether this halo would experience more episodes of starburst or quasar activity later on during these mergers .
therefore , the final black hole mass and other properties of this quasar at the present day are deferred to future simulations that follow its evolution to @xmath37 .
the epoch of reionization ( eor ) is an important landmark event in cosmic history that constrains the formation of the first luminous objects @xcite .
the recent results of wmap3 indicate that the universe was 50% reionized at @xmath329 @xcite , while studies of gunn - peterson absorption @xcite suggest that reionization began as early as @xmath305 and ended at @xmath0 @xcite . at present
, it is believed that the reionization sources are star - forming galaxies since there are insufficient quasars at @xmath330 as indicated by the steep quasar luminosity function @xcite .
the galaxy progenitors of the quasar in our simulations underwent extreme and prolonged starbursts before @xmath3 .
less extreme galaxies in this epoch may also have vigorous star formation histories . detecting these galaxies and determining their contribution to reionization will be crucial to understanding the eor ( @xcite ) .
as reviewed by @xcite , recent observations using both broad - band colors ( e.g. , @xcite ) and narrow - band ly@xmath113 emission ( e.g. , @xcite ) have detected @xmath331 galaxies at @xmath0 and a handful at @xmath332 @xcite .
the low luminosity density of galaxies currently detected at @xmath333 seems insufficient to reionize the universe .
however , ongoing surveys with the hst and spitzer telescopes , and future missions such as the _ dark ages z ly@xmath113 explorer _ ( dazle , @xcite ) and the _ james webb space telescope _ ( jwst , @xcite ) will search deeper and further for high - redshift objects , and may eventually unveil ionizing sources in the eor .
we have presented a model that accounts for the smbh growth , quasar activity and host galaxy properties of the most distant quasar observed at @xmath334 , by following the hierarchical assembly of the quasar halo in the standard @xmath2cdm cosmology .
we employ a set of multi - scale simulations that include large - scale cosmological n - body calculations and hydrodynamic simulations of galaxy mergers , and a recipe for black hole growth self - regulated by feedback .
we first perform a coarse n - body simulation in a volume of @xmath35 to identify the largest halo at @xmath37 , which is assumed to be the descendant of the earliest luminous quasar .
we then `` zoom in '' on the halo and resimulate the region with higher resolution sufficient to extract its merging history starting from very high redshift .
the largest halo at @xmath0 reaches a mass of @xmath335 through 7 major mergers between @xmath208 14.46.5 .
these major mergers are again re - simulated hydrodynamically using galaxy models and recipes for star formation , smbh growth , and feedback .
we find that the quasar host galaxy builds up rapidly through gas - rich major mergers , with star formation rates up to @xmath336 , reaching a stellar mass of @xmath265 at @xmath9 .
the black holes grow through gas accretion under the eddington limit in a self - regulated manner owing to their own feedback .
as the galaxies merge , the black holes coalescence to form a dominant black hole , reaching a peak accretion rate of @xmath337 and a mass of @xmath338 at @xmath339 .
feedback from black hole accretion clears away the obscuring gas from the central regions , making the quasar optically visible from @xmath208 7.56 . at the peak of the quasar phase ,
the star formation rate , metallicity , black hole mass , as well as quasar luminosities of the simulated system are consistent with observations of j1148 + 5251 .
our results demonstrate that rare and luminous quasars at high redshifts can form in the standard @xmath2cdm cosmology through hierarchical , gas - rich mergers , within the available cosmic time up to the early epoch of @xmath340 , without requiring exotic processes .
our model should also provide a viable formation mechanism for other distant , luminous quasars .
moreover , we predict that quasar hosts at high redshifts follow similar @xmath10 correlation observed locally as a result of the coeval evolution of the smbhs and host galaxies
. better measurements of black hole masses and host properties with future observations will therefore be crucial to test our prediction .
furthermore , we predict that the progenitors of the distant quasars undergo strong and prolonged starbursts with rates @xmath341 at higher redshifts @xmath285 , that would contribute to the reionization of the universe . detecting these early galaxies and unveiling the epoch of reionization will be an important goal of current and future missions in observational cosmology .
, m. , taniguchi , y. , murayama , t. , nagao , t. , veilleux , s. , shioya , y. , fujita , s. s. , kakazu , y. , komiyama , y. , okamura , s. , sanders , d. b. , oyabu , s. , kawara , k. , ohyama , y. , iye , m. , kashikawa , n. , yoshida , m. , sasaki , t. , kosugi , g. , aoki , k. , takata , t. , saito , y. , kawabata , k. s. , sekiguchi , k. , okita , k. , shimizu , y. , inata , m. , ebizuka , n. , ozawa , t. , yadoumaru , y. , taguchi , h. , ando , h. , nishimura , t. , hayashi , m. , ogasawara , r. , & ichikawa , s .- i .
2002 , , 576 , l25 , r. h. , fan , x. , white , r. l. , strauss , m. a. , narayanan , v. k. , lupton , r. h. , gunn , j. e. , annis , j. , bahcall , n. a. , brinkmann , j. , connolly , a. j. , csabai , i. , czarapata , p. c. , doi , m. , heckman , t. m. , hennessy , g. s. , ivezi , . ,
knapp , g. r. , lamb , d. q. , mckay , t. a. , munn , j. a. , nash , t. , nichol , r. , pier , j. r. , richards , g. t. , schneider , d. p. , stoughton , c. , szalay , a. s. , thakar , a. r. , & york , d. g. 2001 , , 122 , 2850 , m. , stern , d. , giavalisco , m. , ferguson , h. c. , tsvetanov , z. , chornock , r. , cristiani , s. , dawson , s. , dey , a. , filippenko , a. v. , moustakas , l. a. , nonino , m. , papovich , c. , ravindranath , s. , riess , a. , rosati , p. , spinrad , h. , & vanzella , e. 2004 , , 600 , l99 , x. , hennawi , j. f. , richards , g. t. , strauss , m. a. , schneider , d. p. , donley , j. l. , young , j. e. , annis , j. , lin , h. , lampeitl , h. , lupton , r. h. , gunn , j. e. , knapp , g. r. , brandt , w. n. , anderson , s. , bahcall , n. a. , brinkmann , j. , brunner , r. j. , fukugita , m. , szalay , a. s. , szokoly , g. p. , & york , d. g. 2004 , , 128 , 515 , x. , narayanan , v. k. , lupton , r. h. , strauss , m. a. , knapp , g. r. , becker , r. h. , white , r. l. , pentericci , l. , leggett , s. k. , haiman , z. , gunn , j. e. , ivezi , . , schneider , d. p. , anderson , s. f. , brinkmann , j. , bahcall , n. a. , connolly , a. j. , csabai , i. , doi , m. , fukugita , m. , geballe , t. , grebel , e. k. , harbeck , d. , hennessy , g. , lamb , d. q. , miknaitis , g. , munn , j. a. , nichol , r. , okamura , s. , pier , j. r. , prada , f. , richards , g. t. , szalay , a. , & york , d. g. 2001 , , 122 , 2833 , x. , strauss , m. a. , schneider , d. p. , becker , r. h. , white , r. l. , haiman , z. , gregg , m. , pentericci , l. , grebel , e. k. , narayanan , v. k. , loh , y .- s . , richards , g. t. , gunn , j. e. , lupton , r. h. , knapp , g. r. , ivezi , .
, brandt , w. n. , collinge , m. , hao , l. , harbeck , d. , prada , f. , schaye , j. , strateva , i. , zakamska , n. , anderson , s. , brinkmann , j. , bahcall , n. a. , lamb , d. q. , okamura , s. , szalay , a. , & york , d. g. 2003 , , 125 , 1649 , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , richstone , d. , & tremaine , s. 2000 , , 539 , l13 , m. , dickinson , m. , ferguson , h. c. , ravindranath , s. , kretchmer , c. , moustakas , l. a. , madau , p. , fall , s. m. , gardner , j. p. , livio , m. , papovich , c. , renzini , a. , spinrad , h. , stern , d. , & riess , a. 2004 , , 600 , l103 , a. , parry , i. , bland - hawthorn , j. , cianci , s. , king , d. , mcmahon , r. , & medlen , s. 2004 , in ground - based instrumentation for astronomy . edited by alan f. m. moorwood and iye masanori .
proceedings of the spie , volume 5492 , pp .
1022 - 1032 ( 2004 ) . , ed .
a. f. m. moorwood & m. iye , 10221032 , l. , fan , x. , hines , d. c. , shi , y. , vestergaard , m. , bertoldi , f. , brandt , w. n. , carilli , c. l. , cox , p. , le floch , e. , pentericci , l. , richards , g. t. , rieke , g. h. , schneider , d. p. , strauss , m. a. , walter , f. , & brinkmann , j. 2006 , , in press , astro - ph/0608006 , b. , dickinson , m. , ferguson , h. c. , giavalisco , m. , wiklind , t. , stark , d. , ellis , r. s. , fall , s. m. , grogin , n. a. , moustakas , l. a. , panagia , n. , sosey , m. , stiavelli , m. , bergeron , e. , casertano , s. , ingraham , p. , koekemoer , a. , labb , i. , livio , m. , rodgers , b. , scarlata , c. , vernet , j. , renzini , a. , rosati , p. , kuntschner , h. , kmmel , m. , walsh , j. r. , chary , r. , eisenhardt , p. , pirzkal , n. , & stern , d. 2005 , , 635 , 832 , l. , hinshaw , g. , komatsu , e. , nolta , m. r. , spergel , d. n. , bennett , c. l. , barnes , c. , bean , r. , dore , o. , halpern , m. , hill , r. s. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , peiris , h. v. , tucker , g. s. , verde , l. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , astro - ph/0603450 , g. t. , lacy , m. , storrie - lombardi , l. j. , hall , p. b. , gallagher , s. c. , hines , d. c. , fan , x. , papovich , c. , vanden berk , d. e. , trammell , g. b. , schneider , d. p. , vestergaard , m. , york , d. g. , jester , s. , anderson , s. f. , budavari , t. , & szalay , a. s. 2006 , astro - ph/0601558 , d. n. , bean , r. , dore , o. , nolta , m. r. , bennett , c. l. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , barnes , c. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , astro - ph/0603449 , d. n. , verde , l. , peiris , h. v. , komatsu , e. , nolta , m. r. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , page , l. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2003 , , 148 , 175 , v. , white , s. d. m. , jenkins , a. , frenk , c. s. , yoshida , n. , gao , l. , navarro , j. , thacker , r. , croton , d. , helly , j. , peacock , j. a. , cole , s. , thomas , p. , couchman , h. , evrard , a. , colberg , j. , & pearce , f. 2005 , , 435 , 629 , s. , gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , & richstone , d. 2002 , , 574 , 740 , h. , dickinson , m. , stern , d. , eisenhardt , p. r. m. , chary , r .-
giavalisco , m. , ferguson , h. c. , casertano , s. , conselice , c. j. , papovich , c. , reach , w. t. , grogin , n. , moustakas , l. a. , & ouchi , m. 2005 , , 634 , 109 | the discovery of luminous quasars at redshift @xmath0 indicates the presence of supermassive black holes ( smbhs ) of mass @xmath1 when the universe was less than one billion years old .
this finding presents several challenges for theoretical models , because whether such massive objects can form so early in the @xmath2-cold dark matter ( @xmath2cdm ) cosmology , the leading theory for cosmic structure formation , is an open question .
furthermore , whether the formation process requires exotic physics such as super - eddington accretion remains undecided . here , we present the first multi - scale simulations that , together with a self - regulated model for the smbh growth , produce a luminous quasar at @xmath3 in the @xmath2cdm paradigm .
we follow the hierarchical assembly history of the most massive halo in a @xmath4 volume , and find that this halo of @xmath5 forming at @xmath3 after several major mergers is able to reproduce a number of observed properties of sdss j1148 + 5251 , the most distant quasar detected at @xmath6 @xcite . moreover , the smbhs grow through gas accretion below the eddington limit in a self - regulated manner owing to feedback .
we find that the progenitors experience vigorous star formation ( up to @xmath7 ) preceding the major quasar phase such that the stellar mass of the quasar host reaches @xmath8 at @xmath9 , consistent with observations of significant metal enrichment in sdss j1148 + 5251 .
the merger remnant thus obeys similar @xmath10 scaling relation observed locally as a consequence of coeval growth and evolution of the smbh and its host galaxy .
our results provide a viable formation mechanism for @xmath0 quasars in the standard @xmath2cdm cosmology , and demonstrate a common , merger - driven origin for the rarest quasars and the fundamental @xmath10 correlation in a hierarchical universe . |
monolayer graphene has attracted great amount of interest due to its novel electronic properties @xcite .
it has perfect two - dimensional ( 2d ) geometry and a linear band structure near dirac points . since its discovery ,
a substantial body of work has been done on charge transport properties in graphene @xcite .
electron mobility as high as @xmath0 120,000 @xmath1/v@xmath2s at @xmath0 240 k has been measured in suspended graphene @xcite . at high field bias , epitaxial graphene field effect transistors ( fets ) on sic substrates
have exhibited high current drives , exceeding ~ 3 ma/@xmath3 m @xcite .
the current carrying capability of graphene , and the limiting scattering mechanisms have been the focus of recent research .
the earliest studies of high - field current drive suggested that monolayer graphene should be able to deliver currents in the range of ~ 4 ma/@xmath3 m by considering the intrinsic optical phonon scattering @xcite .
shortly thereafter , current saturation was experimentally achieved in fet structures and the saturation current was observed to be proportional to the square root of the channel carrier density @xcite .
surface - optical ( so ) phonon scattering rather than intrinsic optical phonon scattering was suggested as the dominant scattering mechanism for current saturation in graphene on sio@xmath4 substrates .
the effect of hot phonons on limiting the current drive in graphene was then considered , but found to be masked by other scattering mechanisms for graphene on sio@xmath4 substrate @xcite .
subsequent theoretical and experimental work further confirmed that the limiting mechanism for high - field transport in graphene on sio@xmath4 substrates is so phonon scattering @xcite .
the extrinsic scattering from sio@xmath4 could in principle be diminished by using different substrates , or by suspending graphene .
the saturation current in _
intrinsic _ graphene is still poorly understood .
more recently , optical phonon temperatures as high as 1000 k was reported by raman measurements in graphene sheets under high bias @xcite , and the hot phonon effect was observed in experiments @xcite .
these findings show the necessity of considering hot phonon effect on the intrinsic high field transport properties of graphene .
moreover , although electron - electron ( e - e ) scattering degrades the low field transport properties of graphene @xcite , its effect on high field transport has not been studied yet . as a result , the high - field current carrying capacity of intrinsic graphene , considering the hot phonon effect and e - e scattering , is not clearly understood . in this work
, high field carrier transport in intrinsic graphene is investigated by monte - carlo ( mc ) simulations , by including hot phonon effects as well as e - e scattering .
furthermore , transient high - field transport effects , such as velocity overshoot and transit times are investigated . in the next section ,
the various scattering mechanisms used in mc simulations are described . in the third section ,
mc simulation results are presented , followed by discussions of the results . in the last section ,
we draw conclusions and discuss possible future work .
at finite temperature , the _ intrinsic _ scattering mechanisms that affect carrier transport are due to electron - phonon interactions .
the phonon modes that need to be accounted for are acoustic and optical .
the scattering rates are calculated using fermi s golden rule @xcite .
the acoustic phonon scattering rate from state @xmath5 to state @xmath6 is given by @xmath7 where @xmath8 ev is the deformation potential of acoustic phonon scattering , @xmath9 is the boltzmann constant , @xmath10 is the temperature , @xmath11 is the kinetic energy of the carrier and @xmath12 is the wavevector , @xmath13 is the reduced planck constant , @xmath14 cm / s is the fermi velocity in graphene , @xmath15 g/@xmath16 is the 2d mass density of graphene , and @xmath17 km / s is the acoustic phonon ( sound ) velocity in graphene .
@xmath18 is the angle of electron velocity between initial state @xmath5 and final state @xmath6 .
acoustic phonon scattering in graphene is not isotropic , since the acoustic phonon perturbation is long - range , and the wave functions of carriers in graphene have the ` pseudospin ' symmetry .
the angle dependent factor for acoustic phonon scattering is @xmath19 , which prevents backscattering .
the acoustic phonon scattering rate for electron in state @xmath5 is calculated by summing over all possible finals in @xmath20 space and it is given by @xmath21 where acoustic phonon scattering is treated as quasi - elastic , and both emission and absorption are considered . the optical phonon scattering rate from state @xmath5 to state @xmath6 is given by @xmath22,\end{aligned}\ ] ] where @xmath23 is the mode - specific optical deformation potential of graphene , @xmath24 is the optical phonon frequency and @xmath25 is the optical phonon number .
similarly , the optical phonon scattering rate for state @xmath5 is the summation of all scatterings to final states , and it is given by @xmath26 where @xmath27 is for phonon absorption and emission respectively .
in graphene , the zone edge ( ze ) transverse optical ( to ) mode has the strongest coupling with electrons @xcite and the energy of this mode is around @xmath28 mev .
the ` optical ' deformation potential of the mode is @xmath29 ev / @xcite .
the scattering of zone - edge mode is non - isotropic and has the angle dependence factor @xmath30 @xcite .
forward scattering ( @xmath31 ) is prohibited and back scattering ( @xmath32 ) is preferred due to this factor ; the situation is opposite of what happens in acoustic phonon scattering . in our simulation , we also include the zone center ( zc ) optical phonon .
the zc optical phonon has a higher energy , @xmath33 mev .
the deformation potential for the zc phonon is @xmath34 ev / @xcite .
the zone center optical phonon scattering is isotropic @xcite .
= 2.8 in the effect of e - e scattering in graphene with linear @xmath35 dispersion is different from traditional semiconductors that have parabolic band structure ( @xmath36 ) @xcite .
energy and momentum conservation requirements modify the carrier distribution and center of mass motion in a way that the low - field transport properties ( such as carrier mobility ) is strongly damped , especially at low carrier concentrations .
the e - e scattering rate of electron at state @xmath5 with another electron at @xmath37 is given by @xmath38 where @xmath39 and @xmath40 are the finals states .
the coulomb interaction matrix @xmath41 is given by @xmath42 where wave vector @xmath43 and @xmath44 is the dielectric function of 2d graphene .
@xmath45 and @xmath46 are the scattering angles of the two electrons , which has to be taken into account since coulomb scattering is also long range interaction .
similarly , wave vector @xmath47 is for the scattering process of exchange of two electrons .
the two delta function shows the conservation of momentum and energy in the scattering .
the electron scattering rate is calculated by summation of all @xmath48 state and also all the final states ( @xmath39,@xmath49 ) , and it is given by @xmath50 where the occupation of state @xmath48 is the fermi - dirac distribution and the final states are assumed to be empty .
the e - e scattering rate is incorporated into the mc simulation in this work following the procedure of ref.@xcite with one difference : the screening of coulombic forces is treated as temperature - dependent @xcite .
the scattering rates at 300 k are plotted in fig.[scatteringrates ] . in order to capture high - field effects ,
the scattering rate from 0 to 1 ev are calculated for all the scattering mechanisms in intrinsic graphene .
the intrinsic optical phonon scattering rate increases linearly with energy at states higher than @xmath51 .
acoustic phonon scattering rate is one order of magnitude smaller than the optical phonon scattering at high energies .
electrons are accelerated by electric field and populate high - energy states .
therefore , optical phonon scattering is the dominant energy relaxation channel for hot electrons rather than acoustic phonon .
since the ze optical phonon has lower energy and higher scattering rate , it is the dominant optical mode .
the e - e scattering rate at 300 k is higher than optical phonon scattering .
this scattering thermalizes the carrier distribution , driving it towards a fermi dirac distribution .
the net momentum does not change during e - e collisions , but the average velocity of electrons changes due to the linear band structure of 2d graphene . as a consequence
, e - e scattering has an impact on the current .
the e - e scattering rate increases with carrier temperature so that it become more important under high field . from fig.[scatteringrates ] , the phonon scattering rate at high energies exceeds 10 ps@xmath52 .
optical phonons emitted by electrons have to relax to other phonon modes by anharmonic interaction before the heat propagates into the substrate or contacts .
the characteristic lifetime of optical phonon decay into acoustic phonon modes is @xmath01 to 5ps in carbon - based @xmath53 crystals @xcite .
since the optical phonon lifetime is much longer than the phonon generation time ( @xmath0 0.1 ps ) , the optical phonon population in graphene is out of equilibrium .
the fast generation rate and slow decay rate of optical phonons result in a hot phonon effect in graphene . before presenting the numerical results of the ensemble mc calculation , a heuristic picture of the high field transport properties in graphene
is developed in the next section .
a simple model considering the effect of carrier degeneracy provides analytical estimation of the saturation currents for degenerate electron systems .
the electron distribution function in the @xmath20- space in the steady state at high electric fields can be approximated since the optical phonon emission is much faster than competing energy relaxation mechanisms .
the steady state electron distribution function @xmath54 under this assumption is shown in fig.[mymodel](a ) .
the electrons are distributed between two energy contour circles , the high energy circle @xmath55 and the low energy circle @xmath56 .
all electrons move toward @xmath57 direction under the applied electric field .
once electrons approach the high energy circle on the right , they are scattered back to low energy circle by emitting optical phonon instantaneously .
therefore , the energy difference between the two circles is the optical phonon energy , @xmath58 , as indicated in the figure .
the state occupation inside the low energy circle is full so that optical phonon emission by electrons with energy smaller than @xmath59 are prohibited by the pauli exclusion principle . as optical phonon emission continues and scatters the electrons from high energy states back to the low energy states ,
distribution function @xmath54 is squeezed and elongated along the direction of the electric field . the steady distribution function
is called ` streaming ' distribution .
this streaming model captures the effect of carrier degeneracy , which has not been considered in high field - transport in traditional non - degenerate semiconductors .
= 3.6 in the carrier concentration is proportional to the area of the distribution function @xmath60 in @xmath20 space , @xmath61 , where @xmath62 and @xmath63 are spin and valley degeneracies in graphene .
the area of the distribution function is given by @xmath64 from which the dimensions of the distribution function ( @xmath65,@xmath66 ) can be calculated at any carrier concentration .
the current flowing in the direction of the electric field corresponding to the distribution function in fig.[mymodel](a ) is then found by summing the @xmath67directed group velocities of all states inside the shaded region in the @xmath20 space ; the value is found to be @xmath68 where @xmath69 is the electron energy at low energy circle .
this streaming model shows that the saturation current is proportional to the optical phonon frequency and radius of low energy circle . in fig.[mymodel](b ) , the ensemble saturation velocity @xmath70 is plotted against the electron density .
the saturation velocity varies with carrier concentration and is not a constant .
it approaches the fermi velocity in graphene @xmath14 cm / s as @xmath71 , and falls to @xmath72 cm / s at @xmath73/@xmath1 .
the dotted line shows the closely related square root relation @xmath74 proposed in ref.@xcite .
traditionally , the saturation velocity in non - degenerate semiconductors has been considered a constant , e.g. @xmath75 cm / s in si at room temperature @xcite .
the reason for the carrier - density dependent saturation velocity in graphene is the degeneracy of carriers and the pauli exclusion principle .
close to inverse square - root dependence of the effective ensemble saturation velocity on the carrier density has been measured in graphene on sio@xmath4 substrate @xcite .
the model presented here captures the essence of the high - field transport and is in agreement the experimental results , indicating that the carrier degeneracy plays a major role on high field transport in graphene .
however , the model here can stand on a firmer footing by using a numerical monte - carlo approach to find high field transport properties , which is the topic of the next section .
in the mc implementation of the high - field transport problem , we have considered electron energies ranging from 0 to 10 @xmath76 in the graphene conduction band .
the @xmath77space is divided into around @xmath78 cells . in simulation , ` superelectrons ' move in @xmath20 space according to newton s law @xmath79 @xcite . the number of ` superelectrons ' varies from 1000 to be more than @xmath80 , depending on the carrier density .
the current is calculated by summation of the group velocities over all the ` superelectrons ' .
the screening of the e - e coulombic interaction is dependent on the carrier temperature . in the mc simulation
, the electron temperature @xmath81 is updated every time slot ( 0.5 fs ) .
the temperature was calculated from the thermal energy of electron gas @xmath82 , where @xmath83 is wavevector of the ith electron , and the average wavevector of all electrons is @xmath84 .
the electron temperature was calculated by assuming a fermi - dirac distribution , whereby the thermal energy of the electron gas at temperature @xmath81 is given by @xmath85 where @xmath86 is the area of graphene sheet and @xmath87 . in simulation , we assume the carrier density in graphene conduction band is fixed by ignoring generation processes . as a result , fermi level changes with the temperature , and both of them can be calculated once we know the thermal energy and carrier concentration . in order to capture the effect of hot phonons , the phonon @xmath88-space is discretized in the same fashion as the electron @xmath89-space , and the hot - phonon effect is treated as described in @xcite .
the optical phonon distribution function is updated based on the phonon wavevectors during the emission and absorption events .
optical phonon decay into other phonon modes is governed by the rate equation of phonon occupation function @xmath90 - @xmath91 where @xmath92 is the time slot in mc simulation , @xmath93 is the equilibrium bose - einstein distribution , and @xmath94 is the phonon lifetime .
the time slot @xmath92 is around 0.5 fs , corresponding to a scattering rate 2@xmath95 ps@xmath52 , which is higher than the summation of all scattering rates for electrons .
this time slot was updated according to the phonon population in simulation , since the maximum scattering rate varies with the phonon population .
the degeneracy of carriers is explicitly dealt with by the ` rejection technique ' @xcite .
the simulation is continued till a steady state distribution function is obtained .
this procedure yields the distribution functions of electrons @xmath96 and phonons @xmath97 at every time step , which is then used to find ensemble velocities , currents , as well as transient effects such as velocity overshoot as a function of the carrier density and the applied electric field .
= 3.6 in the steady - state currents calculated by the mc simulation for various strengths of electric fields are shown in figure [ mcresults](a ) for two carrier densities .
the curves of different carrier concentrations show similar behavior .
the currents increase with the field and saturate at a field @xmath98 kv / cm .
the degradation of the saturation current due to the hot - phonon effect is clearly seen by comparing the open and solid symbol lines , which is indicated by the arrows .
hot phonon effect is more significant in degradation of the saturation current at higher carrier concentrations . in order to highlight the effect of e - e scattering at high field
, the transport without e - e scattering is also simulated .
the dot - dash lines are the currents including hot phonon effect but without e - e scatterings . without e - e scattering ,
the currents drop at high fields , which is the indication of negative differential resistance ( ndr ) .
ndr has been proposed and observed in simulations for carbon naotubes ( cnts ) and graphene @xcite , due to the linear dispersion of band structure . in graphene ,
the electrons at high energy state move at the same speed with the low energy state , which is the constant fermi velocity in graphene . at high field , more electrons move into high energy states .
but the increasing of velocity is not as rapidly as conventional parabolic materials . as more and more electrons occupy high energy states the backscattering
is increased , which degrades the average velocity of electron gas and leads to ndr
. however , this effect disappears ( or weakened ) if e - e scattering is considered .
the e - e scatterings exchange electrons momentum and energy , diving electrons into fermi - dirac distribution . as a result ,
high energy electrons exchange their momentum and energy with low energy electrons before substantial backscatterings happen . with e - e scatterings , the hot electrons generation and backscattering are weakened and the ensemble velocity of carriers increases with the applied electric field .
the current - field curves from the mc simulation therefore show that a ) the hot phonon effect acts as a strong current limiting mechanism at high carrier densities and b ) e - e scattering gets rid of ndr effect in graphene . in fig.[mcresults](b ) , the saturation current is plotted against the carrier density .
the solid black curve is the analytical result from streaming model ( equation [ jsatofstreamingmodel ] ) for comparison .
the solid squares are the intrinsic saturation currents without the hot phonon effect .
the saturation current is slightly lower than the analytical result due to hot phonon degradation .
saturation current density of @xmath03 a / mm has been measured in epitaxial graphene on sic substrate ( @xmath99 @xmath100 ) @xcite , which is consistent with the mc result .
the current - carrier concentration curves are not linear , indicating the ensemble saturation velocity is not constant , but decreases with increasing carrier density .
the hot phonon effect is better illustrated by the distribution functions of electrons in @xmath89-space , and phonons in @xmath88-space .
the electron distribution functions are plotted in fig.[distributionfunctons](a ) along the @xmath57 axis which is along the direction of the electric field .
the solid line is the electron distribution at steady state , considering hot phonon and e - e scattering .
the peak of the distribution function is close to 1 , showing the high degeneracy of carriers in graphene .
for comparison , we turned off the e - e scattering or hot phonon effect in the simulation .
the distribution function without e - e scattering has a lower peak and is broader , shown by the dot - dash curve .
compared to the solid distribution curve , the distribution without e - e scattering has more electrons at high energy states , but also more backscattered electrons .
this highlights the role of e - e scattering in maintaining the shape of carrier distribution close to fermi - dirac like ( the solid line ) and thus preventing the ndr effect .
similarly , the effect of hot phonons becomes clear when we compare the distribution functions shown by the solid and dashed curve .
the carrier distribution with hot phonons is shifted less along the field direction , so the saturation current is lowered by the hot - phonon effect .
= 3.6 in the effect of hot phonons on the transport can also be seen from the phonon distribution in @xmath88-space .
fig.[distributionfunctons](b ) shows the 2d phonon distribution functions @xmath101 .
the hot phonon distribution has a peak at positive @xmath102 , which corresponds to the fast optical phonon emission by hot electrons .
the phonon occupation number peak exceeds @xmath103 which corresponds to a phonon temperature of @xmath104 k. besides the steady - state saturation currents , the mc simulation procedure also allows us to study transient phenomena in high - field carrier transport . prior
to approaching steady state , carriers moving short distances in traditional semiconductors are known to undergo transient velocity overshoot , attaining higher velocities than what can be attained in the steady state .
the hot electrons accelerated by high field keep moving in @xmath20 space without significant phonon emission during timescales shorter than the mean phonon emission time . as a result
, a velocity peak is achieved in a short period . in high speed electronic devices ,
the channel length is scaled to tens of nanometers .
the carriers travel from source to drain contacts within ps timescales , and velocity overshoot can occur .
this transient velocity overshoot can be utilized to lower the transit time of carriers in graphene devices , thus resulting in an enhanced speed .
fig.[vdvstime](a ) shows the ensemble transient velocity response to electric fields .
velocity overshoot occurs when the electric field is higher than 5 kv / cm .
the higher the field , the higher is the overshoot velocity peak ; it approaches @xmath105 cm / s at a field of 50 kv / cm for the chosen carrier density ( @xmath106/@xmath1 ) . after reaching the peak overshoot velocity ,
the ensemble drift velocity decreases gradually by dissipating excess energy into the graphene crystal in the form of optical phonon vibrations .
the long tail of decreasing velocity is due to the hot phonon effect - optical phonons are slow in decaying into acoustic modes which heat up the lattice .
velocity overshoot occurs within a time window of @xmath107 ps .
if the channel length ( @xmath108 ) is short enough ( @xmath109 ) , the hot carrier energy is dissipated at the drain contact , and the hot phonon effect can be prevented in a short channel device .
thus , the distance @xmath110 that carriers travel in the velocity overshoot period is crucial for the design of high - speed devices .
the transit times @xmath111 given by @xmath112 is calculated from the result in fig[vdvstime](a ) and shown for various lengths in [ vdvstime](b ) . for short channel lengths , the transit time is sensitive to the magnitude of the electric field , since velocity overshoot increases with electric field . for long channel lengths , a major portion of the transit time
is spent traveling at the steady state drift velocity , and the overshoot velocity makes a minor contribution . in the mc simulation , the transit time is not found to be particularly sensitive to the carrier density , which bodes well for devices requiring high currents and high speeds . in a @xmath113 nm long channel device , the transit time is found to be @xmath114 ps , which is exactly the time window for velocity overshoot .
thus , graphene channels @xmath115 nm can fully take advantage of velocity overshoot for enhancing their speed . the peak overshoot velocity , and overshoot time window ( @xmath116 ) is determined by strength of electron - optical phonon interaction , the optical phonon energy , as well as the low - field mobility .
the optical phonon energy and carrier - phonon interaction are intrinsic properties of material , but the peak overshoot velocity peak can be improved by increasing the cleanliness of the substrate or by using gate dielectrics with less impurities .
we point out some limitations of this work , and suggest topics that need investigation in the future . in this work
we have ignored the band to band processes , e.g. inter - band ( zener ) tunneling and e - e scattering induced carrier generation . the carrier concentration does not change at high electric field under the assumption . however , this is not the case in graphene if the fermi level is close to the dirac point .
since the band gap is zero , tunneling and e - e scatterings generate electrons and holes under high fields at very low carrier concentrations . in experimental measurements , the current increases and
never saturates when the fermi level is close to the dirac point in intrinsic graphene ( @xmath117 ) @xcite .
field - induced carrier generation is also the reason of the low on / off ratio ( @xmath118 ) in graphene fets . in experiments , intrinsic properties of graphene
is difficult to achieve .
so phonon scattering limits the high field transport for graphene in close proximity to substrates with low energy phonon modes , like sio@xmath4 .
suspended graphene is free from environmental scattering effects , but the carrier density is not easy to modulate .
one possible way to measure the intrinsic performance is to use sic substrate graphene which has weak so phonon coupling .
we assumed the lattice temperature to be at 300 k in our simulations , although the optical phonon temperature can be much higher than environment temperature 300 k. this assumption could overestimate the saturation current , since the lattice temperature in graphene increases due to the limited thermal conductance of the substrate and contacts .
the hot phonon effect is dominant , so to a certain extent ignoring the rise of lattice temperature is justified .
the phonon lifetime has been measured for graphene on sic substrate and it was found to be around 2.5 ps @xcite . the hot phonon effect can be diminished by introducing isotopic disorder into graphene @xcite .
isotopic disorder introduces phonon modes that are localized in real space , and therefore spread in @xmath119-space , and more phonon modes are thus involved in the cooling of carriers .
the saturation currents can be improved by this disorder engineered lifetime of optical phonons , with direct consequences of higher saturation currents . in conclusion , we investigated the high field transport in intrinsic monolayer graphene .
mc simulation results show that both hot - phonon effect and e - e scattering have strong impacts on the saturation current .
transient velocity overshoot was studied and the short carrier transit time indicates that graphene is well - suited for high speed electronics .
the authors acknowledge kristof tahy and prof .
eric pop s group for discussions and experimental data on high field transport in graphene , the nsf career and eccs awards , and the src nri mind center for financial support . | transport of carriers in two dimensional graphene at high electric fields is investigated by combining semi - analytical and monte - carlo methods .
a semi - analytical high - field transport model based on the high rate of optical phonon emission provides useful estimates of the saturation currents in graphene . for developing a more accurate picture ,
the non - equilibrium ( hot ) phonon effect and the role of electron - electron scattering were studied using monte carlo simulations .
monte carlo simulations indicate that the hot phonon effect plays a dominant role in current saturation , and electron - electron scattering strongly thermalizes the hot carrier population in graphene .
we also find that electron - electron scattering removes negative differential resistance in graphene .
transient phenomenon as such as velocity overshoot can be used to speed up graphene - based high speed electronic devices by shrinking the channel length below 80 nm if electrostatic control can be exercised in the absence of a band gap .
= 1 |
evolutionary dynamics describes the change of populations over time subject to spontaneous mutation , selection , and other random events @xcite .
different phenotypes in the population can emerge spontaneously by mutation , i.e. through errors during reproduction of wild - types .
in many cases , wild - type and mutant individuals are characterised by heritable differences in behavioural traits or strategies @xcite .
selection acts on different phenotypes and changes the population composition .
changes in the state of the environment can alter these selective pressures over time .
time - varying environments are relevant in the evolution of bacterial populations subject to environment modulations by a host @xcite , or varying antibiotic stress .
an illustrative example is the evolution of normal ( wild - type ) cells and resistant ` persister ' cells ( mutant ) .
this was examined by kussell _
et al . _
@xcite , where periods of antibiosis were turned on and off . during times of antibiotic stress the growth rate of normal cells was reduced , but the resistant cells sustain population levels . in addition , acar _
et al . _
@xcite provided further experimental evidence supporting the deterministic model used in @xcite .
more complicated studies of dynamics in switching environments rely on cells ` sensing ' the environment @xcite and on the history or information of the environment during a cell s life @xcite .
these examples illustrate that the assumption of an interaction structure independent of time is not always realistic . at the same time it is largely an open question how complex interactions between phenotypes together with spontaneous changes in the environment influence the evolutionary dynamics .
the interactions of phenotypes in a population can be formalised in an evolutionary game @xcite .
such games can be used to describe conflict over food or territory , cheating in resource allocation , as well as interactions between variants of a gene @xcite . in an evolutionary normal - form game
each individual can be associated with one out of a finite set of strategies .
a payoff matrix quantifies the reward received by a given individual when it interacts with another individual @xcite .
the dynamics of populations interacting in such a game are often described by deterministic replicator equations or similar differential equations @xcite .
while deterministic dynamics are useful to understand the action of selection _ per se _ , more interesting phenomena arise when stochastic effects are taken into account .
a stochastic approach is appropriate
often even strictly required to understand the impact of fluctuations in finite populations @xcite .
deterministic approaches fail to capture effects such as fixation and extinction , or the convergence to a stationary distribution in systems with mutation @xcite .
there is an increasing body of literature on stochastic evolutionary games .
for example , analytical results for the probabilities of a single mutant to reach fixation have been obtained @xcite .
however , most of this existing work focusses on games played in a fixed environment ; the underlying payoff matrix itself remains unchanged in time .
this assumption may not be appropriate in cases where external factors influence the environment .
external fluctuations in evolutionary games have been previously introduced by adding extrinsic noise to continuous model parameters @xcite , or by letting strategy space itself vary in time @xcite .
environmental variability has also been the subject of investigation in predator - prey models @xcite . in this article
we explore different theoretical approaches that allow calculations of fixation probabilities and mean fixation times of a rare mutation , under fluctuating environmental conditions .
we use a generic birth - death framework , as described in section [ sec : model ] , and our results thus apply to a wide class of population dynamics . in section [ sec : gen ] the theory is developed for an environment which can transition between an arbitrary number of discrete states , and we expand on the two - environment scenario in section [ sec : twostate ] . to illustrate our theoretical results we study the fixation properties in an evolutionary game that stochastically switches between a coexistence game and a coordination game in section [ sec : games ] .
we determine environmental conditions under which the success of a rare invading mutant is maximal .
this is seen to occur at a non - trivial combination of switching rates . for the case
in which mutations occur during the dynamics , as described in section [ sec : mut ] , we explore how the stationary distribution of the population changes in fluctuating environments . we derive approximations for the stationary distribution , valid for a large range of switching rates .
we summarise our findings in section [ sec : sum ] and put them into context .
we seek to model evolutionary dynamics in finite populations of two species that are subject to environmental changes .
the changes in the environment are such that at any given point in time , the system can be in one of a finite set of environmental states . the state that the environment is in determines the details of the birth and death dynamics .
we focus on two cases : first , in the absence of mutations in the dynamics , we derive laws to predict the probability and mean time of the fixation of a mutant . fixation describes the event in which mutants take over the population as opposed to going extinct . fixation or extinction are the only two outcomes of dynamics of a rare mutant in a finite population @xcite . in figure
[ fig : fig1 ] we show a basic sketch of this scenario in which the two monomorphic states of the population are absorbing .
second , we study the case when mutations occur in the dynamics .
there are then no absorbing states .
instead , the dynamics converges to a stationary distribution . , in one time - step it may transition to @xmath0 or @xmath1 , or remain at @xmath2 .
the states @xmath3 and @xmath4 are absorbing in both environments ( no arrows out of these states ) .
the transition probabilities ( birth / death rates ) are dependent on the state of the environment , indicated by solid vs. dashed arrows in environments @xmath5 and @xmath6 , respectively .
the environment switches from state @xmath5 to @xmath6 with probability @xmath7 in any one time - step .
the quantity @xmath8 represents the probability of fixation , as discussed in section [ sec : gen ] .
( online version in colour . ) , scaledwidth=75.0% ] we consider populations consisting of a fixed number of @xmath9 individuals .
each individual can be of one of two types , @xmath10 or @xmath11 , which we refer to as ` mutant ' and ` wild - type ' , respectively .
the population is well mixed ; every individual can interact with any other individual .
the state of the population is fully characterised by the number , @xmath2 , of individuals of type @xmath10 .
the remaining @xmath12 individuals are of type @xmath11 .
we furthermore assume that at any one time the environment can be in one of @xmath13 discrete states , labelled @xmath14 , where @xmath15 is the space of states of the environment ( @xmath16 ) .
hence the state of the entire system at any time is given by the pair @xmath17 .
the discrete - time birth - death dynamics of the population for a given environment , @xmath5 , is then specified by the transition probabilities @xmath18 and @xmath19 of a one - step process .
specifically , if the system is in state @xmath17 the population transitions to state @xmath1 in the next time step with probability @xmath18 .
similarly the state of the population in the next time step is @xmath0 with probability @xmath19 .
these transitions are shown as black arrows in figure [ fig : fig1 ] .
with probability @xmath20 the population remains in state @xmath2 .
we always assume that @xmath21 and @xmath22 for all @xmath17 . with the exception of section [ sec : mut ] we will always assume that the states @xmath3 ( all-@xmath11 ) and @xmath4 ( all-@xmath10 ) are absorbing ( @xmath23 and @xmath24 for all @xmath14 ) . in the absence of further mutation events a type , once absent , can never be re - introduced
. if mutations are present in the dynamics , then the states @xmath3 and @xmath4 are no longer absorbing and the system converges to a unique , non - trivial stationary state .
we consider this case in section [ sec : mut ] . in our approach
the environment evolves from one state to another _ independently _ of the state of the population .
this simplification still captures a wide array of natural scenarios . in the discrete - time setup
we take the dynamics of the environment as a simple markov chain , described by the transition matrix @xmath25 of size @xmath26 .
the entry @xmath7 represents the probability that the environment changes to state @xmath6 in the next time - step , if it is currently in state @xmath5 , as shown in figure [ fig : fig1 ] .
the matrix @xmath27 is a stochastic matrix , @xmath28 for all @xmath14 .
a switch of the environment effectively modifies the birth - death dynamics in the population .
we do not specify the exact type of interaction at this point , but keep the rates @xmath29 general .
let us first consider a discrete - time evolutionary process .
if the system is in state @xmath17 at a given time , it may transition to @xmath30 possible states in any one time - step .
these are given by @xmath31 , @xmath32 and @xmath33 , where @xmath34 can be any of the @xmath13 states of the environment . if we write @xmath35 for the probability of a transition from @xmath17 to @xmath36 , we have @xmath37 no transitions from @xmath17 to @xmath36 can occur when @xmath38 . in this setup
the birth - death probabilities are determined by the state of the environment at the beginning of the discrete time - step .
the fixation probability , @xmath8 , is the probability that the system ends up in the absorbing state with @xmath9 individuals of type @xmath10 , conditioned on initial state @xmath17 .
the probability of fixation of a single mutant , @xmath39 , is of particular interest @xcite .
it is briefly illustrated in figure [ fig : fig1 ] . in our scenario with switching between @xmath13 environmental states , following the lines of refs .
@xcite , the following balance equation for the fixation probabilities can be found @xmath40 .
\label{eq : fixprobbalance}\ ] ] this is to be solved along with the boundary conditions @xmath41 and @xmath42 for all @xmath43 . to obtain a formal solution ,
we introduce @xmath44 , or in matrix form @xmath45 .
the vectors @xmath46 and @xmath47 each have @xmath13 components .
the boundary conditions @xmath41 and @xmath42 translate into @xmath48 and @xmath49 for all @xmath14 . with this notation we have @xmath50 using @xmath51 , we obtain @xmath52 , \label{eq : fixprobbalance3}\ ] ] where @xmath53 .
we stress that the calculation of fixation probabilities and mean fixation times using this formalism requires the matrix @xmath27 to be invertible .
we comment on this further in the context of a specific example in sec .
[ sec : twostate ] . to keep the notation compact we define the variable @xmath54 . using @xmath48 , we have @xmath55 . with this notation
we can write equation ( [ eq : fixprobbalance3 ] ) in the following form @xmath56_\sigma , \label{eq : fixprobbalance4}\ ] ] where @xmath57 is the @xmath26 identity matrix .
this relation expresses the vector @xmath58 in terms of the vectors @xmath59 .
we can therefore express all vectors @xmath60 ( i=@xmath61 ) in terms of @xmath62 .
the constraint @xmath63 then determines @xmath62 self - consistently .
we note that the resulting set of equations is linear in the set @xmath64 .
hence a solution can be obtained in closed form , in principle . in practice one
inverts the linear system using one of the standard algebraic manipulation packages .
once @xmath62 has been found , the other components @xmath60 , with @xmath65 , can be computed via equation ( [ eq : fixprobbalance4 ] ) .
one then uses @xmath66 to find the fixation probabilities starting with @xmath2 individuals of type @xmath10 in environment @xmath5 , @xmath67 .
we note here that algebraically inverting the linear system ( [ eq : fixprobbalance4 ] ) when @xmath9 is large is difficult due to the very large number of terms in the corresponding expressions .
thus , at present , this theory is limited computationally to relatively small @xmath9 . in the case of a single environment , @xmath68 ,
the matrix @xmath27 is simply the @xmath69 identity matrix , and equation ( [ eq : fixprobbalance4 ] ) simplifies to the well - known result for discrete - time birth - death processes @xcite .
we write @xmath70 for the expected number of time - steps taken to reach any one of the two absorbing states , given that the system is started in state @xmath17 .
these fulfil the boundary conditions @xmath71 . with these definitions
we find the following relation @xmath72 + 1 .
\label{eq : uncondbalance1}\ ] ] introducing the variable @xmath73 , we have @xmath74 and with the notation @xmath75 we arrive at @xmath76_\sigma -\frac{1}{\omega^+_{i,\sigma}}. \label{eq : uncondbalance3}\ ] ] this relation allows one to express all vectors @xmath77 ( @xmath65 ) in terms of @xmath78 . the constraint
@xmath79 then determines @xmath78 , and the mean unconditional fixation times are computed using @xmath80 .
we write @xmath81 for the mean fixation time conditioned on absorption in the all-@xmath10 state , given that the system is initially in state @xmath17 . to find this conditional fixation time , we proceed along similar lines as before . introducing the variable @xmath82 , which has boundary conditions @xmath83
, the following balance equation can be found , @xmath84 + \phi_{i,\sigma}. \label{eq : condbalance1}\ ] ] we note that equation ( [ eq : uncondbalance1 ] ) and equation ( [ eq : condbalance1 ] ) appear to be very similar , but the difference is more than just a global pre - factor @xmath8 ; each term in the expression has different indices @xmath2 and @xmath5 .
introducing the variable @xmath85 , we have @xmath86 and introducing @xmath87 we arrive at @xmath88_\sigma-\frac{1}{\omega^+_{i,\sigma}}\phi_{i,\sigma}. \label{eq : condbalance3}\ ] ] the set @xmath89 can then be found using an approach similar to the one described above .
results for the mean conditional fixation time can then be obtained using @xmath90 .
in any of the elementary time steps of the above discrete - time model , _ both _ the state of the population ( @xmath2 ) and the state of the environment ( @xmath5 ) can change .
we next consider a continuous - time setup .
there are two types of discrete events that may occur at any time : ( i ) the state of the environment may change , or ( ii ) a birth - death event may occur .
the rate ( per unit time ) with which a transition from state @xmath5 to state @xmath6 occurs is denoted by @xmath91 .
the occurrence of these events is independent of the state of the population .
the rate with which a birth - death event of the type @xmath92 occurs is @xmath93 , if the environment is in state @xmath5 . the rate with which @xmath94 occurs is @xmath95 .
we write @xmath96 for the probability to find the system in state @xmath17 at time @xmath97 , and find the master equation @xmath98 .
\label{eq : master}\end{aligned}\ ] ] we write @xmath99 for the probability to find the system in state @xmath36 a period of time @xmath97 after it has been started in state @xmath17 .
the corresponding backward master equation @xcite reads @xmath100 + w^-_{i,\sigma } \bigl[q_{j,\sigma';i-1,\sigma}(t ) - q_{j,\sigma';i,\sigma}(t)\bigr ] \nonumber \\ & & + \sum_{\sigma''}m_{\sigma\to\sigma''}\bigl[q_{j,\sigma';i,\sigma''}(t)-q_{j,\sigma';i,\sigma}(t)\bigr ] .
\label{eq : backwardmaster}\end{aligned}\ ] ] we define @xmath101 as the probability that the system has reached fixation in the all-@xmath10 state a period of time @xmath97 after the dynamics has been started in state @xmath17 .
this includes fixation before time @xmath97 . by setting @xmath102 and summing over @xmath6 in equation ( [ eq : backwardmaster ] )
we obtain @xmath103 + w^-_{i,\sigma}\bigl[\varphi_{i-1,\sigma}(t)-\varphi_{i,\sigma}(t)\bigr]\nonumber \\ & & + \sum_{\sigma''}m_{\sigma\to\sigma''}\bigl[\varphi_{i,\sigma''}(t)-\varphi_{i,\sigma}(t)\bigr ] .
\label{eq : backwardmastersum1}\end{aligned}\ ] ] the fixation probabilities are found as @xmath104 , and they can be obtained by setting the time derivative in equation ( [ eq : backwardmastersum1 ] ) to zero . introducing @xmath105 , one finds @xmath106 , \label{eq : fixprobcont}\ ] ] with @xmath107 and where we have used @xmath41 to write @xmath108 . for the case of a single environment
the second term on the right - hand side vanishes and one recovers again the well known results in single environments @xcite .
equation ( [ eq : fixprobcont ] ) has the same general structure as equation ( [ eq : fixprobbalance4 ] ) . keeping in mind that @xmath42 , the fixation probabilities can hence be found by applying the approach outlined in section [ ssec : discretefp ] . calculating the mean fixation time using a diffusion approximation @xcite is not appropriate for our model .
the environmental switching process has no continuum limit .
instead we work with the backward master equation ( [ eq : backwardmaster ] ) and adapt the calculation outlined by antal and scheuring @xcite .
we introduce @xmath109 $ ] , the probability that the system has reached fixation in either of the two absorbing states a period of time @xmath97 after being started in @xmath17 .
again this includes fixation before @xmath97 .
we then have @xmath110 for the probability density to reach fixation exactly at time @xmath97 . from the backward master equation ( [ eq : backwardmaster ] ) we find @xmath111 + w^-_{i,\sigma } \bigl[\rho_{i-1,\sigma}(t)- \rho_{i,\sigma}(t)\bigr ] \nonumber \\ & & + \sum_{\sigma'}m_{\sigma\to\sigma'}\bigl[\rho_{i,\sigma'}(t)-\rho_{i,\sigma}(t)\bigr ] .
\label{eq : backwardmastersum2}\end{aligned}\ ] ] the mean unconditional fixation time is then found via @xmath112 , from which we find @xmath113 + w^-_{i,\sigma } \bigl[t_{i-1,\sigma}- t_{i,\sigma}\bigr ] \nonumber \\ & & + \sum_{\sigma'}m_{\sigma\to\sigma'}\bigl[t_{i,\sigma'}-t_{i,\sigma}\bigr ] .
\label{eq : uncondcont}\end{aligned}\ ] ] a similar iterative equation can be found for the mean fixation time conditioned on absorption in the all-@xmath10 state .
the only difference is the integral of @xmath114 $ ] is given by the fixation probability @xmath8 , and that @xmath115 .
the mean conditional fixation times , @xmath81 , therefore fulfil the relation @xmath116
+ w^-_{i,\sigma } \bigl[\phi_{i-1,\sigma } t_{i-1,\sigma}^a- \phi_{i,\sigma } t_{i,\sigma}^a\bigr ] \nonumber \\ & & + \sum_{\sigma'}m_{\sigma\to\sigma'}\bigl[\phi_{i,\sigma ' } t_{i,\sigma'}^a-\phi_{i,\sigma } t_{i,\sigma}^a\bigr ] .
\label{eq : condcont}\end{aligned}\ ] ] structurally , equations ( [ eq : uncondcont ] ) and ( [ eq : condcont ] ) are of the same form as the corresponding equations for the discrete - time model , and so they can be solved using an analogous procedure . in continuous time , however , the solution procedure no longer relies on the invertibility of the matrix @xmath117 of switching rates between the states of the environment .
this is because we have split up the birth - death dynamics and the changes of the environment into separate events that occur successively .
we now focus on the case of environments which can be in one of two possible states , i.e. @xmath118 .
we label the two states as @xmath119 ( @xmath120 ) .
we focus on the discrete - time scenario .
the matrix @xmath27 can then be written as @xmath121 where the quantity @xmath122 ( @xmath123 ) is the probability that the state of the environment switches to @xmath124 in a given time - step if it is in state @xmath5 at the beginning of this step .
we recall that our theoretical results require the inversion of @xmath27 . excluding the case when @xmath125 vanishes ,
this inversion can be carried out straightforwardly , @xmath126 for the case @xmath127 , we have verified that there is no anomalous behaviour of simulation results .
the general result of equation ( [ eq : fixprobbalance4 ] ) now reduces to the recursion @xmath128 the fixation probability is obtained from the set @xmath129 via @xmath130.\label{eq : fixprob2state}\ ] ] similarly , equations ( [ eq : uncondbalance3 ] ) and ( [ eq : condbalance3 ] ) reduce to @xmath131 and @xmath132 respectively .
the mean unconditional and conditional fixation times are then found as @xmath133 , \\
t_{i,\sigma}^a&= & \frac{1}{\phi_{i,\sigma}}\sum_{j=1}^i\left[\frac{1-p_{-\sigma}}{\delta}\eta_{j,\sigma}-\frac{p_\sigma}{\delta}\eta_{j,-\sigma}\right ] .
\label{eq : condfixtime2state}\end{aligned}\ ] ] the environmental change is fast if the environmental states are short - lived , i.e. much shorter than the mean fixation time in either environment .
then we expect the population dynamics to be controlled by a set of _ effective _ transition probabilities , i.e. weighted averages of the original transition probabilities in the different environmental states .
the weights are given by the fraction of time spent in each environmental state .
as the dynamics of @xmath5 follows a simple a telegraph process @xcite , the asymptotic fraction of time spent in the state @xmath5 is @xmath134 for @xmath135 . using this ,
the effective transition probabilities are given by @xmath136 we note that @xmath122 is the probability that in a given time - step the environment switches _ from _ state @xmath5 to @xmath124 .
hence the time spent in state @xmath5 decreases with increasing @xmath122 if @xmath137 is held fixed .
we anticipate that expression ( [ eq : effectiveprobs ] ) can formally be derived by introducing a relative scaling parameter between the switching probabilities and the birth - death probabilities , and by then taking a suitable limit in which the time scales of both processes are widely separated .
we do not explore this route further here . in this approximation the dynamics of the population are mapped to a simple birth - death process on the set @xmath138 with absorbing states @xmath3 and @xmath4 . for such processes explicit expressions for the fixation probabilities and mean fixation times are known @xcite . in the fast - switching limit
we propose the following approximation for the fixation probability , @xmath139 we have here written @xmath140 .
the corresponding approximations for the mean unconditional and conditional fixation times of a single mutant are @xmath141 respectively .
these expressions _ exactly _ describe the fixation properties of a birth - death system with the effective transition probabilities ; the nature of our approximation is to assume that the birth - death process in quickly changing environments can be described by the effective transition probabilities in equation ( [ eq : effectiveprobs ] ) . finally we note that this theory is _ independent _ of the invertibility of the switching matrix @xmath27 .
as a direct application of the general theory we have developed , we now consider evolutionary game dynamics in well - mixed , finite populations .
any of the @xmath9 individuals can be of one of two types , @xmath10 or @xmath11 .
we limit the discussion to two - player games , but the extension to multi - player games ( e.g. @xcite ) is straightforward . at any point in time
the environment is in one of two discrete states ( @xmath123 ) .
this state fluctuates in time as specified above .
the interaction between individuals is characterised by the payoff matrix @xmath142 the subscript @xmath5 indicates the dependence on the environment .
the matrix focuses on the column player : a type-@xmath10 individual encountering another of its kind receives @xmath143 , and it receives @xmath144 when interacting with a type-@xmath11 individual . in turn ,
an individual of type @xmath11 interacting with an individual of type @xmath10 obtains @xmath145 , and @xmath146 is the payoff for each individual if they are both of type @xmath11 . if the environment is in state @xmath5 , and if there are @xmath2 individuals of type @xmath10 in the population and @xmath12 individuals of type @xmath11 , the expected payoffs for each type of player are @xmath147 the reproductive fitness of any individual is a function of the individual s payoff in the evolutionary game .
we use an exponential mapping @xcite , @xmath148 this is a common natural choice in which fitness is never negative and monotonically increasing with payoff .
any other functional form of the payoff - to - fitness mapping with these two properties would be equally appropriate .
the constant parameter @xmath149 is the so - called intensity of selection .
based on this definition of fitness we model the evolutionary dynamics by the update rules of the moran process @xcite , which has been widely used in evolutionary game theory @xcite .
the moran process represents a simple birth - death process in which the population size remains constant , and by construction it has absorbing states at @xmath3 and @xmath4 . in a discrete - time setting ,
the frequency - dependent moran process is specified by the transition probabilities @xcite @xmath150 where @xmath151/n$ ] is the average fitness in the population .
we note that the framework of the previous section can be applied to microscopic evolutionary dynamics other than the moran process .
this includes , for example , pairwise comparison processes @xcite , or cases with constant selection in any one environment .
rare mutations can introduce a previously absent strategy into the population .
typically there is only one individual of this novel type initially .
we say that @xmath11 is the resident type , and that @xmath10 is the invading mutant type .
all results in this section are based on the initial condition @xmath152 .
we chose @xmath153 for the payoff matrix . the type of game is then determined by the off - diagonal terms .
we chose @xmath154 and @xmath155 , where @xmath156 are real - valued parameters .
thus we have the payoff matrix @xmath157 our parametrisation does not span the entire space of all @xmath158 games , but it covers some of the most common types ( see below ) .
there exist three general types of two player - two strategy evolutionary games .
first , for the coexistence game @xmath159 and @xmath160 , selection drives the population away from the absorbing boundaries .
second , for @xmath161 and @xmath162 , the population dynamics exhibits bi - stability .
this is also known as a coordination game ; selection drives the population towards the monomorphic states . in both cases there
exists an internal point in frequency space for which the direction of selection changes its sign , i.e. at which the gradient of selection is zero .
this point can be calculated by solving @xmath163 ( or equivalently @xmath164 ) for @xmath2 , and broadly speaking it is determined by the relative magnitudes of @xmath165 and @xmath166 .
third , for @xmath167 ( or @xmath168 ) type @xmath10 ( or type @xmath11 ) always has the higher fitness irrespective of the composition of the population .
this type is then always favoured by selection , which never changes direction . for the remainder of this article we focus on switching between coexistence and coordination games .
more precisely we choose @xmath169 and @xmath170 in ( [ eq : payoffmatrix2 ] ) .
the coexistence game corresponds to @xmath171 and the coordination game to @xmath172 . .
white background corresponds to the environment being in the @xmath171 coexistence state , while the shaded background corresponds to the @xmath172 coordination state .
dashed line is the location of the point at which selection balances , which is the same in both states of the environment .
( b ) gradient of selection in the @xmath171 coexistence state ( @xmath173 ) .
solid circle shows location of the point of selection balance , and arrows indicate the direction and magnitude of flow towards this point .
( c ) gradient of selection in the @xmath172 coordination state ( @xmath174 ) .
empty circle shows location of the point of selection balance , and arrows indicate the direction and magnitude of flow away from this point . for the realisation in panel ( a ) and the selection bias shown in ( b ) and ( c ) , the payoff matrix elements are @xmath153 , @xmath175 and @xmath176 , the system size is @xmath177 , the selection intensity is @xmath178 , and the switching probabilities are @xmath179 and @xmath180 .
( online version in colour . ) ] in figure [ fig : fig2](a ) we show a sample trajectory of a simulation in which a single mutant reaches fixation .
the gradient of selection , @xmath181 , for the two games is shown in figures [ fig : fig2](b ) and [ fig : fig2](c ) . during periods
when the environment is in the coexistence state ( light background ) the population fluctuates about the selection - balance point ( dashed line ) , and during periods when the environment is in the coordination state ( shaded background ) the population is driven away from the selection - balance point . in the final period in the coordination state
the mutant is driven to fixation . and circles to @xmath182 ) for fixed @xmath183 , along with the exact theoretical results ( solid lines ) from equation ( [ eq : fixprob2state ] ) , and the effective theoretical result ( dashed line ) of equation ( [ eq : fixationprobtelegraph ] ) .
inset panels show fixation probabilities from equation ( [ eq : fixprob2state ] ) over all combinations of @xmath184 and @xmath185 .
left inset panel : initial condition @xmath186 .
right inset panel : @xmath182 .
the horizontal lines correspond to the data shown in the main panel .
( b ) mean conditional fixation time ( in generations ) of a single mutant as a function of switching probabilities .
the main panel shows simulation results , as described above , for fixed @xmath183 , along with the exact theoretical results ( solid lines ) of equation ( [ eq : condfixtime2state ] ) , and the effective theoretical result ( dashed line ) of equation ( [ eq : condfixtimetelegraph ] ) .
inset panels show mean conditional fixation times from equation ( [ eq : condfixtime2state ] ) over all combinations of @xmath184 and @xmath185 .
left inset panel : initial condition @xmath186 .
right inset panel : initial condition @xmath182 .
the horizontal lines correspond to the data shown in the main panel .
the payoff matrix elements are @xmath153 , @xmath175 and @xmath176 , the system size is @xmath187 , and the selection intensity is @xmath188 .
( online version in colour . ) ] in figure [ fig : fig3 ] we show the effect of switching the environment on the fixation dynamics .
we choose @xmath189 . by equating the reaction probabilities ( i.e. , setting @xmath190 ) , or equivalently equating the expected payoffs in equation ( [ eq : meanpayoff ] ) , and looking at leading order terms in @xmath9
, the gradient of selection is seen to change sign at @xmath191 .
this point is closer to the extinction state of the mutant ( @xmath3 ) than to the fixation state ( @xmath4 ) .
we next describe the key observations we make from these results , before we turn to their interpretation .
_ fixation probability ( figure [ fig : fig3](a ) ) : _ + the fixation probability in this example depends non - trivially on the rates with which the environment switches states ; we find an optimal combination of switching rates , @xmath192 , for which fixation of a single mutant is most likely .
the fixation probability is dependent on the initial state of the environment for @xmath193 . _ fixation time ( figure [ fig : fig3](b ) ) : _ + mean conditional fixation times show very little dependence on the initial state of the environment .
the fixation time is small for @xmath194 when the environment is found mostly in the coordination game , and large when the environment is mostly in the coexistence state .
_ validity of the theoretical approach : _ + as seen in both panels of figure [ fig : fig3 ] the theoretical predictions of equations ( [ eq : fixprob2state ] ) and ( [ eq : condfixtime2state ] ) , indicated by solid lines , are in convincing agreement with simulation data .
theoretical results from the model with effective transition rates ( section [ ssec : telegraph ] ) reproduce the simulation data qualitatively .
quantitative agreement is obtained in the limit of large switching rates , but unsurprisingly , there are systematic deviations when switching is slow .
we now proceed and give an intuitive explanation for the observed effects .
mean conditional fixation time is reduced as more time is spent in coordination environment : _ + the behaviour of the fixation time can intuitively be understood from the deterministic gradient of selection of the two games ( figure [ fig : fig2](b ) and [ fig : fig2](c ) ) .
if fixation happens , it will generally be quicker in the coordination game than in the coexistence game @xcite .
this is due to the adverse selection bias in the coordination game at low mutant numbers ( see figure [ fig : fig2](c ) ) .
the more time the system spends in this region of adverse selection the less likely it is for the mutant to reach fixation .
thus if fixation happens in a coordination game then it happens fast . in the coexistence game on the other hand the direction of selection is towards the balance point , as shown in figure [ fig : fig2](b ) .
the system can ` afford ' to spend significant time in the region of small mutant numbers and still reach fixation eventually even after repeated excursions throughout frequency space .
there is thus no need for fixation to occur quickly , and conditional fixation times can be long .
these observations make it plausible that the mean conditional fixation time will generally decrease when less time is spent in the coexistence game , which is exactly what we find in figure [ fig : fig3](b ) .
we have tested other choices of the parameters @xmath165 and @xmath166 for which the two games are a coexistence game and a coordination game , and we find that the behaviour of the mean conditional fixation times is robust under these changes .
_ mean conditional fixation time is largely independent of the initial state of the environment : _
+ systems started in the coordination environment will tend to reach extinction relatively quickly due to initial adverse selection , unless the environment switches to the coexistence state early on .
thus the sample of runs that reach fixation started from the coordination - game environment will be dominated by runs in which the environment switches soon after the start of the run .
then we expect that the value of the mean conditional fixation time is close to the one obtained when starting in the coexistence game .
_ dependence of fixation probability on initial state of the environment : _ + the data in figure [ fig : fig3](a ) show that initiating the dynamics in the coexistence game favours fixation of the mutants for @xmath195 ( when @xmath183 is fixed ) , however above this threshold the initial state of the environment has relatively little effect .
the reason for this is as follows .
when starting in the coordination game , selection pushes the mutant towards extinction .
hence fixation is more likely if the initial state is the coexistence fitness landscape . above @xmath196 ,
the switching process of the environment is too fast for the initial condition to have any significant effect on the population dynamics .
it is this regime in which we expect the effective description ( section [ ssec : telegraph ] ) to approximate the system well .
this is indeed confirmed in figure [ fig : fig3 ] , the theoretical prediction of the effective theory , equation ( [ eq : fixationprobtelegraph ] ) , agrees well with our simulation results in this fast - switching region .
at which @xmath39 is maximal given @xmath183 as a function of @xmath165 and @xmath166 .
remaining parameters are @xmath188 and @xmath187 .
( online version in colour . ) ] _ behaviour of fixation probability depends on location of the selection - balance point : _ + if the environment is fixed to the coexistence - game state , fixation is more likely the closer the point of selection balance is to the fixated state , see figure [ fig : fig4](a ) .
the location of this balance point is approximated by @xmath197 , and so the fixation probability increases as @xmath198 is decreased . in a fixed coordination - game environment
the reverse is the case .
the range of adverse selection is to the left of the balance point , and so fixation is less likely the closer the point of selection balance is to the fixated state . for @xmath199 ,
i.e. a selection - bias point close to @xmath3 , we therefore expect that the fixation probability will increase the more time is spent in the coordination - game environment , i.e. @xmath39 is an increasing function of the probability @xmath184 with which the system leaves the @xmath171 state ( coexistence game ) .
this is indeed what we find in simulations ( data not shown ) . for @xmath200 ,
i.e. @xmath201 close to @xmath4 , the reverse is the case .
fixation is more likely in the coexistence game ( @xmath171 ) , and the fixation probability is hence a decreasing function of @xmath184 at fixed @xmath185 . although we do not show the data here , this is again confirmed in simulations . for @xmath202
the situation is less clear .
the fixation probability will be comparable in both games if the environment is frozen .
two effects here conspire to produce a non - trivial outcome : * consider the case in which the system is mostly in the coordination - game state , i.e. @xmath203 .
it is plausible that an occasional switch to a coexistence game will make fixation more likely than in a constant coordination game .
this is because the coexistence - game environment pushes the system away from extinction at low mutant numbers . in the regime of @xmath203
we thus expect the fixation probability to increase as @xmath184 is lowered .
in other words , @xmath204 is a decreasing function at large @xmath184 . * similarly , if the system is mostly in the coexistence - game environment ( @xmath205 ) , short periods of time in the coordination game can make fixation more likely .
this is because selection at large mutant numbers is directed towards fixation in the coordination game . at @xmath205
we expect @xmath39 to be an increasing function of @xmath184 .
these two effects taken together generate a maximum of the fixation probability at intermediate values of @xmath206 , which is exactly what we find in figure [ fig : fig3](a ) .
we would like to stress that the effect ( i ) is only present provided the selection - balance point is not too close to the extinct state .
the phenomenon discussed under ( ii ) is only present if the selection - balance point is not too close to the fixated state .
if the balance point is located too close to either boundary the corresponding effect will be suppressed and the remaining effect dominates .
one then finds monotonically increasing or decreasing dependences @xmath204 . to confirm our picture we varied the payoff parameters @xmath165 and @xmath166 , and find the value of @xmath184 that maximises fixation probability for a given @xmath183 , as a function of @xmath165 and @xmath166 in figure [ fig : fig4](b ) .
the point of selection balance is approximately @xmath207 ( up to system - size corrections ) .
the presence of diagonal structures in figure [ fig : fig4](b ) shows that the behaviour of the fixation probability is dependent on the location of the selection - balance point .
if this point is close to the fixation state @xmath4 ( @xmath200 , bottom - right in figure [ fig : fig4](b ) ) , then the fixation probability is maximal for vanishing @xmath184 .
if this point is close to the extinction state ( @xmath199 , top - left in figure [ fig : fig4](b ) ) , then the fixation probability is maximal for large @xmath184 . for intermediate locations of the selection balance point
( @xmath202 ) fixation is maximised at a non - trivial combination of environment states .
the features observed in figure [ fig : fig3 ] , i.e. the peak in the fixation probability and shape of the mean conditional fixation time as a function of @xmath184 , are found to be robust when the system size is increased .
fixation probabilities generally decrease with system size but the observed peak becomes sharper .
the mean conditional fixation times in a frozen coexistence environment scale exponentially with @xmath9 , whereas they scale as the logarithm of @xmath9 in the coordination environment @xcite .
we observe these scalings in our system , with the mean conditional fixation time increasing exponentially with @xmath9 for small @xmath184 , and increasing sub - linearly with @xmath9 for large @xmath184 .
we now consider systems with mutations occurring during the dynamics .
this removes the possibility of fixation and extinction .
the combination of mutation , selection , and noise can lead to non - trivial stationary states .
we introduce mutation by modifying the discrete - time transition probabilities of equation ( [ eq : transitionprob ] ) and now use @xmath208 where @xmath209 is the mutation rate .
the transition probabilities @xmath210 are now non - zero , and so the states @xmath3 and @xmath4 are no longer absorbing . the stationary probability @xmath211 of finding the system in state @xmath17 ( @xmath212 , @xmath14 ) is obtained as the solution of the balance equation @xmath213 .
\label{eq : statdist}\ ] ] this equation is of the form @xmath214 , and it is solved by finding the eigenvector corresponding to the eigenvalue @xmath215 of the linear operator @xmath216 . in principle
this can be done analytically , but we use standard numerical packages to find the eigenvector . the stationary distribution for the state of the population is found by summing over all states of the environment , @xmath217 .
this solution is exact .
if the environment states are long - lived , the population will relax to the stationary state of the current environment before the next switching event . with this
, one might expect that the overall stationary distribution is the weighted average of the stationary distributions one would obtain in the respective stationary environments .
the stationary distribution in a single fixed environment , @xmath5 , can be found explicitly as @xmath218 this can be derived for example from equation ( [ eq : statdist ] ) assuming that the transition matrix of the environment is diagonal , @xmath219 , where @xmath220 is the kronecker delta . the average stationary distribution over many slow - switching environments
can then be written as @xmath221 where @xmath222 is the probability that the environment is in state @xmath5 . alternatively ,
if the switching probabilities per time - step are large one might expect the stationary distribution to be approximated by the distribution found in a system controlled by the effective transition rates , @xmath223 .
these are obtained as described in section [ ssec : telegraph ] , with suitable modifications to account for mutation .
the resulting stationary distribution is found as @xmath224 the distributions @xmath225 and @xmath226 are both approximations . to evaluate the validity of the assumptions leading to these approximations
we compute the distance @xmath227 and similarly for @xmath226 , where @xmath228 is distribution of the population at time @xmath97 obtained from simulations . to confirm our analytical approach
we also compute the distance of simulation data from the exact solution for @xmath229 .
we allow the system to run for a fixed time @xmath230 , and we then use the time - averaged distance , @xmath231 to evaluate the accuracy of the approximations .
we ignore the first half of the time series to remove remnants of the initial condition .
we note that the time @xmath230 , measured in generations , is equivalent to @xmath232 simulation time - steps and is chosen to be long enough such that the system relaxes to the stationary state before measurements start at time @xmath233 .
( dotted line ) and coordination game @xmath234 ( dashed lines ) calculated from equation ( [ eq : statdistsingle ] ) , along with the exact solution @xmath229 [ equation ( [ eq : statdist ] ) , evaluated numerically ] and the ` average ' @xmath225 [ equation ( [ eq : statdistave ] ) ] and ` effective ' @xmath226 [ equation ( [ eq : statdisteff ] ) ] approximate stationary solutions ( solid lines ) .
these distributions are for switching probabilities @xmath235 .
( b ) the time - averaged distance [ equation ( [ eq : distance ] ) ] between distributions obtained from an ensemble of @xmath236 simulations , @xmath237 , and the analytic stationary distributions for symmetric switching , @xmath238 .
simulations run for @xmath239 generations .
inset plots show the time - averaged distances over all switching - parameter space .
left inset panel shows the distance to the ` averaged ' stationary solution [ equation ( [ eq : statdistave ] ) ] , and right inset panel shows the distance to the ` effective ' stationary solution [ equation ( [ eq : statdisteff ] ) ] .
the payoff matrix elements are @xmath153 , @xmath175 and @xmath176 , the system size is @xmath187 , the selection intensity is @xmath188 , and the mutation probability is @xmath240 .
( online version in colour . ) ] we present results for the two - world scenario , where the environment switches between a coexistence game and a coordination game as described above .
the stationary distribution of the environmental state is given by @xmath241 and @xmath242 .
the stationary distributions of the population for the fixed environments ( calculated using equation ( [ eq : statdistsingle ] ) ) are shown in figure [ fig : fig5](a ) . in a constant coexistence game ( @xmath171 ) the stationary distribution
is peaked about the point at which the gradient of selection changes sign , and in a fixed coordination - game environment ( @xmath172 ) we find a distribution which is strongly peaked near the @xmath4 state .
the asymmetry is due to the imbalanced payoff matrix used , such that the basin of attraction for the @xmath243 state is much larger than for the @xmath244 state .
for the parameters chosen in the figure [ fig : fig5 ] , the selection - balance point is at @xmath245 . for equal switching rates , @xmath238 ,
the averaged stationary distribution @xmath225 lies exactly in between the two single - environment distributions .
the effective distribution is approximately uniform in the centre of the domain , with a lower probability of being found close to the domain boundaries .
this reflects the fact that for equal switching probabilities the effective game is close to neutral , but frequent mutations push the population to the interior . the exact solution [ equation ( [ eq : statdist ] ) ] matches the features of the single - environment distributions , with a peak at @xmath243 and at the coexistence point @xmath201 .
interestingly , this solution also predicts a peak at the @xmath244 state , a feature which is not seen in the single - environment distributions , or in the effective distribution . as seen in the main panel of figure [ fig : fig5](b ) ,
the exact solution is confirmed by simulations across many orders of magnitude of switching probabilities .
any deviations can be attributed to incomplete equilibration of the measure used in equation ( [ eq : distance ] ) . for large switching probabilities , the effective stationary distribution , @xmath226 , matches the simulations .
as expected the effective theory becomes inaccurate for slow switching , roughly below @xmath246 , in our example .
the weighted - average stationary distribution , @xmath225 , shows the opposite behaviour .
it is in reasonable agreement with simulations for slow switching , but shows systematic deviations when the switching process is too fast for the population to react adiabatically .
this picture is further corroborated by the data shown in the inset panels of figure [ fig : fig5](b ) .
the weighted - average and the effective distributions accurately predict the stationary distribution obtained from simulations when the two switching rates are very disparate , i.e. @xmath205 or vice versa ( top - left and bottom - right corners of the two insets ) . in these regions , the environment spends most of its time in one state , so that the model effectively reduces to the single - environment case .
simulation data , the exact solution , and both approximations then all collapse to the same result , the stationary distribution obtained in a single fixed environment .
the approach based on effective transition rates ( right inset of figure [ fig : fig5](b ) ) is found to be accurate over a large range of switching probabilities away from the slow - switching scenario .
conversely , the weighted - average distribution ( left inset of figure [ fig : fig5](b ) ) becomes increasingly accurate if the dynamics of the environment is slow ( @xmath247 ) .
the dynamics of a population evolving under changing environmental conditions is an important concept in the study of bacterial populations .
previous work has focused on deterministic analyses @xcite , or on an environment following a continuous stochastic process @xcite .
here we have taken a different route and assumed that the environment switches between discrete states .
we have developed the mathematical formalism to describe fixation properties in a general birth - death process in an environment fluctuating between an arbitrary number of states .
the main results of this investigation are self - consistent expressions for the fixation probability of a mutant in a fixed - size population , as well as for the mean unconditional and conditional fixation times . for short - lived environments we put forward an approximation based on effective transition probabilities . as a specific application
we discuss the fixation properties in the context of an evolutionary game in a two - world scenario .
the two states of the environment then correspond to two different payoff matrices of the underlying games .
simulations confirm our exact solution over a wide range of switching probabilities .
the approximation based on effective transition probabilities is seen to reproduce simulation data in the limit of fast switching .
focussing on the case of switching between a coexistence game and a coordination game we find unexpected non - trivial behaviour of the fixation probability of a single mutant .
we observe in our analytical results and in simulations that fixation can be more likely in a scenario in which the fitness landscape switches between the two games than in either of the two constant environments .
we provide an intuitive explanation for this effect , and we have investigated in detail the circumstances under which this phenomenon can occur . adding mutations to the dynamics removes the possibility of fixation , but introduces non - trivial stationary states .
we develop a method for the calculation of this distribution , along with approximations for long - lived and short - lived environmental states respectively .
these approximations are shown to agree well with simulations in their respective limits .
the general theory developed here now allows further investigation of evolutionary dynamics in time - varying environments .
it provides a first mathematical characterisation of the effects one may expect in such systems .
the closed form self - consistent solutions will help to speed up future studies , and they may remove the need for extensive computer simulations . while our work is mainly mathematical , we think that our theory can be used to interpret existing experimental studies such as those studied by acar _
et al . _
@xcite . for some biological systems it may be more appropriate to use constant selection in each environment , as opposed to frequency - dependent selection .
our example of switching between coexistence and coordination games was chosen to illustrate the theory .
we have also seen such cases lead to unexpected effects .
we note that both types of game have been observed in systems of experimental evolution @xcite .
we hope the formalism we have developed will be useful to analyse models closer to other biological applications , and potentially to guide future experiments on evolutionary systems in time - dependent environments . on a more general level constructing a mathematical theory of evolutionary dynamics
is very much work in progress .
an integral part of the evolution of microbes and higher organisms alike is frequency - dependent selection . at the same time
external factors determining the detailed mechanics of selection may vary in time . in this work we have combined frequency - dependent selection , fluctuating environments , and stochastic dynamics in finite populations into one model , and we have provided the analytical tools for its analysis .
this , we hope , is a contribution toward a more complete understanding of evolutionary processes .
p.a . acknowledges support form the epsrc .
p.m.a . gratefully acknowledges financial support from deutsche akademie der naturforscher leopoldina , grant no .
lpds 2012 - 12 , as well as discussions with whan ghang and christian hilbe .
franzenburg s , fraune s , altrock pm , kuenzel s , baines jf , traulsen a , & bosch tcg .
bacterial colonization of hydra hatchlings follows a robust temporal pattern .
_ isme journal_. 2013;7:781790 .
( doi 10.1038/ismej.2012.156 ) mcfall - ngai m , hadfileld mg , bosch tcg , carey hv , domazet - loso t , douglas ae , dubilier n , eberl g , fukami t , gilbert sf , _
et al . _
animals in a bacterial world , a new imperative for the life sciences .
_ proceedings of the national academy of sciences usa_. 2013;110:32293236 .
( doi 10.1073/pnas.1218525110 ) maclean rc , fuentes - hernandez a , greig d , hurst ld , & gudelj i. a mixture of `` cheats '' and `` co - operators '' can enable maximal group benefit .
_ plos biology_. 2010;8(9):e1000486 .
( doi 10.1371/journal.pbio.1000486 ) arnoldt h , timme m , & grosskinsky s. frequency - dependent fitness induces multistability in coevolutionary dynamics . _
journal of the royal society interface_. 2012;9:33873396 .
( doi 10.1098/rsif.2012.0464 ) | the environment in which a population evolves can have a crucial impact on selection .
we study evolutionary dynamics in finite populations of fixed size in a changing environment .
the population dynamics are driven by birth and death events .
the rates of these events may vary in time depending on the state of the environment , which follows an independent markov process .
we develop a general theory for the fixation probability of a mutant in a population of wild - types , and for mean unconditional and conditional fixation times .
we apply our theory to evolutionary games for which the payoff structure varies in time .
the mutant can exploit the environmental noise ; a dynamic environment that switches between two states can lead to a probability of fixation that is higher than in any of the individual environmental states .
we provide an intuitive interpretation of this surprising effect .
we also investigate stationary distributions when mutations are present in the dynamics . in this regime , we find two approximations of the stationary measure .
one works well for rapid switching , the other for slowly fluctuating environments .
[ firstpage ] * keywords : * evolutionary dynamics , fluctuating environments , stochastic processes |
, extremely low error probability has been required in many application areas such as wireless communication systems without feedback and data storage systems . low - density parity - check ( ldpc ) codes@xcite,@xcite have become one of promising error - correcting codes in these areas due to their near capacity - approaching performance . however , since finite - length ldpc codes show a serious error - floor problem@xcite in high signal to noise ratio ( snr ) region , it may be difficult to achieve an extremely good error correction performance .
many researches on the iterative decoding of ldpc codes ( e.g. @xcite -@xcite ) have been made to achieve good error correction performance by improving only the decoding algorithm while not changing the code structure .
however , the requirement for the performance and complexity of decoder is getting more strict .
thus , more powerful decoding schemes for ldpc codes are required . in this paper
, we propose a decoding scheme for ldpc codes using simple product code structure to improve the error correction performance together with error floor .
the proposed decoding scheme is based on combining two independently received soft - decision data for an unsuccessfully decoded codeword , which achieves good decoding performance with relatively low additional decoding complexity . in the encoding procedure of the proposed scheme ,
codewords of the ldpc codes are stacked in rows as horizontal codewords of the product codes .
then vertical codes which are used to help decoding of the ldpc codewords are applied to the stack of ldpc codewords , which results in a codeword matrix like product code structure . in the decoding procedure ,
decoding of each horizontal ldpc codeword in a codeword matrix is performed firstly according to decoding algorithm of ldpc codes .
if some codewords fail to be successfully decoded , re - decoding based on combining two received soft - decision data for the same codeword will be performed .
also , we define and analyze the decoding capability of the proposed decoding scheme using the parity - check matrices of vertical codes and derive the _ combined - decodability _ for the case of single parity - check ( spc ) and hamming codes being used as vertical codes .
especially , the proposed decoding scheme shows better error - correcting capability in high snr region .
any linear codes can be used as vertical codes in the proposed scheme . however , since vertical codes are only used to help decoding of ldpc codes in the proposed scheme , short codes with high rate are preferable as vertical codes .
thus , simple algebraic codes such as spc or hamming codes are appropriate for the vertical codes .
even though the vertical codes are assumed to be systematic in this paper , they do not have to be systematic for the proposed decoding scheme .
the proposed decoding scheme can be applied to any other linear codes with soft - decision decoding than ldpc codes . however , undetected errors in the horizontal codewords cause a serious problem for the proposed scheme . if undetected errors seldom occur and soft - decision decoding is used for the horizontal codes such as ldpc and turbo codes@xcite , the proposed decoding scheme can be effectively applied .
this paper is organized as follows : in section [ sec : proposedscheme ] , after introducing and defining some notations and operations , a new decoding scheme for ldpc codes using simple product code structure is proposed .
two simple examples for the proposed decoding scheme are given in section [ sec : simpleexamples ] . in section [ sec : correctioncapability ] , the combined - decodability of the proposed decoding scheme is analyzed .
the numerical results are shown in section [ sec : simulationresults ] and the conclusion is given in section [ sec : conclusions ] .
in this section , we propose a new decoding scheme for ldpc codes using the concept of product codes . in the proposed scheme ,
ldpc codes act as horizontal codes of the product codes and simple codes such as spc or hamming codes are used for vertical codes of the product code to help the decoding of the horizontal codes .
product codes are serially concatenated codes @xcite , which were introduced in @xcite . by using the concept of product codes ,
a long block code can be easily constructed by using two or more short block codes .
now , we consider two systematic linear block codes , @xmath0 with parameters @xmath1 and @xmath2 with parameters @xmath3 , where @xmath4 , and @xmath5 denote the code length , the number of information bits , and the minimum hamming distance , respectively .
the product code @xmath6 is obtained by the following steps as illustrated in fig .
[ fig : structureprod ] : 1 . placing @xmath7 information bits in an array of @xmath8 rows and @xmath9 columns ; 2 . encoding each of @xmath8 rows using code @xmath2 ; 3 . encoding each of @xmath10 columns using code @xmath0 . the codes @xmath0 and @xmath2 are also called as vertical code and horizontal code of the product code @xmath11 , respectively
. the parameters of the product code @xmath11 are @xmath12 and the code rate is given as @xmath13 , where @xmath14 and @xmath15 are the code rates of @xmath0 and @xmath2 , respectively .
in this paper , for simplicity , only binary block codes are considered but the proposed decoding scheme can be directly applied to the nonbinary codes .
let @xmath16 and @xmath17 , @xmath18 , be the number of unsuccessfully decoded horizontal codewords in a codeword matrix after first decoding of each @xmath19 horizontal codewords .
parity - check matrix and its modifications of a vertical code in the product code are defined as : * @xmath20 : @xmath21 parity - check matrix of a vertical code in the product code ; * @xmath22 : @xmath23 extended parity - check matrix whose rows are all possible nonzero linear combinations of rows in @xmath20 ; * @xmath24 : @xmath25 punctured parity - check matrix constructed by selecting @xmath17 columns from @xmath22 .
it can be assumed that @xmath20 contains no all - zero column .
it is clear that the length @xmath19 of the vertical code also denotes the number of horizontal codewords in a codeword matrix of the product code .
the column indices of @xmath24 are same as the row indices of the unsuccessfully decoded horizontal codewords in the codeword matrix .
let @xmath26 , @xmath27 , denote @xmath19 binary horizontal codewords and @xmath28 denote the _ soft - decision vector _ for @xmath29 , where @xmath30 is the log likelihood ratio ( llr ) for @xmath31 calculated from the corresponding received data @xmath32 defined by @xmath33 then , the check equations ( or rows ) in @xmath34 $ ] describe all possible relations among @xmath29 s , that is , @xmath35 .
note that these check equations ( or rows ) of @xmath22 are not linearly independent .
since the vertical code is a systematic code , we can use @xmath36 , @xmath37 , and @xmath38 , to denote the _
systematic ( horizontal ) codewords _ including the systematic part of the vertical code and the _ parity ( horizontal ) codewords _ including the parity part of the vertical code as given in fig .
[ fig : structureprod ] , respectively .
special algebra for llr values in @xcite can be used with small modification .
we use the operation @xmath39 for combining two llr values defined as @xmath40 let @xmath41 , @xmath27 , denote the vector of llr values computed from the horizontal codeword @xmath29 itself , i.e. , @xmath42 if @xmath43 and @xmath44 if @xmath45
. then it is easy to show that @xmath46 can be calculated using the sign change operation as @xmath47 in the proposed decoding scheme , the received codewords of the ldpc codes are stacked as horizontal codewords for the product code .
the number of codewords in a stack is determined by the vertical codes which are used to help decoding of the horizontal codes .
next , we attempt to decode each horizontal codeword in the codeword matrix .
if all the systematic codewords @xmath48 , @xmath49 , are successfully decoded , the decoding process will be successfully finished . however , if some horizontal codewords fail to be decoded , then re - decoding of the horizontal codewords using @xmath24 of the vertical codes will be performed .
the details of this procedure are explained below .
suppose that there are @xmath17 unsuccessfully decoded horizontal codewords in a codeword matrix and the @xmath50-th row of @xmath24 of the vertical codes has the minimum nonzero hamming weight @xmath51 among all rows of @xmath24 .
clearly , it implies that @xmath51 unsuccessfully decoded horizontal codewords corresponding to the nonzero elements are involved in the @xmath50-th check equation of @xmath22 of the vertical code , which is the check equation affected by the minimum number of unsuccessfully decoded horizontal codewords .
therefore , @xmath24 is used to decide which check equation of @xmath22 is most easily solvable , i.e. , which check equation contains the fewest unsuccessfully decoded horizontal codewords . in the proposed decoding scheme ,
the check equation which contains the fewest unsuccessfully decoded horizontal codewords is first utilized for re - decoding . according to the value of @xmath51 ,
the proposed decoding scheme is performed by considering the following three cases .
case 1 ) @xmath52 : if there is only one unsuccessfully decoded horizontal codeword participating in a check equation of @xmath22 , say the @xmath50-th row of @xmath22 , of the vertical code , we can correct it simply by the modulo 2 addition ( denoted by @xmath53 ) of the other successfully decoded horizontal codewords participating in the same check equation .
let @xmath29 be the unsuccessfully decoded horizontal codeword participating in the @xmath50-th check equation of @xmath22 of the vertical code .
then we can recover @xmath29 by @xmath54 where @xmath55 is the set of indices of the nonzero elements in the @xmath50-th row of @xmath22 .
as @xmath29 is successfully decoded , the corresponding column of @xmath24 is removed and the number of columns in @xmath24 reduces by one .
this process is repeatedly applied to the check equations which satisfy the @xmath56 condition for updated @xmath24 .
case 2 ) @xmath57 : if there are only two unsuccessfully decoded horizontal codewords participating in a check equation ( i.e. , a row ) of @xmath22 of the vertical code , we can re - decode the combined vector of two independently received soft - decision vectors for the same horizontal codeword similar to the type - i hybrid - automatic repeat request ( h - arq ) scheme @xcite .
we already know that among horizontal codewords participating in a check equation of @xmath22 , any codeword is equivalent to the sum of all the other codewords .
thus we can derive two independently received soft - decision vectors for any of two unsuccessfully decoded horizontal codewords ; ( a ) one from an unsuccessfully decoded horizontal codeword , ( b ) the other one from the soft - decision vector of the other unsuccessfully decoded horizontal codeword by changing signs of its elements according to the successfully decoded horizontal codewords participating in the same check equation .
let @xmath58 and @xmath59 be the unsuccessfully decoded horizontal codewords participating in the @xmath50-th check equation of @xmath22 of the vertical code .
then , from , we can have @xmath60 in addition to the soft - decision vector @xmath61 for @xmath58 , we can derive the other soft - decision vector for @xmath58 by using the relation as @xmath62 since it can be regarded that the horizontal codeword @xmath58 is independently received twice through the channel , we can just add two soft - decision vectors as @xmath63\ ] ] and re - decode it by achieving approximately 3db performance gain similar to type - i h - arq scheme . if the re - decoding is successful , then the @xmath50-th check equation of @xmath22 contains only one unsuccessfully decoded horizontal codeword @xmath59 and it can be recovered by the method in case 1 ) .
case 3 ) @xmath64 : if there are three or more unsuccessfully decoded horizontal codewords participating in a check equation of @xmath22 of the vertical code , combining the soft - decision vectors for further re - decoding is still possible .
first of all , the unsuccessfully decoded horizontal codewords participating in the same check equation of @xmath22 are partitioned into two groups and a soft - decision vector for each group is obtained by summing all the unsuccessfully decoded soft - decision vectors in that group using the operation .
then , for these two soft - decision vectors , the same method for two unsuccessfully decoded horizontal codewords in case 2 ) is applied .
we can try this process to all possible partitioning of unsuccessfully decoded horizontal codewords until the decoding succeeds and then reduces the number of unsuccessfully decoded codewords in the same check equation .
however , it should be noted that more combined horizontal codewords result in less performance gain .
let @xmath65 be the @xmath66 unsuccessfully decoded horizontal codewords participating in the @xmath50-th check equation of @xmath22 of the vertical code . then can be rewritten as @xmath67 two soft - decision vectors for @xmath68 can be obtained as @xmath69 and @xmath70 since it can be regarded that one horizontal codeword is independently received twice through the channel , we can add two soft - decision vectors in and and re - decode it . if the re - decoding succeeds , then @xmath71 and @xmath72 are known . by repeating the previous process , @xmath73
can also be split into two groups as @xmath74 and the same process can be applied to decode @xmath75 .
if the re - decoding is unsuccessful , then different grouping of the unsuccessfully decoded codewords can be tried .
this process is repeatedly performed until all errors are corrected .
the flowchart of the proposed decoding scheme is shown in fig .
[ fig : flowchart_main ] , where only @xmath76 is considered .
this flowchart is based on the assumption that the vertical code is a linear systematic code .
note that four steps in the left bottom part of fig .
[ fig : flowchart_main ] are equivalent to the erasure decoding under binary erasure channel .
it is well known that the _ error correctability _
@xmath77 of the vertical code is determined by the minimum hamming distance @xmath78 of the code , i.e. , @xmath79 under binary erasure channel .
assume that an spc code with length @xmath19 is used as a vertical code of the product code and an ldpc code is used as a horizontal code .
then , the parity - check matrix @xmath20 of the vertical code and its @xmath22 can be written as @xmath80 from the parity - check matrix , it holds @xmath81 after decoding each of @xmath19 horizontal codewords , there may exist @xmath17 unsuccessfully decoded horizontal codewords .
since @xmath20 , @xmath22 , and @xmath24 have only one row whose entries are all 1 , @xmath51 is always equal to @xmath17 .
if @xmath82 and the unsuccessfully decoded codeword is parity codeword @xmath83 , we already successfully decode the whole systematic horizontal codewords @xmath84 . if @xmath82 and the unsuccessfully decoded codeword is a systematic codeword @xmath48 , this codeword can be easily corrected by the even parity condition for the vertical code as @xmath85 if @xmath86 and two systematic codewords @xmath87 and @xmath88 are unsuccessfully decoded , can be rewritten as @xmath89 we can exactly calculate @xmath90 because all these codewords are successfully decoded . with this result and the soft - decision vector for @xmath88 , we have another independently received soft - decision vector @xmath91 for a codeword @xmath87 as @xmath92 similar to the type - i h - arq , we can add two soft - decision vectors @xmath61 and @xmath91 and then re - decode it to obtain @xmath87 .
if one of two unsuccessfully decoded codewords is @xmath83 , it is easy to show that the same procedure can be applied .
if @xmath93 and three systematic codewords @xmath87 , @xmath88 , and @xmath94 are unsuccessfully decoded , can be rewritten as @xmath95 then , we can combine the soft - decision vectors for @xmath88 and @xmath94 by and exactly calculate @xmath96 . by using these vectors , we derive another independently received soft - decision vector @xmath91 for the codeword @xmath87 as @xmath97 finally , we add two soft - decision vectors @xmath61 and @xmath91 and re - decode it for @xmath87 .
if the re - decoding fails , then we can retry the same decoding procedure for @xmath88 or @xmath94 instead of @xmath87 . in the case of @xmath98
, the similar process can be applied .
assume that a @xmath99 hamming code is used as a vertical code of the product code and an ldpc code is used as a horizontal code .
then , the parity - check matrix @xmath20 of the vertical code and its @xmath22 can be written as @xmath100 1 & 1 & 1 & 0 & 0 & 1 & 0 \\[-3 mm ] 0 & 1 & 1 & 1 & 0 & 0 & 1 \\ \end{bmatrix } \rightarrow h_e = \begin{bmatrix } 1 & 0 & 1 & 1 & 1 & 0 & 0 \\[-3 mm ] 1 & 1 & 1 & 0 & 0 & 1 & 0 \\[-3 mm ] 0 & 1 & 1 & 1 & 0 & 0 & 1 \\[-3 mm ] 0 & 1 & 0 & 1 & 1 & 1 & 0 \\[-3 mm ] 1 & 1 & 0 & 0 & 1 & 0 & 1 \\[-3 mm ] 1 & 0 & 0 & 1 & 0 & 1 & 1 \\[-3 mm ] 0 & 0 & 1 & 0 & 1 & 1 & 1 \\ \end{bmatrix}.\ ] ] then , the rows of @xmath22 give the following check equations for 7 horizontal codewords @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath83 , @xmath105 , and @xmath106 ; @xmath107 after decoding each of seven horizontal codewords , there may exist @xmath17 horizontal codewords , @xmath108 , which fail to be correctly decoded . if @xmath109 , at least one check equation from @xmath110 contains that unsuccessfully decoded horizontal codeword , which can be easily corrected .
it is easy to see that selecting proper check equation to correct an erroneous horizontal codeword is equivalent to finding a weight-1 row in @xmath24 .
the same approach can be used to correct more than one error .
for example , let @xmath101 , @xmath102 , and @xmath103 be three unsuccessfully decoded horizontal codewords .
then , we have @xmath111 1 & 1 & 1 \\[-3 mm ] 0 & 1 & 1 \\[-3 mm ] 0 & 1 & 0 \\[-3 mm ] 1 & 1 & 0 \\[-3 mm ] 1 & 0 & 0 \\[-3 mm ] 0 & 0 & 1 \\ \end{bmatrix}\ ] ] and each recovery process for @xmath101 , @xmath102 , and @xmath103 can be conducted using , , and , which correspond to the sixth , the fourth , and the last rows of @xmath24 , respectively .
however , for three unsuccessfully decoded codewords @xmath101 , @xmath102 , and @xmath104 , there is no weight-1 row in the following @xmath24 constructed by selecting the first , second , and fourth columns of @xmath22 , which has @xmath57 as @xmath111 1 & 1 & 0 \\[-3 mm ] 0 & 1 & 1 \\[-3 mm ] 0 & 1 & 1 \\[-3 mm ] 1 & 1 & 0 \\[-3 mm ] 1 & 0 & 1 \\[-3 mm ] 0 & 0 & 0 \\ \end{bmatrix}.\ ] ] instead , as the case 2 ) in the subsection [ sec : proposedscheme]-b , by using corresponding to the first row of the above @xmath24 , we can modify the sign of each element of the soft - decision vector of @xmath104 using the correctly decoded codewords @xmath103 and @xmath83 , which results in another soft - decision vector for @xmath101 .
this newly created soft - decision vector can be added to the original soft - decision vector for @xmath101 and the re - decoding can be performed for this combined vector .
if the re - decoding fails , we can try the same decoding procedure for another check equation corresponding to a weight-2 row in @xmath24 .
now , we consider more severe case , for example , there is only one successfully decoded horizontal codeword in a codeword matrix ( @xmath112 ) , say @xmath103 .
using , we have the following re - decoding steps : 1 .
combine the soft - decision vectors of @xmath104 and @xmath83 using the operation to have the soft - decision vector for @xmath113 ; 2 .
modify the sign of each element of the resulting soft - decision vector using @xmath103 to obtain the soft - decision vector for @xmath114 .
the resulting soft - decision vector for @xmath114 can be considered as the other independently received vector for @xmath101 .
thus this newly created soft - decision vector can be added to the original soft - decision vector for @xmath101 and the re - decoding can be performed for this combined soft - decision vector .
if the decoding is successful , then it gives @xmath115 case .
if the decoding is not successful , we can also try other check equation until the decoding is successful or there is no remaining check equation available .
in this section , the combined - decodability of the proposed decoding scheme is defined and the relation between the combined - decodability of the proposed decoding scheme and the vertical code with parity - check matrix @xmath20 is investigated .
although the soft information combining of the proposed decoding scheme can be applied for any value of @xmath51 , the analysis of combined - decodability is performed by only considering the cases of @xmath116 in this section . for a given @xmath20 of a vertical code @xmath0 and a fixed @xmath117 ,
if every possible @xmath118 matrix @xmath24 contains at least one row of hamming weight one or two , the vertical code @xmath0 is said to be @xmath117 _
combinable_. if a vertical code @xmath0 is @xmath117 combinable for all @xmath119 , then @xmath0 is said to be @xmath120 _ combined - decodable_. if @xmath0 is not @xmath121 combined - decodable but @xmath120 combined - decodable , then @xmath120 is called the combined - decodability of @xmath0 . using the above definitions , the length of vertical code can be bounded as in the following lemmas and theorems .
+ if a vertical code @xmath0 with a parity - check matrix @xmath20 is @xmath117 combinable , any vertical code @xmath122 whose parity - check matrix @xmath123 is constructed by selecting @xmath117 or more columns from @xmath20 is also @xmath117 combinable .
a set of all possible @xmath124 s which are constructed by selecting @xmath117 columns from @xmath125 is contained in the set of all possible @xmath24 s which are constructed by selecting @xmath117 columns from @xmath22 .
all vertical codes @xmath0 are @xmath126 and @xmath127 combinable . since there is no all - zero column in the parity - check matrix @xmath20 of any vertical code @xmath0 , all vertical codes @xmath0 are both @xmath126 and @xmath127 combinable .
if a vertical code @xmath0 with an @xmath21 parity - check matrix @xmath20 is @xmath128 combinable , then @xmath129 , where @xmath130 .
moreover , when @xmath131 , @xmath0 is @xmath128 combinable if and only if @xmath20 is constructed by using every nonzero @xmath132-tuple column vector exactly twice as its columns .
if @xmath133 , there should exist three identical columns in @xmath22 .
if these three identical columns are selected to construct @xmath24 , each row weight of @xmath24 is either zero or three .
thus @xmath0 can not be 3 combinable .
+ ( only if part ) : for a parity - check matrix @xmath20 of a 3 combinable code @xmath134with @xmath135 , suppose that all the nonzero @xmath132-tuple column vectors do not appear exactly twice in @xmath20 . then @xmath22 should have at least three identical columns .
if these three identical columns are selected to construct @xmath24 , it contradicts the assumption that @xmath0 is 3 combinable . + ( if part ) : clearly , @xmath131 because the number of nonzero @xmath132-tuple column vectors is @xmath136 .
if we construct @xmath24 by choosing any three columns from @xmath22 , there should be at least one distinct column from the others and therefore we can always find a row of weight one or two in @xmath24 . from lemmas 4 and 5 , the following theorem can be stated without proof .
+ the combined - decodability of spc codes is two . in order to find the combined - decodability of hamming codes ,
the following lemmas are needed .
+ if a vertical code @xmath0 with an @xmath21 parity - check matrix @xmath20 is 4 combinable , then @xmath137 , where @xmath130 .
moreover , when @xmath138 , @xmath0 is @xmath139 combinable if and only if @xmath20 is constructed by using every nonzero @xmath132-tuple column vector exactly three times as its columns .
if @xmath140 , there should exist four identical columns in @xmath22 .
if these four identical columns are selected to construct @xmath24 , each row weight of @xmath24 is either zero or four .
thus @xmath0 can not be 4 combinable .
+ ( only if part ) : for a parity - check matrix @xmath20 of a 4 combinable code @xmath0 with @xmath141 , suppose that all the nonzero @xmath132-tuple column vectors do not appear exactly three times in @xmath20 .
then @xmath22 should have at least four identical columns .
if those four identical columns are selected to construct @xmath24 , there is no row of weight one or two in @xmath24 .
thus it contradicts the assumption that @xmath0 is 4 combinable . + ( if part ) : since each nonzero @xmath132-tuple column vector appears exactly three times in @xmath20 , @xmath138 .
if we choose any four columns from @xmath22 , there should be at least one distinct column from the others .
suppose that there is neither a weight-1 row nor a weight-2 row in @xmath24 for @xmath142 .
since both @xmath22 and @xmath24 are closed under the row - wise addition , @xmath24 can be of two distinct types , that is , the first type has @xmath143 as its nonzero rows and the second type has one 4-tuple of weight three , say @xmath144 , as its nonzero rows such that @xmath145 \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 0 & 0 & 0 & 0 \\[-3 mm ] 1 & 1 & 1 & 1 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 1 & 1 & 1 & 1 \\ \end{bmatrix } ~\textrm{or}~ \begin{bmatrix } 0 & 0 & 0 & 0 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 0 & 0 & 0 & 0 \\[-3 mm ] 1 & 1 & 1 & 0 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 1 & 1 & 1 & 0 \\ \end{bmatrix}.\ ] ] however , note that the latter matrix can not be obtained because there is no all - zero column in @xmath24 .
therefore the former matrix is the only possible matrix for @xmath24 to have rows of weight zero or four .
based on this result , we can see that there should be a row of weight one or two in @xmath24 .
the following corollary is given without proof .
+ if a vertical code @xmath0 is 3 combinable , then @xmath0 is also 4 combinable and thus @xmath0 is 4 combined - decodable .
suppose that vertical code @xmath0 has minimum distance @xmath146 which is larger than or equal to 3 . if @xmath0 with an @xmath21 parity - check matrix @xmath20 is 5 combinable , then @xmath147 , where @xmath130 .
moreover , when @xmath148 , @xmath0 is 5 combinable if and only if @xmath20 is constructed by using every nonzero @xmath132-tuple column vector only once as its columns .
since @xmath146 is larger than or equal to 3 , @xmath20 should not contain identical columns .
therefore , @xmath149 is trivial and the first part is proved .
+ ( only if part ) : for @xmath148 and @xmath150 , @xmath20 is uniquely constructed by using every nonzero @xmath132-tuple column vector only once as its columns . + ( if part ) : suppose that there is neither a weight-1 row nor a weight-2 row in @xmath24 for @xmath151 .
since both @xmath22 and @xmath24 are closed under the row - wise addition , there are only two types of @xmath24 such as 1 .
@xmath24 should contain @xmath152 as its only nonzero rows ; 2 .
@xmath24 should contain @xmath153 , @xmath154 , and @xmath155 at least once as its nonzero rows .
note that the second type represents all the cases of three nonzero rows having 1 , 2 , and 2 zeros in distinct positions , respectively .
then we have @xmath156 \vdots & \vdots & \vdots & \vdots & \vdots \\[6 mm ] 0 & 0 & 0 & 0 & 0 \\[-3 mm ] 1 & 1 & 1 & 1 & 1 \\[6 mm ] \vdots & \vdots & \vdots & \vdots & \vdots \\[6 mm ] 1 & 1 & 1 & 1 & 1 \\ \end{bmatrix } ~~\textrm{or}~~ \begin{bmatrix } 0 & 0 & 0 & 0 & 0 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 0 & 0 & 0 & 0 & 0 \\[-3 mm ] 1 & 1 & 1 & 1 & 0 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 1 & 1 & 1 & 1 & 0 \\[-3 mm ] 1 & 1 & 0 & 0 & 1 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 1 & 1 & 0 & 0 & 1 \\[-3 mm ] 0 & 0 & 1 & 1 & 1 \\[-3 mm ] \vdots & \vdots & \vdots & \vdots & \vdots \\[-3 mm ] 0 & 0 & 1 & 1 & 1 \\ \end{bmatrix}.\ ] ] therefore , in order for @xmath0 to be 5 combinable , @xmath20 should contain neither five identical columns nor two distinct pairs of columns .
if @xmath20 is constructed by using every nonzero @xmath132-tuple column vector only once as its columns , it is clear that @xmath20 has neither five identical columns nor two distinct pairs of columns and @xmath0 is 5 combinable . thus the proof is done .
+ for 5 combinable vertical code @xmath0 without restriction on the minimum distance , it is not difficult to verify that the upper bound in lemma 9 becomes @xmath157 .
this bound can be proved by similar method used in the proof of lemma 9 . in this subsection
, we will derive the combined - decodability of hamming codes .
hamming codes as vertical codes are neither 6 combinable nor 7 combinable . for a @xmath99 hamming code
, @xmath22 is given in subsection [ sec : examplehamming ] .
the first three rows in @xmath22 show all possible nonzero column vectors of length 3 and the rows in @xmath22 are all possible linear combinations of the first three rows except all - zero row .
since all rows in @xmath24 have weight four , a @xmath99 hamming code is not 7 combinable . for a @xmath158 hamming code , the parity - check matrix consists of all nonzero @xmath132-tuple column vectors . using only column permutation
, we can reorder the parity - check matrix as @xmath159 \vdots & \vdots & \vdots & \vdots & \vdots & \vdots & \vdots & \cdots \\[-3 mm ] 0 & 0 & 0 & 0 & 0 & 0 & 0 & \cdots \\[-3 mm ] 1 & 0 & 1 & 1 & 1 & 0 & 0 & \cdots \\[-3 mm ] 0 & 1 & 0 & 1 & 1 & 1 & 0 & \cdots \\[-3 mm ] 0 & 0 & 1 & 0 & 1 & 1 & 1 & \cdots \\[-3 mm ] \end{bmatrix}.\ ] ] note that the first seven column vectors have zeros in the first @xmath160 positions and they should be all different from each other . since there is no all - zero column vector in @xmath20 , there should appear all nonzero 3-tuple column vectors in the last three rows corresponding to the first seven columns .
then , we can construct @xmath22 by using all possible non - trivial linear combinations of rows in @xmath20 , and @xmath24 by selecting the first seven columns from @xmath22 .
it is clear that such @xmath24 is made up of all - zero rows and the rows of @xmath22 for a @xmath99 hamming code .
since the weight of rows of @xmath24 is either 0 or 4 , hamming codes are not 7 combinable .
it is also easy to check that hamming codes are not 6 combinable because we can always find an @xmath24 for @xmath161 which consists of weight-0 , weight-3 , and weight-4 rows by choosing any 6 columns from the above @xmath24 for @xmath162 .
the combined - decodability of hamming codes as vertical codes is five . from lemmas 3 , 4 , 5 , 7 and 9 ,
hamming codes are 5 combined - decodable .
the theorem is directly proved by this result and lemma 10 .
.distribution of @xmath24 for all possible horizontal codeword errors in the proposed decoding scheme with hamming codes [ cols="^,^,^,^,^ " , ] table [ tab : distributionhamming ] shows the distribution of @xmath24 for all possible unsuccessfully decoded horizontal codewords when hamming codes are used as vertical codes .
it clearly shows that combined - decodability of any hamming code is five .
table [ tab : distributionnonhamming ] shows the distribution of @xmath24 for all possible unsuccessfully decoded horizontal codewords when double parity - check ( dpc ) codes are used as vertical codes .
it shows that the combined - decodabilities of @xmath163 and @xmath164 dpc codes are four and two , respectively . to analyze the combined - decodability of the proposed decoding scheme , some simple linear codes such as spc , dpc , and hamming codes are considered as vertical codes , the parameters of which are listed in table [ tab : candidates ] .
in this section , numerical analysis of the proposed decoding scheme is performed .
the performance of ldpc codes with @xmath165 and @xmath166 in two different lengths @xmath167 and @xmath168 in ieee 802.16e standard@xcite are compared with that of the proposed decoding scheme . the maximum number of iterations for iterative belief propagation ( bp ) decoding is set to 50 for the entire simulations . for the proposed decoding scheme , the @xmath169 spc code is selected as the vertical code .
to maintain the same overall code rate , the ldpc codes obtained by puncturing the above codes in ieee 802.16e standard are used as the horizontal codes . by using the puncturing rate @xmath170 , the overall code rates for the proposed decoding scheme
are also set to @xmath171 and @xmath172 .
+ fig .
[ fig : performance ] shows that the proposed decoding scheme with @xmath169 spc vertical code outperforms the conventional ldpc codes in ieee802.16e standard for @xmath165 and @xmath166 , especially in high snr region , where ` proposed-@xmath17 ' means that up to @xmath17 horizontal codeword errors are combined - decoded .
error correction performance of each decoding scheme is shown in fig .
[ fig : performance ] by word error rate ( wer ) of ldpc codewords for the conventional scheme and horizontal ldpc codewords for the proposed scheme . even though the proposed decoding scheme does not correct all error patterns of @xmath17 or less horizontal codeword errors , fig .
[ fig : performance ] shows remarkable performance improvement .
the simulation results also show that both the waterfall and error - floor performance of the ldpc code can be improved by constructing a product code and performing the proposed decoding scheme . additional computational complexity for the proposed decoding scheme may be significant for low snr region .
however , for high snr region , decoding failure of more than one horizontal codeword in a codeword matrix is rarely occurred .
thus , in spite of remarkable performance improvement , additional computational complexity per codeword becomes negligible for high snr region .
we showed the numerical results only for the case of spc codes being used as vertical codes .
also , some of possible grouping methods for the erroneous horizontal codewords , i.e. , grouping into one codeword and others , are considered for the case of @xmath173 .
if we use more powerful codes such as hamming codes as vertical codes or consider more various grouping methods for the erroneous horizontal codewords , the performance of the proposed decoding scheme is believed to be improved but their decoding complexity increases .
a new decoding scheme for ldpc codes using simple product code structure is proposed based on combining two independently received soft - decision data for the same codeword .
the proposed decoding scheme can achieve very low error probability , which is very difficult to achieve only by using single channel coding scheme .
especially , the proposed decoding scheme can achieve better error - correcting capability in high snr region .
also , the decoding capability of the proposed decoding scheme is analyzed using parity - check matrices of vertical codes and the _ combined - decodability _ is exactly derived when spc and hamming codes are used as vertical codes .
it should be noted that the proposed decoding scheme can be applied to any other linear codes with soft - decision decoding than ldpc codes and will work well for the codes which rarely have undetected errors such as ldpc codes
. 1 r. g. gallager , low density parity check codes .
cambridge , ma : mit press , 1963 .
d. mackay and r. m. neal , `` near shannon limit performance of low - density parity - check codes , '' _ elec .
1645- 1646 , aug .
t. richardson , `` error floors of ldpc codes , '' in _ proc .
allerton conf .
commun . , control , comput .
1426 - 1435 , oct .
m. p. c. fossorier , m. mihaljevic , and h. imai , `` reduced complexity iterative decoding of low - density parity check codes based on belief propagation , '' _ ieee trans . commun .
673 - 680 , may 1999 .
m. p. c. fossorier ,
`` iterative reliability - based decoding of low - density parity check codes , '' _ ieee j. sel .
908 - 917 , may 2001 .
j. chen and m. p. c. fossorier , `` near optimum universal belief propagation based decoding of low - density parity check codes , '' _ ieee trans .
3 , pp . 406 - 414 , mar .
m. r. yazdani , s. hemati , and a. h. banihashemi , `` improving belief propagation on graphs with cycles , '' _ ieee commun .
_ , vol . 8 , no . 1 , pp .
57 - 59 , jan . 2004 .
s. laendner and o. milenkovic , `` algorithmic and combinatorial analysis of trapping sets in structured ldpc codes , '' in _ proc .
wireless networks ( wirelesscomm2005 ) _ , vol .
630 - 635 , dec . 2005 .
y. han and w. e. ryan , `` low - floor decoders for ldpc codes , '' _ ieee trans .
57 , no . 6 , pp . 1663 - 1673 , jun .
s. k. chilappagari , m. chertkov , m. g. stepanov , and b. vasic , `` instanton - based techniques for analysis and reduction of error floors of ldpc codes , '' _ ieee j. sel .
27 , no . 6 , pp .
855 - 865 , aug . 2009 .
s. k. planjery , d. declercq , s. k. chilappagari , and b. vasic , `` multilevel decoders surpassing belief propagation on the binary symmetric channel , '' in _ proc .
inform . theory _ , pp .
769 - 773 , jun . 2010 .
r. asvadi , a. h. banihashemi , and m. ahmadian - attari , `` lowering the error floor of ldpc codes using cyclic liftings , '' _ ieee trans .
inform . theory _
2213 - 2224 , apr . 2011 .
s. k. planjery , d. declercq , l. danjean , and b. vasic , `` finite alphabet iterative decoders for ldpc codes surpassing floating - point iterative decoders , '' _ elec .
919 - 921 , aug . 2011 . c. berrou , a. glavieux , and p. thitimajshima , `` near shannon limit error - correcting coding and decoding : turbo codes , '' in _ proc .
_ , 1993 , pp
. 1064 - 1070
. f. j. macwilliams and n. j. a. sloane , _ the theory of error - correcting codes .
_ amsterdam , the netherlands : north - holland , 1978 , pp .
567 - 580 .
p. elias , `` error - free coding , '' _ ire trans .
inform . theory _
it-4 , pp .
29 - 37 , sep .
j. hagenauer , e. offer , and l. papke , `` iterative decoding of binary block and convolutional codes , '' _ ieee trans .
inform . theory _
429 - 445 , mar .
s. lin , d. costello , and m. miller , `` automatic - repeat - request error - control schemes , '' _ ieee commun . mag .
12 , pp . 5 - 17 , dec
802.16 , `` ieee standard for local and metropolitan area network part 16 : air interface for fixed and mobile broadband access systems , '' 2004 . | in this paper , a new decoding scheme for low - density parity - check ( ldpc ) codes using the concept of simple product code structure is proposed based on combining two independently received soft - decision data for the same codeword .
ldpc codes act as horizontal codes of the product codes and simple algebraic codes are used as vertical codes to help decoding of the ldpc codes .
the decoding capability of the proposed decoding scheme is defined and analyzed using the parity - check matrices of vertical codes and especially the _ combined - decodability _ is derived for the case of single parity - check ( spc ) and hamming codes being used as vertical codes .
it is also shown that the proposed decoding scheme achieves much better error - correcting capability in high signal to noise ratio ( snr ) region with low additional decoding complexity , compared with a conventional decoding scheme .
b. shin : new decoding scheme for ldpc codes based on simple product code structure combined - decodability , decoding , hamming codes , low - density parity - check ( ldpc ) codes , product codes , single parity - check ( spc ) codes . |
the magnetic properties of small transition metal clusters have been of growing interest in the past few years .
however , there has not been any study about the dependence on temperature so far . as shown in a large variety of papers ( e.g. refs . ) small clusters exhibit , unlike the bulk , smooth structural transformations , which one might call _ isomer hopping _ , and which occur between pure solid and fluid phases over a relatively wide temperature range , in a so - called coexistence phase . in the following
we will be concerned with the effect of those properties on the magnetic behaviour of clusters .
the basic idea is that a cluster of a specific size might have two or more structures with different magnetic moments and that these structures or _ isomer states _
occur with their statistical probability .
as shown below , this simple assumption causes a strong deviation of the paramagnetic behaviour of the magnetic susceptibility from the curie law . as an interesting feature , a dependence on the strength of an external magnetic field might occur , even at relatively small fields .
transition metal clusters seem to be natural candidates for the occurrence of such effects because of their different possible electron configurations with very similar energies , which in turn yield quite different average magnetic moments .
we consider to have @xmath0 isomers with magnetic moments @xmath1 and ground state energies e@xmath2 , calculated using an appropriate spin - density functional hamiltonian . in the presence of an external magnetic field the energy changes .
first order quantum mechanical perturbation theory yields @xmath3 where g is the land factor and the magnetic field is pointing in the @xmath4-direction ( see refs . ) .
the first term describes a paramagnet , which will be considered exclusively in what follows and can be described with a purely classical description @xcite , as @xmath5 now we are able to write down the partition function as @xmath6 indeed this is an extremely simple partition function , which can be improved to any complexity by adding all degrees of freedom of the clusters , e.g. by substituting the summation over the isomers by integration over the whole configuration space over the atomic positions or by adding a summation over different spin multiplicities of the clusters . from eq .
( [ part ] ) the average magnetic moment @xmath7 and the magnetic susceptibility @xmath8 can be calculated from eq .
( [ part ] ) easily @xcite , using @xmath9 @xmath10 if @xmath11 is taken as the number of particles per unit volume . by the dissipation - fluctuation theorem is @xmath12 . ] in the case of small magnetic fields @xmath13 h @xmath14 we get @xmath15 and @xmath16 here , @xmath17 is the partition function defined in eq .
( [ part ] ) for zero magnetic field .
( [ avchi ] ) reveals simply that @xmath8 is a linear superposition of the susceptibilities of the individual isomers weighted by their thermal probability .
the general behaviour of the paramagnetic susceptibility according to eq .
( [ avchi ] ) is plotted in fig . 1 for a hypothetical system with only two major isomer states with magnetic moments @xmath18 and @xmath19 and their energy difference @xmath20 . unlike a purely paramagnetic behaviour according to the curie law , the plots reveal local minima and maxima .
a similar behaviour has recently been experimentally found by cowen _
@xcite for fe@xmath21 clusters in supercages of nay zeolite .
the locations of the extrema only depend on the ratio between the magnetic moments @xmath22 and @xmath23e . for a given ratio @xmath22 ,
the position of the minimum depends in an almost linear fashion on @xmath23e , as can be seen in fig . 2 .
the extraordinary sensitivity of these results on @xmath23e is illustrated by the fact that a change of @xmath23e by only a few mev moves the minimum by hundreds of degrees kelvin . in fig .
3 the average magnetic moments of the canonical ensemble are plotted for various magnetic fields .
these plots reveal the fact , that in the presence of a magnetic field the occupation probability of the isomers changes dramatically due to the additional magnetic energy . if one views h to be small in the sense of eq .
( 5 ) , the magnetic susceptibility can be easily calculated .
for varying magnetic fields this calculation yields a behavior which is very similar to that of a varying ground state energy difference ( see fig . 4 ) . at first sight ,
configurations of different symmetry such as bcc versus fcc structures of fe clusters seem to be the best candidates to show the effects discussed above .
even if the energy difference between two isomers might be very small , the transition from one isomer to the other may involve a complex concerted motion of atoms and may be associated with a nonzero activation energy .
thus a thermal relaxation might take quite a long time , and a hysteresis might be observable .
lee and callaway @xcite have found interesting results for cr@xmath24 and v@xmath24 bcc clusters , which are given in table i. with varying lattice spacing the average magnetic moment changes by a factor of up to five , whereas the ground state energy changes are only @xmath25 ry / atom .
this is especially interesting since even a simple spatial expansion of the transition metal clusters intuitively is more probable than a structural transformation . from our results
we infer that it is extremely hard to obtain phenomenological results for the magnetic behaviour of clusters from spin - density functional methods , since the results depend very sensitively on ground state energy differences , which are often very small and comparable in magnitude to the precision of these _ ab initio _ methods .
in addition , for larger clusters , the number of relevant isomers increases dramatically .
nevertheless , we have shown that in the model case of two cluster isomers , the rich behaviour of the magnetic susceptibility as a function of temperature and the external magnetic field is a sensitive tool to probe structural properties of clusters . on the other hand , once the general magnetic response due to the above discussed effects is understood , one might use the susceptibility of the clusters as a very sensitive thermometer .
moreover , by applying a sufficiently strong magnetic field , one might be able to suppress one of the transition metal cluster isomers , provided the clusters have ample time for structural rearrangement .
an extended study of these effects , including specific examples , will be published elsewhere .
we thank professors cowen and dye for making their experimental data available to us prior to publication .
one of us ( dt ) acknowledges financial support by the national science foundation under grant number phy-92 - 24745 and the office of naval research under grant number n00014 - 90-j-1396 . | we investigate the magnetic properties of small transition metal clusters using a simple statistical model , which requires some input data from _ ab initio _ spin - density functional calculations . in our study , we consider a thermodynamically equilibrated ensemble of clusters with different structures , spin multiplicities , and ground state energies .
we calculate the physical properties of this system by weighting the individual configurations according to the boltzmann statistics .
we find that presence of isomers with very similar ground state energies , yet very different magnetic properties , gives rise to a rich magnetic behaviour of the system which differs significantly from what would be expected for single configurations .
we apply the present model to determine the magnetic susceptibility of a cluster ensemble of langevin paramagnets .
our results show that some of the anomalies in the magnetic behaviour of transition metal clusters might be understood in the framework of our model which is , of course , limited by the extremely high computational effort needed to obtain the input data . |
Subsets and Splits